A Literature Review of Food Analytics

doi:10.21203/rs.3.rs-4121495/v1

Download PDF

Research Article

A Literature Review of Food Analytics

https://doi.org/10.21203/rs.3.rs-4121495/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

This research examines the potential use of modern technologies such as big data, data science, artificial intelligence, and machine learning, which have penetrated several aspects of our lives, to address food concerns and problems, forming the nowadays called food analytics. We discuss the potential use of such technologies in relation to food problems and shortages. We analyze the opportunities and challenges associated with the use of such technological advancements and the potential benefits for the global food system. We also provide a research agenda with future directions for the application of big data, data science, artificial intelligence, and machine learning to the food ecosystem.

big data

data science

food analytics

food ecosystem

With the world population expected to rise dramatically over the next few decades, and in the face of extra stress caused by climate change and urbanization, one of the most pressing challenges that the world faces today is securing basic resources such as food, energy, and water (Pitts et al., 2020). A food system can be broadly defined as encompassing "all the elements (environment, people, inputs, processes, infrastructures, institutions, etc.) and activities related to the production, processing, distribution, preparation, and consumption of food, and the outputs of these activities, including socio-economic and environmental outcomes" (HLPE, 2017, p.23).

By engaging in activities such as deforestation, urbanization, transportation, and modifying the landform and vegetation cover, humans alter the ecological, hydrological, and thermal features of their surroundings. Consequently, there is an interdependent relationship between human and natural systems. Further, population growth, climate change, and development have intensified natural resource usage and environmental degradation to levels that jeopardize the stability and security of food, energy, and water system (FEWS) services. Nevertheless, unsustainable resource usage causes biodiversity loss and natural resource degradation, both of which have serious consequences for vulnerable communities (Pitts et al., 2020). Nowadays, the importance of data and analytics is escalating, as well as their implementation in several industries, for they enable practitioners to come up with data-driven (DD) solutions. Therefore, enhancing products and services while saving scarce resources is a win-win!

Data and analytics are of great importance to researchers, for they help them come up with data-driven decisions (DDD) to address life-threatening situations. For this reason, we believe that this cross-disciplinary research would help bridge a gap, as well as serve one of the United Nations’ (UN) Sustainable Development Goals (SDG)- goal number 2- “Zero Hunger”. The food industry is massively diverse and includes lots of intricate processes thus, with the help of data analytics (DA), we could serve the food industry’s three main sectors; the primary, which is concerned agriculture, the secondary, which is concerned with manufacturing and production, and finally the tertiary which mainly involves the finished products sale.

This paper aims to answer the following research question: how could the use of big data and analytics algorithms help identify, address, or contribute to the solution of some of the food system’s challenges and concerns?

The paper is structured as follows. Section 2 describes the overall methodology which has been used to review the literature. Section 3 provides an overview of the theoretical foundation of this study such as DS, ML, etc. Section 4 outlines the theoretical background of the study followed by the introduction of the context of this report, namely, food ecosystem, food shortage, food safety, and lastly, institutions and food. Then, section 5 provides examples of using BD and DA in food sector within European projects and how it impacts them. Section 6 follows and includes some policies regarding food. The results of the literature review including methods used in food system, data in food system and the identified challenges are then presented. Finally, the paper concludes with an agenda for future research on this topic.

To address the aim of this research, we followed a concept-centric literature review approach as outlined by Webster and Watson (2002). This approach contrasts with the author’s centric approach in which the readers are usually familiar with the main topic and there are already available studies that discussed the main topic in detail. We chose the concept-centric method as it allows us to systematically synthesize the literature and enables us to create a preliminary classification on the food analytics literature.

Accordingly, the literature review process was started by identifying the core journals and conferences in food analytics. We then went through the table of contents of each of these core journals and conferences and manually searched for the relevant articles by reviewing the title, abstract, and keywords of the articles. In addition to the core journals, we searched for the articles in online databases (namely, Scopus, Web of Science, EBSCO, PubMed, MDPI, and Taylor & Francis), using the search terms for literature search. The keywords that are used for this systematic literature review were: food, analytics, big data, AI, and IoT. Any meaningful combinations of these keywords were included as a search term. Examples of the search terms used were food analytics, big data for food, AI and food, AI in food, IoT in food analytics. Other search techniques were also applied, including logical combination of the keywords, such as: “food” AND “IOT” AND “analytics”; “big data” AND “food”; “food” AND “AI”; and so forth.

Finally, to identify further relevant studies, backward and forward citation analysis based on Webster and Watson’s (2002) recommendation were conducted. We employed this approach because the number of relevant findings in the previous steps was too few to obtain reliable results. In this article, due to the emerging nature of food analytics, we also included the secondary resources (i.e., sources that analyze, interpret, or summarize primary literature), such as industrial and consultancy reports, project deliverables, press, and so on. However, only publications in English language were considered in this review and no time limitation was set. Figure 1 summarizes the research methodology followed in this literature review and Appendix A shows an overview of the outlets, publication year, etc.

Many disciplines are now attempting to use data science (DS), which entails the use of algorithms running on top of big data (BD) following a process, to extract information and insights from structured and unstructured data and thus be able to address traditional problems, e.g., food safety, via innovative technologies. We present theoretical background on big data (BD), data science (DS), artificial intelligence (AI) and machine learning (ML), and finally, the internet of things (IoT) in this section so that we can use them later in the paper as a lens to review current breakthroughs in food that have resulted in what is known as food analytics.

Analytics is a multidisciplinary subject that depends on a wide range of skills and knowledge from several domains, such as BD, DS, AI & ML, and IoT, all of which work in a contextual domain and expertise. The ability to deal with and analyze diverse types of data, including structured, unstructured, and streaming data, is a critical part of analytics (Elgendy, Elragal & Päivärinta, 2021). Analytics has permeated many of our disciplines, allowing it to be at the forefront of current advances in society, enabling digitalization, in food, enabling food analytics, in healthcare, enabling digital healthcare, and so on. A representation of some of these areas can be found in Fig. 2 below. In the next sections, we provide theoretical background on the analytics domain’s components.

3.1 Big Data

In today's popular publications, whether online or in print, it has become increasingly common to come across references to DS, analytics, BD, or a combination of these terms (Agarwal & Dhar, 2014). BD is data of such volume, spread, diversity, and velocity that necessitates the employment of technical structures, analytics, and tools to enable insights that uncover hidden information and create value for enterprises. BD is defined by three key characteristics: volume, variety, and velocity (aka the three V’s). The vastness and immensity are referred to as its volume. The rate at which data changes or is created is referred to as its velocity. Variety comprises the various formats and types of data, as well as the discrete types of uses and methods of interpreting the data. Lastly, the primary attribute of BD is data volume (Lau et al., 2016). BD can be quantified by size in terabytes or petabytes, as well as the number of records, transactions, tables, or files. Furthermore, one of the things that makes BD really big is that it comes from numerous sources than ever before, including IoT data, logs, clickstreams, and social media. When these sources are used for analytics, typical structured data is joined by unstructured data such as text and human language, as well as semi-structured data such as extensible markup language (XML), JSON, or rich site summary (RSS) feeds. Moreover, multidimensional data from a data warehouse can be retrieved to provide historical context to BD. Therefore, variety is equally as crucial as volume in BD. Additionally, BD can be defined by its velocity or speed. This is the frequency at which data is generated or delivered. Streaming data, which is acquired in real-time from websites, is at the cutting edge of BD. Some researchers and organizations have proposed adding a fourth V, veracity. The quality of the data is the emphasis of veracity. Due to data inconsistency, incompleteness, ambiguity, delay, deception, and approximations, this classifies BD quality as good, bad, or undefined (Elgendy & Elragal, 2014).

Actually, not just corporations and governments generate data; we are all data generators now (McAfee & Brynjolfsson, 2012). We generate data using mobile phones, social network interactions, GPS, etc. Nevertheless, most of such data is not formatted in such a way that it can be stored and/ or processed in a standard database management system (DBMS). This necessitates the use of BDA approaches to make sense of such data. However, BD cannot be handles using traditional methods and approaches due to its high volume, variety, velocity, value, and veracity (Elgendy & Elragal, 2014). To begin, volume refers to the sheer amount of data, whereas variety refers to the many sorts of data acquired from structured and unstructured data sources. Velocity refers to the rate at which data is collected, processed, updated, and analyzed. Value, on the other hand, relates to the strategic and informational benefits of BD, whilst veracity refers to the dependability of the data sources. Variability and visualization have also been incorporated in recent years. Variability refers to how the insights alter as the perception of information changes or as new data is added to the mix. Figure 3 depicts the BD Vs covered in this study, with the three primary ones highlighted in blue.

By making more types of information available and useful at a higher frequency, BD may unlock significant value by increasing product and service development, boosting performance, improving decision-making (DM), and leading to better and more informed management decisions (Manyika et al., 2011). To collect, analyze, link, and compare such datasets, however, appropriate technology, processing capacity, and algorithmic precision are required. As a result, BD can provide a greater level of intellect and knowledge that can provide previously inconceivable discoveries with the appearance of truth, objectivity, and accuracy (Boyd & Crawford, 2012). Yet, simple ownership of BD is insufficient to generate a sustained competitive advantage, which requires the ability to compile structured and unstructured data, analyze massive amounts of such data, and use the insights to inform decisions (Amankwah-Amoah & Adomako, 2019). As a result, it is well acknowledged in the literature that BD differs from traditional data and requires different approaches for storage, management, and processing than previously available data and information (Elgendy, Elragal & Päivärinta, 2021).

3.2 Data Science

The systematic extraction of non-obvious and useful patterns and knowledge from data is known as DS (Dhar, 2013; Kelleher & Tierney, 2018). Its goals are to advance research, support organizational DM, and enable a data-driven society (Ahalt, 2013). The definition demonstrates how closely DS is related to science, scientific methodologies, and business analytics. Naturally, much of the definition of science and scientific method, that is, the “principles and procedures for the systematic pursuit of knowledge involving the recognition and formulation of a problem, the collection of data through observation and experiment, and the formulation and testing of hypotheses”, is shared by DS (Merriam-Webster, 2022). As a result, what we designate as data, the means of knowledge extraction, and the end goal distinguish DS. “Knowledge” refers to insights and claims that can be produced through examining factual data in the DS domain and adjacent domains, such as decision support systems (Rizk & Elragal, 2020). DS is predicated on the reality that enormous amounts of data are now generally available, and that the analytical methods we use to uncover insights, useful patterns, and hidden information from data are now widely available in the form of tools and libraries, such as python libraries.

3.3 Artificial Intelligence and Machine Learning

The use of AI in our daily lives exponentially increases. According to Luckin et al. (2016, p.14), “AI is a computer system that has been designed to interact with the world through capabilities e.g., visual perception & speech recognition, and intelligent behaviors e.g., assessing the available information and then taking the most sensible action to achieve a stated goal, that we would think of as essentially human”. Another definition of AI states that it is an area of research concerned with assisting machines in finding more human-like answers to challenging problems. This basically entails taking traits from human intellect and applying them to algorithms. Despite its close link with computer science, AI is also associated with a wide range of other key disciplines, including mathematics, psychology, cognition, biology, and philosophy (Tirgul & Naik, 2016).

AI is the science of programming machines to behave intelligently. Although there are no widely agreed definitions of AI, it may be characterized as the study of ideas that enable computers to be intelligent. Maybe the most intriguing definition was the slogan of Thinking Machines, Inc., a now-defunct computer company: "making machines that will be proud of us" (Murphy, 2000). The acronym AI is divisive, sparking ongoing philosophical discussions about whether a machine can ever be intelligent. As a result, to escape the controversy surrounding the term AI, many researchers define their work as “intelligent systems” or “knowledge-based systems” (Murphy, 2000). Every day, new AI applications, spanning from ML to robotics and other AI-enabled technology, enter our lives.

At the heart of AI is ML, which is a collection of algorithms used to analyze datasets. By utilizing powerful analytics algorithms, ML strives to process and evaluate large datasets. It enables the development of practical ideas for assessing performance, gaining a competitive edge, and functioning as a new platform for efficiency, creativity, and improved data-driven DM (Wamba et al., 2016). Furthermore, analytics include the technology, processes, tools, and techniques, as well as analytical approaches, that may be applied to datasets to produce meaningful insights and actionable prescriptive, descriptive, and predictive results (Mikalef et al., 2018). The analytics engine, at its core, employs a series of complex algorithms to discover (hidden) patterns in (big) datasets (Russom, 2011).

The top 10 data mining algorithms were determined during the IEEE 2006 International Conference on Data Mining (ICDM) based on expert nominations, citation counts, and a community survey. In order, those algorithms are: C4.5, k-means, support vector machine (SVM), Apriori, expectation maximization (EM), PageRank, AdaBoost, k-nearest neighbors (kNN), Naïve Bayes, and CART. Classification, clustering, regression, association analysis, and network analysis are all covered (Elragal & Klischewski, 2017).

AI involves developing theories and computer systems that can perform tasks traditionally requiring human intelligence, such as sensory perception and decision-making. To achieve desired goals, AI utilizes external data collected from sources like IoT and big data, and employs knowledge-based rules provided by developers or identifies underlying rules and patterns using machine learning (ML). ML is a pivotal aspect of AI as it enables computers to learn from the environment without explicit programming, mimicking a key trait of human intelligence.

Simply defined, ML algorithms distill and agglomerate knowledge from unstructured data in such a way that their outputs are executed by computer programs capable of doing helpful tasks such as alerting users or triggering crucial processes. It delves into the study and design of algorithms that can learn from and forecast data; such algorithms bypass strictly static programming instructions by generating DD predictions/ decisions. Figure 4 depicts the general classification of ML algorithms into three main categories (Misra et al., 2022; Nychas, Panagou & Mohareb, 2016).

However, there is a need to bridge the gap between the numerous emerging and very promising gadgets that may be used in the food industry in conjunction with the appropriate data mining and analysis. This multidisciplinary and multidimensional data paradigm will benefit food safety management systems by integrating and crossing several scientific fields and sub-disciplines such as process chemistry development, information technology, food science, food microbiology, molecular biology, process analytical chemistry, vibrational spectroscopy, bioinformatics, ML, chemical engineering, process systems, and control engineering (Nychas, Panagou & Mohareb, 2016).

AI is considered as improving scientists and governments, as well as traditional political and economic problem-solving tactics. Sustaining and nurturing an expanding population through the provision of nutritious and safe food is a challenging and complex undertaking for the food industry. To limit the risks, forecasting hazards, risk assessment, and prevention connected to food safety necessitate a simple and efficient heuristic prediction approach such as BDA. Various businesses and farmers could benefit from the usage of BD by evaluating and extracting productive value information from it, increasing their efficacy, and opening the door to smart farming. There is an urgent need to fill the void created by all food and agricultural problems, as well as optimize current processing methods with future sustaining automation technology.

ML provides statistical methods for analyzing data, such as labeled data from the past or clustering algorithms based on data similarity. Deep understanding provides architecture or a platform for the development of multi neural networks (NN) as a tool for mimicking the human brain in artificial NN (ANN), convolutional NN, and recurrent NN. ANNs are concerned with a problem statement in numbers, whereas convolutional NN are concerned with input in visuals or images. Recurrent NN can handle time-series-related issues (Sharma et al., 2021).

3.4 Internet of things (IoT)

IoT connects physical objects like home lights, heating, and even refrigerators to a shared network. It enables for the storage of large amounts of data, which can then be used later or in real-time analytics. Kevin Ashton first coined the term “Internet of Things” in 1998. Things in IoT refer to any existing object that can communicate or cannot communicate. IoT architecture is thought to be a 3-layer technology. The perception layer, the network layer, and the application layer are the three layers. Nonetheless, two new levels are frequently incorporated into the architecture, the middleware layer, and the business layer. IoT is built on machine-to-machine (M2M) connectivity. The IoT connects two machines without the need for human intervention (Aazam, Hung & Huh, 2014).

The IoT is a network of physical objects- "things"- embedded with sensors, software, and other technologies for connecting and exchanging data with other devices and systems through the internet. It is a massive network of interconnected things and people, all of which collect and share data about how they are utilized and the environment around them. These gadgets range from common domestic items to complex industrial machines (Clark, 2020; Oracle, n.d.). They can communicate with one another and with the physical world by exchanging relevant data between the physical and virtual worlds. These things can also respond autonomously to happenings in their surroundings. All these processes can activate some actions and create services by human intervention or M2M communication (Ahmadi et al., 2019).

IoT has its application in several areas, such as connected industry 4.0, smart cities, smart homes, smart energy, self-driving cars, smart agriculture, and healthcare. IoT aims to integrate the physical world with the virtual world by using the internet as a medium to communicate and exchange information. In Fig. 5, you will see some of the IoT real-life applications.

IoT devices are made up of embedded systems that communicate with sensors, actuators, and require wireless connectivity. These IoT gadgets are also known as IoT sensors. The embedded system is made up of field programmable gate arrays (FPGAs) or microprocessors, communication modules, memory, and input/ output interfaces. The sensors are used to monitor and measure various farm variables (for example, soil nutrients and weather data) and production factors. Location sensors, optical sensors, mechanical sensors, electrochemical sensors, and airflow sensors are the various types of sensors. Air temperature, soil temperature at various depths, rainfall, leaf wetness, chlorophyll, wind speed, dew point temperature, wind direction, relative humidity, solar radiation, and atmospheric pressure are all measured with these sensors (Elijah et al., 2018).

Sensory devices are inexpensive to operate after the initial expense of purchase and installation, and they create an abundance of data that can be processed for free (Nychas et al., 2021). The impact of IoT stems from its potential to enable robust connection between the physical and digital worlds, a concept known as the 4th industrial revolution. Indeed, the application of IoT in industry is sometimes referred to as Industrial IoT (IIoT). Remote sensors receive information provided by machines in the IIoT framework to boost efficiency, encourage better DM, and establish competitive advantages, independent of industry or company size (Misra et al., 2022). Furthermore, it would be imprudent not to use the emerging technologies that collect data, from smart sensors to satellites that can be crunched using various cloud-based analytic software program gear such as AI, to make aquaponics operations more efficient and environmentally friendly (Paul et al., 2022).

IoT, BD, and AI are arguably old terms in the tech industry that have only recently gained traction. Figure 6 depicts an overview of what each signifies. According to Google Trends search histories, IoT and BD have grabbed the interest of a wide spectrum of internet users in the last couple of years, although AI has been a topic of discussion for well over a decade. Moreover, as the number of communication devices increases, so does the volume of data generated, and AI is continuing to integrate in some form or another into the lives of a substantial section of the world's population. Unlike AI, IoT, which is basically an industrial technology, lugubriously, remains of little interest to the general population (Misra et al., 2022).

Advances in ML and BD research hold great promise for international development institutions such as UN agencies to gain new insights and provide analytics that can improve DM and systematize evidence around strategic themes by leveraging the vast information generated by accountability mechanisms. In a ML environment, computers employ various algorithms to mimic human learning and accomplish tasks, in a process that begins with data collection and processing and concludes with the use of statistical models for analysis and visualization of findings. ML models and their predictive capacity are subsequently upgraded based on their ability to learn new things from data and human inputs (Garbero, Carneiro & Resce, 2021).

Today, when we talk about the agri-food sector, we must also talk about the “food system”. This section will outline the primary components that make up the contemporary food system. Firstly, we will provide some statistics and background information to contextualize the topic. Following that, we will delve into the food ecosystem, which will provide a concise overview of the food chain and the interconnectedness of its various components. Additionally, food shortage, here we explain some consumer buying trends which brings about the problem, and the concept of circular economy, which helps in some ways save resources. Moreover, food safety and how to ensure it in efficient ways. Finally, we will discuss the relation between some institutions and food. In the next subsection we will state the reality behind the food context.

4.1 The Food Context

The UN reports that one-third of the world’s food is thrown away each year, which adds up to $750 billion that is completely wasted. This means that about 28% of the world’s agricultural land is used to produce food that is eventually wasted. The supply chain management in a food business is very challenging owing to the need for advanced control systems for coping with perishables, fluctuating supply–demand variations and narrow food safety, and sustainability goals (Misra et al., 2022). Food losses in developed nations are comparable to those in developing countries, although in poor countries, more than 40% of food losses occur at the post-harvest and processing stages, whereas in developed countries, more than 40% of food losses occur at the retail and consumer levels. Agriculturists throughout the world are now recognizing the pressing need for DS skills to unlock and act on the intelligence housed in agribusiness’ amassed BD and information to maintain food security. Ashford (2015) mentioned in a report that “The internet of things could be key to the farming industry meeting the challenge of increasing food production by 70% by 2050”. According to literature, the farming industry must incorporate IoT if it is to feed the 10 billion global population predicted by 2050. IoT is commonly defined as a network that connects all relevant elements to the internet via radio frequency identification (RFID), global positioning systems, and sensors, among other information detection devices, in line with the accepted protocol. Sensors/ actuators can be linked, invoked, deactivated, or disabled via the internet using the IoT framework. The system uses IoT to provide 2-way communication between the sensor/ actuator and other devices. Enabling the agriculture domain with IoT would provide real-time data input on various assets, but it highlights the challenge of being able to evaluate, make sense of, and derive intelligence from massive amounts of data streams coming from the agriculture sector via IoT framework (Nychas et al., 2021; Parvin et al., 2019). Consequently, the use of IoT networks involving humidity, temperature, light, and microbiological and product quality sensors for real-time monitoring of products in transit is useful for the food industry in rescheduling, recalling, or taking the appropriate constructive actions.

4.2 Food Ecosystem

Humans instinctively are born to thrive for survival. Nowadays, with the increase in famine, capable people intuitively secure themselves with over-buying of food, which results in lots of waste. This brings about a negative impact on the ecosystem! Food wastes are known to release harmful gases, one of which is methane (Sadiku et al., 2020). As methane production increases, air pollution increases and soil nutrients decrease, which affects crop production, and causes a disturbance in the ecosystem (Arif & Verstraete, 1995). Moreover, methane is known to be a tropospheric ozone precursor, which obstructs natural plant growth and decrements the agricultural production, thus disrupting the food chain (Cutting Agricultural Methane Emissions Can Help Save the Planet while Increasing Yields and Improving Lives, 2021). Hence, controlling crop and food wastes that release harmful gases ought to be addressed for it serves the production efficiency as well as sustainability (Sadiku et al., 2020).

A natural area of focus for agri-food scientists and engineers would be maximizing the implications of developing information technology (IT) for feeding the world sustainably (Misra et al., 2022). Soil is one of the world's most valuable natural resources, it acts as a carbon storage site. Soil offers food, fiber, fuel, and aids with water and climate regulation. Although most organizations are aware of its benefits, they are not aware of the risk they may face due to its degradation. Soil organic carbon (SOC) is abundant in soil and is necessary for retaining and improving soil fertility and production. Nowadays, globally cultivated soils have been claimed to have lost 50–70% of their original carbon content. This suggests an ideal chance to return carbon to the soil to elicit a positive response, so improving soil health and addressing a variety of issues such as climate change, food, and water security. Given SOC's position as a key parameter in many soil functions, SOC data evolution is the need of the hour. These measurements are required to forecast SOC feedback on climate change. BDA has evolved into one of the most potent ways for extracting usable information from data. With geospatial data of very high quality and quantity available, a combination of data assimilation and ML algorithms can also improve SOC dynamic prediction and analysis (Hinge et al., 2021).

4.3 Food Shortage

As the population continues to grow, the daily struggle to combat hunger becomes more difficult. It is estimated that the global population will reach 10 billion by 2050, leading to an increased demand for food and a decreased availability of resources. The initial COVID-19 shutdown saw customers stockpiling commodities like dried pastas and frozen meats, placing an unprecedented strain on the procurement businesses that employ a just-in-time business approach. Ever since, the temptation to raise stocks to match consumer needs has never been higher (Nychas et al., 2021). For so, unveiling solutions is the uttermost concerning thing of all!

Food waste has reached unprecedented proportions. Every year, more than 1.3 thousand million tons of food are wasted around the world. It is socially unacceptable to squander food while many regions are still suffering from famine and malnutrition. As per the Food and Agriculture Organization (FAO) of the UN, between 720 and 811 million people faced hunger in 2020 (FAO, 2021). Food waste involves production and logistics activities that do not result in product consumption, both environmentally and economically. According to the FAO, the societal costs of food waste were around $2.6 trillion in 2014, of which $700 billion are associated with environmental damages and $1 trillion are associated with economic losses owing to wasted and lost production. Given its importance, some institutions have called on societies and businesses to reduce food waste. For instance, the UN listed food waste reduction along supply chains as a target in their SDG framework (Altintzoglou, Honkanen & Whitaker, 2021).

On the bright side, positive initiatives that aim to successfully deal with food surplus can be implemented. For example, circular economy (CE) views “waste as a resource” and proposes a solution to the food waste problem. In this regard, the food waste hierarchy (FWH) provides a comprehensive framework for prioritizing potential actions ranging from surplus food prevention to the extraction of new value in the form of new edible or non-edible items. Prevention is the highest priority in FWHs since it eliminates food waste by addressing the underlying causes. Food waste prevention necessitates expertise (e.g., technological skills) that the actors in the core food supply chain may lack. As a result, relationships with technology providers can be critical. Alternatives at the FWH's lower layers include re-use for animal feed, re-use for composting, and recovery. This collaboration drives relevant changes in present value generation, emphasizing the significance of rethinking company models through collaboratively exploiting circular systems. Thus, food waste prevention requires changes that extend outside the supply chain to embrace business model innovation. The main causes of food waste can be divided into supply and demand-related trends and quality and process control variables (e.g., planning, forecasting, marketing management, inventory management) (Ciccullo et al., 2022).

A shift to a stronger form of food waste prevention typically requires a significant change to the company's operations and should help them stay afloat for the foreseeable future. Process improvement results in a better fit of food supply and demand, which leads to lower long-term profitability because customers end up buying less. For example, the idea of “zero waste” apps are to make customers at various levels of the supply chain aware of how much food they truly need to buy to meet their needs. The apps recommend quantities that are significantly lower than what clients would have purchased without app support. As a result, these apps decrease the food waste resulting from over-purchasing and spoilage of foods. In other words, shifting the “zero waste” philosophy downstream to consumers will result in a fall in revenue for the entire supply chain, without necessarily compensating in terms of cost reduction (Mourad, 2016). The agri-food business is a never-ending source of growth for feeding a large population, but there is a significant need to develop high-standard procedures using intelligent and creative technology like AI and BD.

4.4 Food Safety

At any point in the farm-to-fork food chain, pollutants that are chemical, biological, or physical can contaminate food. Traditional in-lab safety evaluations are neither appropriate nor practical across the chain of food production and distribution. Thus, to contribute to food safety assurance, the on and postprocessing food sectors, as well as food business operators, focus on the installation of an effective food safety management system (FSMS), which is based on managing, monitoring, and documenting essential parameters for the purpose of identifying potential threats to food safety. The process analytical technology (PAT) concept, which arose from the desire of biopharmaceutical industry regulators to shift product quality control towards a science-based approach, is proposed for the food industries with the following goals in mind: optimization of food quality and reduction of food waste through more efficient control of the processes, considering all processing steps and integrating sensors at the critical control points (CCP) (Nychas, Panagou & Mohareb, 2016). A variety of non-destructive methods using sensors have been utilized for identifying hazardous compounds within the food processing and at the end of production (finished product), these techniques include gas chromatography (GC), high-performance liquid chromatography (HPLC), HPLC-mass spectrometry (HPLC-MS), GC-MS, nuclear magnetic resonance (NMR), fourier transform infrared spectroscopy (FT-IR), and Raman spectrometry. These sensors ought to be strictly used within controlled conditions. Sensors having spectra in the ultraviolet (UV), near-infrared (NIR), mid-infrared, and visible (VIS) ranges have been used in different sensor spectroscopy methodologies to this aim (Nychas et al., 2021).

Furthermore, given the expensive nature of using herbicides, and the heightened awareness of the potential harm caused by over-reliance on chemicals, there is an increased push to utilize cameras as weed-detection sensors in agricultural settings. There have historically been two ways to weed control: chemical and mechanical. Mechanical control has the advantage of being environmentally friendly yet labor expensive, whereas chemical control has the opposite characteristics. Some firms have recently emerged as providers of intelligent and autonomous weeding equipment that use cameras to detect weeds in the field. This makes building commercial harvesters a significant task, despite the necessity for such robots to save labor and optimize productivity (Misra et al., 2022).

Another method used is collective expertise, social co-governance in the food sector offers sustainable answers to the problems of food security and food quality. Crowdsourcing is typically used in social co-governance to collaborate with customers or experts and create value. Over 40% of new food product developments fail, and these failures typically have an impact on the viability of small and medium-sized businesses. The path of product development has been determined by large food firms after analyzing BD and trying to crowdsource consumer preference information. Indeed, food security has a great deal of potential with social co-governance. To prevent food deterioration and waste, the intelligent control of food inventory can be done by integrating mobile consumer group data with food shelf-life. Monitoring food-borne illnesses, spotting contaminated goods, reducing the likelihood of food rumors, and improving food safety are some applications of social co-governance as well (Tao et al., 2021).

In addition, a proposed food safety solution includes applications that contain safety information regarding safe food transit and storage practices, however this requires active customer participation. DS seeks to uncover hidden patterns within enormous volumes of data by combining advanced statistical methodologies such as ML, AI, and pattern recognition with current data integration, storage, and visualization technologies. With rapid advancements in sensors, mobile technologies, and computer processors, DS is gaining traction across a wide range of industries, encompassing practically every field of research and everyday applications. For instance, mobile phones, with the ability to run instructive apps and built-in sensory equipment such as cameras, are becoming an increasingly useful, albeit an underutilized scientific research tool (Nychas et al., 2021).

4.5 Institutions & Food

Due to processing contaminants, existing and emerging risks, the European Union (EU) has, recently, mandated the use of a mechanism called process analytical technology in the food manufacturing process and production lines. This mechanism addresses a broad demand highlighted by the agri-food industry. In the EU, the primary responsibility for ensuring food safety lies with food business operators; this responsibility is monitored by regulatory personnel; as well as by consumers, who must understand that they are also accountable for the safe-handling, preparation, and storage of food (Nychas et al., 2021). Moreover, the World Health Organization (WHO) has lately entailed the BD approach to endorse DM in food safety, which brought about the food safety platform “FOSCOLLAB”, which integrates numerous sources from different disciplines. This platform integrates and makes user-accessible dashboards from both structured and unstructured data from a variety of industries, including animal, agricultural, food, public health, and economic indicators (Marvin et al., 2017).

The European Commission has devised a BD strategy and advocates for a data-driven economy in Europe. They encourage data openness, such as free internet access to EU-funded research outputs, including scientific papers and research data. This includes projects supported by the EU on crop monitoring for developing countries (e-Agri), product lifecycle monitoring (LinkedDesign), and increasing the efficiency and quality of the product development process (iprod). Innovative consumer support tools for selecting nutritious food will be developed as part of the RICHFIELDS project (www.richfields.eu) (personalized nutrition). The developed tools will make use of food product data, food intake data, lifestyle, and health data, as well as real-time data. Data generated by consumers using mobile apps or technology-wear (Marvin et al., 2017).

The Food, Energy, Water nexus is a UN paradigm that emphasizes the interdependence of food, energy, and water. Pressures or development in one area will have an impact on others. The study of FEWS interactivity is becoming more popular as a framework for research, technology, and policy to address ongoing sustainability challenges that necessitate a multidisciplinary understanding of feedback loops and interactions between human and natural systems. FEWS research necessitates new and innovative scientific methodologies from complexity science to address the integrative, complex, and multi-scale interdependencies spanning space and time, as well as the dynamics of their interactions. FEWS research focuses on understanding the linkages between each system to gain a comprehensive understanding of the causes and consequences of changes within and across systems. If the goal is to establish a sustainable system, the FEWS nexus indicates points of overlap or conflict among the system parts required to produce those outputs, with the ultimate goal of "raising efficiency, eliminating trade-offs, building synergies, and strengthening governance" across the systems. These data are used to inform models of ecosystem service flows and trade-offs, demonstrating what is lost and gained under alternative DM FEWS scenarios. Furthermore, modeling trade-offs can be projected over place and time, giving critical information to guide SDGs and policy decisions (Pitts et al., 2020).

Another impactful project is the UN’s World Food Programme (WFP) provides food assistance to about 100 million people in over 80 countries each year. The UN’s WFP is the world's leading humanitarian agency, saving and improving lives by providing emergency food aid and working with communities to enhance nutrition and resilience. “For its efforts to combat hunger, for its contribution to bettering conditions for peace in conflict-affected areas and for acting as a driving force in efforts to prevent the use of hunger as a weapon of war and conflict”, WFP received the Nobel Peace Prize in 2020 (Peters et al., 2022).

4.6 Policy-making in Food

Food policies are normally determined at the national level, but they may also be developed at the regional level depending on the degree of harmonization between the countries in the region. According to The World Bank, there has also been a recent and significant increase in trade agreements between countries, and while this does not always result in recognizing each country's individual food policies and legislation, it does open the door to a better understanding of individual country needs. The Food and Agriculture Organization (FAO) of the UN was established in 1945 with four express purposes: (1) to improve nutrition and living standards in member nations; (2) to improve the efficiency of production and distribution of all food and agricultural products; (3) to improve the conditions of rural populations; and (4) to expand the world economy in such a way that it would ensure humanity's freedom from hunger. For many years, international organizations such as Codex Alimentarius (CODEX) have struggled to provide a level playing field in terms of food policy and regulations. CODEX has developed several global “standards” and “guidelines,” but they are often adopted exclusively by nations that lack their own internal standards or those desiring to trade on open terms with like-minded countries (Smith, 2016).

The WHO's principal mission within the UN system is to coordinate international health issues by assisting UN member nations in achieving improved health results. According to the WHO charter, “the enjoyment of the highest attainable standard of health is one of the fundamental rights of every human being without distinction of race, religion, political belief, economic, or social condition,” and this charter represents a critical element of good food policy (Smith, 2016).

Sweden, however, has its own “National Food Strategy” as per the Government bill 2016/17:104. Its overall objective is to have “a competitive food supply chain that increases overall food production while achieving the relevant national environmental objectives, aiming to generate growth and employment and contribute to sustainable development throughout the country. The increase in production – of both conventional and organic food – should correspond to consumer demands. An increase in production of food could contribute to a higher level of self-sufficiency. Vulnerability in the food supply chain will be reduced”. This calls for laws and regulations to be developed to promote the overall goal of a competitive and sustainable food supply chain with increased production. Where it should be accomplished through appropriate charges and taxes, regulatory simplification, decreased administrative load, and other measures to boost competitiveness and profitability. The national Swedish goal to expand food production will aid in the creation and communication of long-term conditions for food supply chain firms to thrive and grow, as well as increase incentives for enterprises to invest in Sweden (Government Offices of Sweden, n.d.).

5.1 Big Data in Food Analytics

Since 2013, BD has been a study emphasis, mostly in food safety, food security, and agriculture. Although its application to food safety is still in its early stages, technology is affecting the entire supply chain. Regulatory, food enterprise (encompassing data generated at every link of the industrial chain from farms to restaurants), and media data are the most common food BD sources (including food-related news, video, pictures, and audio). High-quality BDA can help the food business thrive, however low-quality DA can harm managers' forecasting of market demand and social stability (Tao et al., 2021).

The recent introduction of advanced technologies has changed practitioners' and academics' attention to the deployment of enormous amounts of created data as a critical and efficient instrument for addressing modern food supply chain management concerns. With food and beverage firms increasingly focusing on acquiring, processing, and analyzing useful information derived from different sources within their respective food systems, data management has grown into a crucial tool in today's food supply chain. Within this paradigm, BD has received a lot of attention in recent decades, referring to massive amounts of data assets that are highly dispersed and heterogeneous in nature. Surprisingly, the annual added value that could be realized using BD in food supply chains is projected to be between $120 and $150 billion (Margaritis, Madas & Vlachopoulou, 2022). BD is mostly used in observing the concerns controlling global problems such as food security, safety, feasibility, and growth in efficiency, which expands the scope of BD beyond farming and includes the entire food supply chain. The interconnection of agricultural and supply chain devices results in real-time data evolution via hassle-free wireless IoT. To predict insights in agricultural and business process improvements, the data sets are process-oriented, machine-oriented, and human-sourced (Sharma et al., 2021).

Sailing to the other harbor, the analytics use, and monetization of data are increasingly crucial for the profitability and sustainability of multi-sided platforms (MSPs). Previous literature shows that apart from revenue growth and cost optimization, DA can decrease customer acquisitions costs, retain valuable customers, help predict customer behavior, improve customer experience, reduce fraud, provide real-time offers, and enhance DM. However, data on its own is not a source of competitive advantage since all firms can collect hordes of data from a variety of sources. Rather, data must be purposely analyzed, and activated. Nonetheless, firms face a host of issues - organizational, financial, physical, and human resources - in their attempts to create a competitive capability from the use of data and may easily fail to exploit the benefits of DA. Regarding the growth of data, along with the trends in digital business models and expected benefits from DBMS, a recent global survey of ~ 400 companies showed that 77% of companies do not have strategies to use BD effectively. Many companies are thus failing to benefit from integrating BD into their business models (Isabelle et al., 2020). The literature offers several reasons for such failures. According to Morabito (2015), BD emphasizes “utility from” data rather than “ownership of” data. This means that access to purposeful data is key. Further, raw data is useless unless it is purposely analyzed. Jones (2019) notes that there is a difference between data that can be recorded and data that actually gets recorded, as well as between the results from data analyses that get extracted, understood, and exploited for business benefits.

Further, by integrating information on environmental parameters with pathogen development and/ or hazard occurrence, BD can be utilized to forecast the existence of diseases or contaminants. For example, by monitoring agricultural conditions in the field, locations with a high frequency of aflatoxins can be discovered before they enter the food chain. Another study used a range of models and databases, including weather data, to construct quantitative models to forecast the contamination of the mycotoxin deoxynivalenol (DON) on wheat in northwestern Europe. The presence of Listeria monocytogenes could be anticipated by defining the presence of pathogens on farm fields and integrating this with environmental and meteorological data (Marvin et al., 2017).

This brings us to the realization that BD is a largely underutilized instrument in the food safety and quality sectors. Researchers stated that “big data is high volume, high velocity, and high variety information assets that require new processing forms to enhance DM, insight discovery, and process optimization”, which corresponds to the description of sensory data generated in agriculture (Sharma et al., 2021). Ciccullo et al. (2022), claimed that approximately 80% of the world’s data is unstructured. Hence, information and insights can be derived from BD by applying advanced analytics (i.e., BDA), which brings unprecedented opportunities to benefit from it. In fact, previous literature has found that firms using BDA are 36% more likely to surpass their competitors in revenue growth, operating efficiency, and can decrease their customer acquisition costs by 47% (Isabelle et al., 2020).

To conclude, cloud computing, IoT, BD, and blockchain can transform food supply chain discrete production lines into data-driven interconnected intelligent systems. Each action is automatically integrated by semantic active technology, increasing the efficiency of precision agriculture and company administration. Farmers may improve crop planting and animal growth cycles by using sensors and drones to collect data on weather, topography, and animal and crop behavior. Intelligent gadgets collect actionable data and make decisions that reduce downtime for equipment. By 2026, smart agriculture will save 4–6% of agricultural costs while increasing market value by 3%. The use of BD can help firms not only deal with issues in food production, but also find more economical raw materials, lowering manufacturing costs. It also encourages the development of smart agriculture, which saves water, preserves soil, reduces carbon emissions, and increases output (Tao et al., 2021).

5.2 Data Science in Food Analytics

According to literature, BDA can be used to improve sustainability performance of supply chains (Margaritis, Madas & Vlachopoulou, 2022). Studies also demonstrate how BD collected along all phases of agricultural supply chains, from pre-production to distribution, and analyzed using ML approaches can improve supply chain performance and environmental sustainability. For instance, accurate weather forecasts can help with water resource management. Furthermore, the capacity to extract meaningful insights from BD may increase supply chain visibility and facilitate collaboration for long-term sustainability. In general, supply chain environmental sustainability can be improved by employing BDA to estimate the environmental impact of supply chain-related choices (Ciccullo et al., 2022).

Additionally, several BD collecting and analytics platforms, such as SemaGrow (http://www.semagrow.eu/), have been developed to assist farmers in DM. This system makes use of algorithms and tools to efficiently query large-scale data collections and independent data sources. It concentrated on the agriculture domain and its cases by combining and integrating massive and heterogeneous spatiotemporal data sets. Regional agro-climatic modeling in the context of climate adaptation is one of the use cases investigated by SemaGrow. Food tracking and tracing is required in the supply chain to ensure fast recalls. This can be expanded by incorporating GPS, sensor-based, and RFID technology. This allows for the collection of near-real-time data on the location and other characteristics of the food. Such monitoring data is supposed to aid in the early detection of a problem, allowing for prompt preventive steps and, as a result, the prevention of an outbreak (Marvin et al., 2017).

Climate change, population growth, water shortages, soil degradation, and food security are all important concerns confronting traditional agriculture (Pitts et al., 2020; Sadiku et al., 2020). The data collected from sensors, such as the pH sensor and the temperature sensor, is processed using ML techniques to determine the system's health. Countries with a diverse land distribution and scarcity of water for home and agriculture use require a viable solution to food and environmental issues. The Aquaponics farming method approach promises to be a prominent alternative. For effective growth and information usage, cloud-based sustainable smart aquaponics farming employs IoT-based predictive analytics. Aquaponics is a closed-loop symbiotic system that allows aquatic life and plants to coexist. Aquatic animals produce a nutrient-rich byproduct (ammonia), which plants utilize as fertilizer. This system involves several mechanical and biological operations like heating, pumping, filtering, and so on (Paul et al., 2022). DA and sensor technology enable the creation of optimal conditions for both plants and fish, ensuring high output and improving resource efficiency by regularly monitoring the water quality, water level, temperature levels, fish health, salinity, pH level, humidity, and sunlight. Accurate data can boost fish and plant yields by allowing the nutrients transfer between plants and fish. The data can also be utilized to automate tasks that require less human participation (Elijah et al., 2018).

Ciccullo et al. (2022) analyzed the business models of a couple of startup companies in the food industry to know how they leverage BDA for food waste prevention and management. In the long run, BD could be an asset to the company in terms of ensuring business stability and high operational performance since BDA is a potential supporter in addressing food waste issues. These businesses tend to reduce food waste by leveraging other solutions, even though BDA may have theoretically provided extra assistance, even allowing them to advance to a higher stage in the FWH. Companies with a high BDA exploitation level attempt to maximize material efficiency. Companies that create value from waste, on the other hand, have poor BDA capabilities today, but if they start implementing it in the future, they may be able to tap unexploited food waste prospects. In the advanced BDA and food waste leverage cluster, firms prevent the formation of food surplus, hence preventing food waste at its source. Despite their high performance, these organizations are interested in making their food waste avoidance methods more efficient, as obtaining 100% efficiency is an ideal and demanding goal. Hence, by leveraging BDA, companies can make better decisions to combat food waste from the source, improve forecasting and management of recycling input flows, and address higher stages of the food waste hierarchy.

Food, like any living entity, demonstrates huge diversity among samples, even within the same batch of products, and this issue is exacerbated when the environmental impact, contamination impact, storage and processing conditions, and others are considered. As a result, huge volumes of data are clearly necessary for the construction of robust and efficient models. Nychas et al. (2021) have proposed an idea where the development of a distributed data repository, which will integrate a vast amount of heterogeneous data (such as microbiological, metabolomics, and multi/ hyperspectral fingerprints taken from a wide range of handling, storage, and distribution conditions), can be based on the quantification and correlation of changes associated with agri-food products. A robust and reliable monitoring service can only be produced and used profitably by correlating different food quality criteria, therefore create reliable models of product quality and safety. This information could be tracked through production, supply, and distribution, then uploaded on a cloud data storage, and be made accessible to people via live tracking systems combining mobile and web technologies by using uniquely designed QR codes on food packages or food safety apps that provide personnel with the necessary product information, such as the product origin, nutritional value, food miles, safety profile, and predictive hazards. Hence, the incorporation of such apps drives the agri-tech industry to depend on DS, a multidisciplinary field which combines data, mathematics, statistics, algorithms, and computing to model discrete aspects of the whole supply chain to identify and improve the current industry practices as inefficient.

A major driver for food producers to alter food items and ingredients unlawfully, is to enhance the quantity of the finished product with inexpensive replacements, keeping net costs low and net revenues high. Sorrowfully, vulnerabilities in the food supply chain still exist, and with elevating pressures to maintain food both cheap and in stock to meet demands, the risk of adulteration increases (Nychas et al., 2021).

There are various innovative but underutilized DS technologies available today that have the potential to revolutionize and impact the future of food safety, such as BD methods, AI, blockchain, IoT, and the digital twin (DT) (a digital representation of a physical object such as a city or factory) (Margaritis, Madas & Vlachopoulou, 2022; Nychas et al., 2021; Tao et al., 2021). The combination of these technologies would have a significant impact on the food sector's so-called Industry 4.0. Each technology, while not necessarily complimentary to the others, has essential limitations and qualities that, when combined, result in an improved system, an example is AI coupled with DT for predictive modeling risk assessment. When combined with IoT technology, AI will be boosted by the volume and intrinsic variability in IoT data; additionally, combining blockchain with AI will strengthen the transparency and traceability of such data. IoT is critical for considerably improving traceability, food safety, and waste reduction since it is built on the interconnection of all things (sensors, devices, machines, computing devices) via communication mediums (e.g., WiFi, Bluetooth, RFID). Embedded sensors, low-power wireless communications, and signal processing techniques have recently received attention, allowing devices such as mobile phones to collect and transfer data to repositories via transferring channels such as WiFi, allowing for real-time monitoring and control. In other words, the linked sensors could be put in, on, or near the production and distribution lines, where multidimensional and multivariate data is transferred over the air to a cloud-based central or distributed data repository. Validated mathematical models on the cloud server can accept POST (power-on self-test) requests from sensors as input variables and return the safety profile and risk parameters to the sensor in real-time (Nychas et al., 2021).

5.3 Artificial Intelligence & Machine Learning in Food Analytics

AI’s significance has been deemed deserving though poorly implemented. Monitoring optimum environmental settings avails the world’s growing population. AI helps in detection using sensors and sets optimum farming conditions, for greater and better crop yields, which is critical to meeting the needs of the growing inhabitants of the world. In addition, BDA helps farmers track down problems early and hence take necessary constructive and/ or preventive actions, to enhance production (Sadiku et al., 2020).

BDA and allied technologies in food science and related domains improve performance through IoT for food safety and supply chain traceability to solve food security challenges. Modern digitalization creates large amounts of data at a breakneck pace to make better decisions about agricultural produce, investigate complicated agrarian potential, and monitor equipment performance. Some AI developments in the supply chain are AI-based demand forecasting, risk-management, resilience, transportation, supplier selection, inventory management, etc. Data collection and implementation can aid in the efficient and responsible use of resources, the improvement of DM, and the application of a circular economy strategy in the food chain. Automation and intelligent systems can also be utilized to choose appropriate chemicals for food safety, and multitasking robots can help to speed up the process while maintaining quality (Sharma et al., 2021).

Blue River Technology, a start-up recently acquired by John Deere, has invented yet another sort of weed management system that utilizes cameras, computers, and AI to differentiate crops from weeds. The tractor-propelled machine, which is now functioning on a limited scale in cotton weeding, uses chemical methods to attack weeds by spraying herbicides particularly on weed-infested areas. The fundamental advantage of this method is that it reduces the amount of chemicals required in agriculture, which has both economic and environmental benefits (Misra et al., 2022).

According to Garbero, Carneiro & Resce’s (2021), a number of ML algorithms were used: those from Natural Language Processing (NLP), which encompasses algorithms capable of understanding the content of documents, extracting insights, as well as organizing and categorizing the information itself; and LASSO (least absolute shrinkage and selection operator), which is a regression analysis method in statistics and ML that performs both variable selection and regularization to enhance prediction accuracy and interpretability of the statistical model. The initial stage in data preparation was to make any non-suitable documents appropriate for analysis. This involved transforming the project report files into machine-readable documents so that data extraction could take place. The food systems perspective has become increasingly important as meeting the targets of SDG 2 to end hunger, achieve food and nutrition security, improve nutrition, and promote sustainable agriculture necessitates more holistic approaches to understanding food insecurity and nutrition deficiencies that critically analyze interactions between food production and consumption and their socioeconomic and institutional contexts. One significant challenge is exploiting integrated data that can talk about the growth of food systems and the interconnections between their components. Through project-level data analysis, machine-driven analytics can play an important role in expediting evidence generation around strategic themes, hence reinforcing the usefulness of these methodologies for portfolio systematization (Garbero, Carneiro & Resce, 2021).

In recent years, the success of AI in other scientific areas has led to deep NN and convolutional NN (CNNs) providing superior results in machine vision compared to more traditional ML approaches such as support vector machines, genetic algorithms, and partial least-squares discriminant analysis. CNNs have been used to recognize and classify food categories. While this technology is still in its primitive stages, it has yielded encouraging results with food type recognition utilizing image databases such as Food101 and UECFood (Zhou et al., 2019). Howbeit, one obstacle to the next step of applying DL image-based ML algorithms is the lack of reference photos in the libraries for all food types in varied levels of quality, adulteration, and contamination. To summarize, CNNs mimic the human visual system to extract information from visual material, where color and form are the major aspects of an image that assist in identifying its subject. These can be extracted using mathematical approaches such as edge detection and image segmentation, both of which can be automated using ML methods to gather data relating to regions of interest within an image. Edge detection distorts an image by boosting specific features within it, such as pixel intensity gradient shifts in specific directions, using convolutional kernels. These characteristics are utilized to define edges, and various kernels can extract various types of visual data, such as corners, straight lines, textures, and blank spaces. The expression of these features creates a profile of an image subject, which a NN can utilize to predict the subject's class/ type (Shelhamer, Long & Darrell, 2017). See Fig. 7 below.

5.4 IoT in Food Analytics

The introduction of advanced IoT technology has led to the adoption of more efficient and sustainable agricultural processes, progressively changing traditional agriculture into precision agriculture (PA). PA is considered one of the fastest developing fields of technology for both large and small-scale farming operations. It is now a major participant in the drive to increase global food supply. Because the central focus of this concept is focused on the collection and effective data management in the context of DM, PA is essentially the cornerstone for smart agriculture development (i.e., data-driven agriculture activities) and naturally the use of BDA in agriculture. Indeed, the application of BDA fosters opportunities in the agriculture supply chain, not only in terms of process and results benchmarking, tracking, and tracing of agricultural products throughout the entire supply chain, and eventually predictive modeling for agricultural product production and distribution. According to Wolfert et al. (n.d.), BDA are incorporated into the agricultural BD value chain, which is essentially a system comprised of sequential activities that are interconnected with each other from data collection and storage to data transformation and utilization through DA, eventually leading to data visualization and presentation to agriculture value chain stakeholders. Data in smart agriculture is typically recorded, enriched, upgraded, or combined as materials and information travel through the downstream supply chain using wireless sensor network (WSN) technology. The latter is based on interconnected sensor units that collect data from the environment and transmit digital signals via a radio frequency (RF) communication unit relevant to agricultural product storage, monitoring, and tracking across the IoT network. RFID technology, which could also be considered a wireless node application because it formulates the very basic example of interconnected “things”, is probably the most common technology that fosters the capitalization on data collection within tracking and tracing networks. RFID tags contain electronic product code (EPC) data that is accessible via RFID readers and provide information on item identification, inventory management, quality tracking, and agricultural product lifecycle evaluation. Essentially, the operation of RFID tags is related to the upgrading of barcodes, allowing for the tracking of agricultural products while encouraging information enrichment anytime RFID is recorded along the supply chain network. The use of RFID technology within a WSN via BD usage has been investigated in automated irrigation systems to identify the optimal allocation of water. Another intriguing RFID use within a WSN is smart monitoring of agricultural infrastructure via scalar sensor units that enable remote picture capture to analyze agricultural production processes and spot the appearance of pest problems or plant diseases in real-time (Margaritis, Madas & Vlachopoulou, 2022). Consequently, the use of IoT networks involving humidity, temperature, light, and microbiological and product quality sensors for real-time monitoring of products in transit is useful for the food industry in rescheduling, recalling, or taking appropriate actions.

In the agricultural sector, IoT and DA are employed to improve operational efficiency and productivity. The usage of WSN as a main driver of smart agriculture is giving way to the use of IoT and DA. IoT aspires to integrate the physical and virtual worlds by utilizing the internet as a communication and information exchange platform. IoT is defined as a networked system of interconnected computing devices, mechanical and digital machinery, items, animals, or people with unique identifiers and the ability to transfer data without requiring human-to-human or human-to-computer interaction. As the world population is expected to reach 9.7 billion in 2050, there will be a higher demand for food. To meet the world's food demands in the next years, the globe is adopting the use of IoT paired with DA. Smart agriculture will be possible using IoT and DA, which is predicted to produce great operational efficiency and yield (Elijah et al., 2018).

There are numerous instances of IoT use in agriculture (Misra et al., 2022). Crop and livestock, tracking, machinery, irrigation and water quality monitoring, weather monitoring, soil monitoring, disease and pest management, automation, greenhouse production, and precision are examples of such applications. Several environmental conditions influence crop yield in agricultural cultivation (See Fig. 8). Acquiring such data aids in understanding the farm's trends and processes. Rainfall, leaf wetness, temperature, humidity, soil moisture, salinity, climate, dry circle, solar radiation, pest migration, and human activities are examples of such data. The acquisition of such extensive records enables optimal DM to increase farm produce quality, reduce risk, and optimize profitability (Elijah et al., 2018).

Smart farming enables farmers and growers to increase output while decreasing waste, from the amount of fertilizer used to the number of trips made by farm vehicles. Food waste is a major concern to food security, with waste estimated to account for up to 50% of all food produced due to poor management from production to retail. The advent of wireless communication technology, mobile devices, and ubiquitous services has paved the road for widespread internet connectivity. The amount of linked agricultural equipment is predicted to increase from 13 million at the end of 2014 to 225 million by 2024, according to a Machina research report. Data may be accessed whenever and wherever by connecting IoT devices to the Internet. However, data transfer through the internet necessitates security measures, real-time data support, and flexibility of access. The internet managed to clear the path for cloud computing, which collects massive amounts of data for storage and processing. Cloud computing entails managing user interfaces, services, organizing and coordinating network nodes, computation, and data processing. The usage of cloud IoT platforms enables the storage of large amounts of data collected from sensors in the cloud. However, there are several critical technical criteria to consider while deploying IoT devices. The following factors should be examined for wireless connectivity: communication distance, data rate, battery life, mobility, latency, security and resilience, and gateway modem cost (Elijah et al., 2018).

Lately, a vast amount of data is generated from discrete sources, such as food safety monitoring systems, IoT, media, and other devices (Nychas et al., 2021; Sadiku et al., 2020; Tao et al., 2021). These data ought to be used for the betterment of the food supply chain and DS, undoubtedly, plays a vital role to fulfil this. Along with poor communication and coordination between the various participants in the food supply chain, customers’ attitudes regarding existing food standards are other major contributors to food loss and waste. The food industry extensively relies on regulatory inspection for monitoring the quality and safety of food where analyses are performed via conventional methods (e.g., International Organization for Standardization for total viable count). To maintain the safety standards of food, such processes are costly, timely, destructive to foods, retrospective, and preclude real-time information providence as per remaining shelf-life and safety throughout the foods’ lifecycle (Nychas et al., 2021).

Sensing is the source of all data in the IoT. With the deployment of numerous IoT devices, the agri-food sector generates a vast number of heterogeneous datasets in terms of content, structure, and storage type. BD is characterized by variability, variety, unstructured nature, noise, and excessive redundancy. To extract important information from such massive amounts of data, complex methods for data curation and storage, as well as expensive statistical methodologies and programming models, are required. Conditioning and preprocessing of primary data yield the knowledge needed to comprehend the state of the (agri-food) system. A system can be made capable of making independent localized judgments and taking appropriate actions by employing advanced algorithms and measuring the system's performance in relation to the desired outcome. This amount of autonomy in sensing, DM, and actuation is what distinguishes an IoT system as intelligent (Misra et al., 2022).

Millions of global monitoring data entries are stored in the Global Environment Monitoring System (GEMS/ food) database. Given the relatively high number of entries (600–800 entries per month), the data is organized logically and is easily retrievable. Information on chemical properties, microorganisms’ growth conditions, and weather data can be useful for food safety studies, for it can be utilized in models to anticipate the presence of hazards, such as mycotoxins in wheat, for instance. Data is retrieved in numerous ways, food safety authorities and food-related groups are already using social media platforms like Facebook, Twitter, and YouTube to communicate with the general public about food safety issues. Food agencies will have a better understanding of their audience and may discover new challenges by monitoring consumers' social media interactions. Approaches to web mining and social media analysis are being developed to use the massive amounts of data as an early warning system for identifying potential health and food safety hazards that could lead to a crisis. As per data storage, it generally is accomplished using data management systems such as MySQL, Oracle, and PostgreSQL. However, such systems are insufficient to handle large amounts of data. In these instances, much more speed, flexibility, and dependability are required than standard systems can provide. As a result, next-generation databases known as NoSQL have been developed that are nonrelational, open source, and horizontally scalable. Examples of such systems are MongoDB, Cassandra, and HBase. Following storage, the next difficulty is to move large amounts of data from many sources into a NoSQL cluster for processing. Transfer software is required for this, and examples of such software used to manage massive amounts of data include Aspera and Talend (Marvin et al., 2017).

To conclude, BDA assists companies in the food industry to understand customers’ needs and preferences by gathering real-time information about consumers, products, brands, and competitors. As a result, it plays a vital role in the expansion of the food industry. DA is a complicated process where raw data is translated into meaningful insights, that drives business decisions. It is just a matter of employing the proper tools and understanding what to do with the gathered data (Trends, 2021). For this, decision-makers leverage new technologies to further support their businesses. McCurdy (2022), claims that DA unlocks great value by increasing the employees’ efficiency, reducing waste, ensuring consistent quality, optimizing product portfolios, and finally becoming a DD organization.

Food safety is an indispensable part of our everyday lives, yet mournfully it is one of the greatest challenges facing the food industry (Tao et al., 2021). It is important to understand the potential risks associated with the food we consume to take the necessary steps to maintain safe food handling practices. Food-borne illnesses can be prevented by knowing which foods are safe to eat, how to store them correctly, and how to prepare them. To guarantee that our food is always safe to eat, it is also critical to be aware of the most recent news and updates regarding food safety. With the right knowledge and approach, we can all enjoy scrumptious and safe meals.

There is no universally accepted definition of food safety, however it could be described as the assurance that food will not cause harm to those who consume it when it is prepared, eaten, or stored as intended. On the other hand, food quality entails all other attributes that influence the products’ value to consumers (Nychas et al., 2021). Transparency and trust in food are paramount drivers for integrity control and improvements in efficiency and economic growth. Plaintively, all food commodities throughout the food chain cycle are prone to face numerous hazards, which results in a high contamination probability. For instance, biological and chemical hazards could be naturally present at any of the food production stages, whether accidentally introduced or fraudulently imposed, they both risk the consumers’ health and their trust in the food industry. Nowadays, in a world where “healthy nutrition” has surpassed “well-fed” as the prevailing paradigm of consumption, traditional food science has been unable to meet the rising demands for food. Howbeit, big data offers a solution to regulatory challenges by assisting businesses in better understanding customer demand and identifying patterns in the food industry (Tao et al., 2021).

Nowadays, to assess the quality or safety of raw or processed materials and food products, a variety of audits and inspections are performed (Margaritis, Madas & Vlachopoulou, 2022). This is largely due to the efficient design of the processes, products, and procedures used, where testing the final product (by analyzing certain hazards) is regarded as the production process's control measure. Both physicochemical and microbiological analyses are used to monitor food safety. These analyses are time-consuming and provide retrospective results. They are also expensive, occasionally require high-tech molecular tools, highly skilled personnel, and they are typically destructive to test products, therefore, limiting their ability to be used on, in, or at the line. Since it is physically, financially, and logistically impossible to inspect and sample 100% of food products, it is obvious that this strategy cannot adequately guarantee the consumer’s protection (Nychas, Panagou & Mohareb, 2016).

Marvin et al. (2017), offered extensive public information regarding food safety oversight and sampling inspection in a variety of countries. This material contains reports on animal and plant disease monitoring, risks, and food-borne diseases, as well as assistance for deep-risk researchers. The Rapid Alert System of Food and Feed (RASFF) is a widely used online food safety database in the EU for industry and scientific research. Other countries’ food safety databases include the Import Rejection Report (IRR) and the Inspection Classification Database (ICD) in the US, as well as the State Administration for Market Regulation (SAMR) warnings in China. The SAMR routinely posts its sampling inspection results of imported and exported food on official websites, allowing consumers to know the quality of food on the market. Likewise, the FSIS system is used by the US government to share food sample analysis reports. Additionally, the EFSA database contains information on food consumption trends and patterns within the EU (Tao et al., 2021).

The term “omics” refers to a broad range of scientific fields, including genomics (which deals with the effects of nucleotide variations within genes), transcriptomics (mRNA expression), metabolomics (levels of metabolites), and proteomics (which deals with the levels of peptides and proteins). The primary objective for constructing toxicogenomics-based predictive assays for chemical safety, and specifically for the aim of hazard detection, entails the creation of vast genomic databases from cell or animal exposure to recognized toxins. By using "animal-based" and in vitro (cellular) models, toxicogenomics seeks to understand the molecular processes that underlie the manifestation of toxicity and to generate molecular expression patterns (i.e., molecular biomarkers) that forecast toxicity in vitro and in vivo. This so-called read-across approach is based on the notion that identical physiological responses are determined by similar gene expression profiles, which are then used to determine the toxicological characteristics of a biological or chemical entity (Marvin et al., 2017). In addition, DNA sequences and DNA barcoding are thought to be enticing food fraud surveillance technologies. Analytical platforms have proven quite helpful because they offer significant details on the make-up of foods. Where AI, proteomics, and other new omics disciplines (such as metabolomics, lipidomics, volatilomics, and fingerprinting) have all shown to be useful tools in the food industry. Moreover, advanced high-throughput sequencing techniques, which are regarded as either non-targeted or targeted quick detection approaches, as well as non-destructive and non-invasive sensors have recently been utilized to gauge food safety (Nychas et al., 2021).

A challenging issue for food safety is the enormous volumes of data produced by multiple analytical and high-throughput platforms (Elijah et al., 2018). To provide a flexible method of keeping these very heterogeneous and varied data kinds, monitoring and real-time sensor data should be kept in a central repository with a semi-structured design. An adaptable, smartphone and cloud-enabled web interface can complement this data repository. To extract information from visual media, CNN mimics the human visual system. Color and shape are the primary characteristics of a picture that aid in identifying it. These characteristics can be extracted using mathematical techniques, which can then be automated using machine-learning techniques to collect information about regions of interest within an image. Moreover, edge detection enhances specific aspects of an image, such as gradient changes in pixel intensity along specific axes, by means of convolutional kernels. Different kernels may extract various types of visual data, including corners, straight lines, textures, and empty spaces, these properties are together utilized to create edges. These traits are expressed in an image subject profile that a neural network can utilize to determine the subject's class or kind (Nychas et al., 2021).

Unfortunately, there are multiple challenges facing the food industry, among which are limited shared data, lack of system standards, and the differences in the names and categories of the same food. The sharing and circulation of data among food regulatory departments aids in the development of intelligent food supply chain supervision. However, there is a limited data sharing, hence government departments have been unable to completely provide thorough monitoring and sampling data, causing firms to repeat testing of product quality indicators, resulting in significant waste of social resources and higher operating expenses. Additionally, the independence of departments further contributes to the relative independence of the food supervision system due to a lack of inadequate system standards. Furthermore, the limiting of responsibility increases the independence of departments. As a result, it is critical that a new manner of interdepartmental data sharing be investigated. Lastly, due to the inconsistent food standards, there are variances in the names and categories of the same item, which makes data sharing difficult (Tao et al., 2021).

Moreover, traceability can be achieved by real-time monitoring of production, supply, and distribution chains, which can be made available to primary production and final (products) food business operators and retailers via a live tracking system based on, for example, blockchain technology that combines mobile and web technologies. In an online food chain management platform, live monitoring mechanisms for labeled food products will be deployed to enable continuous food quality control. Because biological, chemical, and environmental dangers can develop, change, and differ in type and amount at different phases of food processing, sensitive sensors, microbiological testing, and metagenomics must be adopted at important production stages. The desired scientific breakthrough is the implementation of the aforementioned food safety tracking system in conjunction with DS, BD, data analytics, IoT concepts (defined as a network of devices that gather and transmit data via the Internet), and real-time data feeds acquired from various sensors under a common infrastructure where, for example, blockchain can be used to efficiently record transactions, execute dynamic actions via smart contracts, and so on (Nychas et al., 2021).

The diversity in agriculture has a significant impact on non-renewable resources, which are expected to be depleted in a few decades. To maintain consistency in input resources and ensure a consumer-friendly supply chain approach, this industry requires strict monitoring. Overuse of resources has resulted in global warming, which poses a serious threat due to CO2 emissions and deforestation. As a result of new environmental circumstances adaptability, food insecurity may rise, and resource scarcity may occur in many countries. To maintain efficiency and meet demand, selecting a practical sustainable modern technique would yield better outcomes. In response to consumer demand and to ensure quick production, the food industry has developed a limited set of food processing processes (Sharma et al., 2021).

Food commodities may be subjected to multiple risks along the food chain from farm to fork, which may induce physical, biological, or chemical contamination of food and, as a result, raise the risk of consuming contaminated food. These hazards, such as pathogenic microorganisms, mycotoxins, biogenic amines, or potentially carcinogenic substances such as caramel colors, have produced a long-term distrust in governments and industry among European consumers (Nychas, Panagou & Mohareb, 2016).

Finally, massive advances in computing power, storage capacity, and software have resulted in a rise in BD technology. This increase has raised various privacy issues. Similarly, it has led to an increase in the number of smart gadgets, and the rate of that expansion poses problems to society and peoples' freedom, notably in terms of security and privacy, as governments strive to set regulations, standards, and laws that define this progress (Ogbuke et al., 2022). According to Kshetri (2014), because there is so much data, security breaches and privacy violations are possible, which can have serious implications and losses, including reputational damage, legal liabilities, ethical problems, and other exacerbated technical issues. Hence, customers are increasingly concerned about the lack of privacy protection for their data provided by organizations' data gathering practices, particularly the use of monitoring technology such as cookies and GPS trackers (Ogbuke et al., 2022).

The synthesis provided by our literature review enables further analysis and insights regarding the future research directions in the intertwining of big data, analytics and food forming food analytics. Several studies have discussed the adoption barriers of innovative solutions in food and farming sector, when they use big data and analytics algorithms. These aspects may include but are not limited to trust, data privacy and security, data complexity, lack of adequate knowledge of IoT and its application especially among farmers, high cost for smart solutions when BDA comes into play, and so on. So, several important questions could be researched, such as:

To what extent increasing knowledge and awareness among food industry users my enhance the adoption rate of smart food and farming solutions?
What risks may smart food production using BDA create for users in terms of data privacy and security? How to overcome these adoption barriers?
How BDA may contribute to build trust for food industry users to increase the adoption of smart food and farming solutions (e.g., use of IoT sensors)

The increased sharing and exchange of data poses a significant challenge to the EU agri-food sector. Agriculture data has a very specialized and diverse nature (e.g., livestock and fish stock data, land and agronomic data, climate data, machine data, financial data, compliance data etc.). When we consider the entire supply chain, the problem gets much more confusing, because information about food origin, quality, and safety, including nutritional information, is mixed up with information about production, energy, distribution, and waste management. So, we need to address:

How to reap all the potential benefits of data sharing between many stakeholders? How to ensure a fair and transparent regulations?
How to ensure that the code of conduct for data sharing in agriculture provides the prerequisites for defining the soft infrastructure based on contractual agreements, as well as guidelines on fair and transparent data use?

To achieve a European progress, a set of data and system interoperability protocols and standards must be agreed upon. This will aid in avoiding being tethered to existing platform architectures. One critical problem is whether farm management system (FMS) manufacturers are willing to allow interoperability and federation of their data platforms with other systems. So, there is a need-to-know answer to several questions, such as:

How to enable interoperability and data platforms federation of the food industry and ecosystem?
What are, and how to implement, a set of design principles allowing the creation and maintenance of data space in food industry where data is governed, managed and exchanged seamlessly and safely?

Even though some studies have discussed utilization of AI techniques and the intelligent optimization algorithm for food production management, further research is necessary to address these kinds of managerial aspects, when food industry benefits from AI based solutions. Therefore, the question that should be answered is:

How and in what ways BDA and more particularly AI based techniques may contribute to the smart food industry in managerial level (concerning the processes and so on)

Some of the reviewed studies highlighted the fact that using BDA and smart solutions such as IoT and AI technologies may affect the quality of the produced food. This in turn will affect the quality of humans’ life in the long run. Therefore,

What are the direct and indirect effects of using BDA on quality of life for humans in a long-term period?

When it comes to environmental sustainability, several studies have highlighted the importance of applying BDA in the food industry. However, these articles have not thoroughly investigated this aspect and further research is required in this regard. For example,

In what ways does employing BDA-based techniques such as ML, IoT, or cloud-based analytics in the food industry affect environmental sustainability?

Regarding prediction of the food situation, various studies have discussed this from various perspectives such as prediction of customer behavior, predictive modeling risk assessment for food production, IoT-based predictive analytics, and so forth. While most of them have discussed “why” it should be done, there is a need for a more comprehensive study regarding how these predictions can be used in the decision-making processes in food industries in different process phases. Thus, the questions are:

How the results of applying BDA in food industry may help us to make DDD to mitigate risks in food industry
How IoT-based predictive analytics in the food industry may empower DD decision-making?

Agriculture is entering a digitally enhanced farming era, with data generated at various phases of agricultural production and related operations, and IoT solutions, robotics, and sophisticated decision-support systems being used. Data-driven approaches provide unprecedented opportunities to 1) improve resource efficiency, productivity, and environmental sustainability; 2) dynamically adapt business plans to changing markets and consumer expectations; 3) reduce administrative costs and enable science-based policies; 4) provide more prosperous living conditions for rural communities; and 5) improve the consumer's relationship with the various actors across the value chain. This study provides a comprehensive review of the application of big data, data science, AI, ML algorithms, and IoT in the food industry, and discovers areas which require further research efforts and investigation, as well as posing questions for researchers, practitioners and governmental food agencies and officials, with regards to the technical challenges, the privacy and security concerns, and the legislative restraints. The review conducted in this study is invaluable in helping to advance the food industry, and its findings could be used to inform future developments in the industry, in a cross-disciplinary fashion.

Disclosure Statement

The authors report there are no competing interests to declare.

Author Contribution

R.E. Conceptualization, Review the articles, Writing original draft sections introduction, Food systems, research gapA.E. Theoretical background, supervisionA.H. Methodology, Research agendaAll authors: review and revise the paper in several iterations

Acknowledgments

This work was supported by Luleå University of Technology under Grant (383211).

Aazam M, Hung PP, Huh EN (2014) Cloud of Things: Integrating Internet of Things with Cloud Computing and the Issues Involved. Applied Sciences and Technology (IBCAST)
Agarwal R, and Dhar V (2014) Big Data, Data Science, and Analytics: The Opportunity and Challenge for IS research. Inform Syst Res 25(3):443–448
Ahalt S (2013) Why Data Science? Presented at the National Consortium for Data Science. Chapel Hill
Altintzoglou T, Honkanen P, Whitaker RD (2021) Influence of the Involvement in Food Waste Reduction on Attitudes towards Sustainable Products Containing Seafood By-products. J Clean Prod 285:125487. https://doi.org/10.1016/j.jclepro.2020.125487
Amankwah-Amoah J, Adomako S (2019) An Organizing Framework. Comput Ind 105:204–212. https://doi.org/10.1016/j.compind.2018.12.015. Big Data Analytics and Business Failures in Data-rich Environments:
Arif MA, Verstraete W (1995) Methane Dosage to Soil and its Effect on Plant Growth. World J Microbiol Biotechnol 11(5):529–535. https://doi.org/10.1007/BF00286368
Ashford W (2015), February 4 IoT Could be Key to Farming, says Beecham Research. Accessed January 02, 2023. https://www.computerweekly.com/news/2240239484/IoT-could-be-key-to-farming-says-Beecham-Research
Boyd D, Crawford K (2012) Critical Questions for Big Data: Provocations for a Cultural, Technological, and Scholarly Phenomenon. Inform Communication Soc 15(5):662–679
Ciccullo F, Fabbri M, Abdelkafi N, Pero M (2022) Exploring the Potential of Business Models for Sustainability and Big Data for Food Waste Reduction. J Clean Prod 340:130673. https://doi.org/10.1016/j.jclepro.2022.130673
Clark J (2020), August 28 What is the Internet of Things, and how does it work? IBM Business Operations Blog. Accessed February 09, 2023. https://www.ibm.com/blogs/internet-of-things/what-is-the-iot/
Cutting Agricultural Methane Emissions Can Help Save the Planet while Increasing Yields and Improving Lives. (2021) Climate & Clean Air Coalition. Accessed December 14, 2022. https://www.ccacoalition.org/en/news/cutting-agricultural-methane-emissions-can-help-save-planet-while-increasing-yields-and
Dhar V (2013) Data Science and Prediction. Commun ACM 56(12):64–73
Elijah O, AbdulRahman T, Orikumhi I, Leow CY, Nour-Hindia MHD (2018) An Overview of Internet of Things (IoT) and Data Analytics in Agriculture: Benefits and Challenges. IEEE Internet Things J 5(5):3758–3773. https://doi.org/10.1109/JIOT.2018.2844296
Elgendy N, Elragal A (2014) Big Data Analytics: A Literature Review Paper. The 14th Industrial Conference on Data Mining (ICDM). Petersburg: Springer-LNCS
Elgendy N, Elragal A, Päivärinta T (2021) DECAS: A Modern Data-driven Decision Theory for Big Data and Analytics. Journal of Decision Systems, 337 – 73. https://doi.org/10.1080/12460125.2021.1894674
Elragal A, Klischewski R (2017) Theory-driven or Process-driven Prediction? Epistemological Challenges of Big Data Analytics. J Big Data 4–19. https://doi.org/10.1186/s40537-017-0079-2
FAO (2021) https://www.fao.org/state-of-food-security-nutrition
Garbero A, Carneiro B, Resce G (2021) Harnessing the Power of Machine Learning Analytics to Understand Food System Dynamics Across Development Projects. Technological Forecast Social Change 172:121012. https://doi.org/10.1016/j.techfore.2021.121012
Government Offices of Sweden (2016) A National Food Strategy for Sweden: More Jobs and Sustainable Growth throughout the Country. Accessed on April 03, 2023. https://www.government.se/contentassets/16ef73aaa6f74faab86ade5ef239b659/livsmedelsstrategin_kortversion_eng.pdf
Hinge G, Surampalli RY, Goyal MK, Gupta BB, Chang X (2021) Secur Environ Manage Technological Forecast Social Change 169:120823. https://doi.org/10.1016/j.techfore.2021.120823. Soil Carbon and its Associate Resilience using Big Data Analytics: For Food
HLPE A (2017) Nutrition and Food Systems. A Report by the High Level Panel of Experts on Food Security and Nutrition of the Committee on World Food Security. CFS, Rome
Internet of Food and Farm 2020. (n.d.). IoF2020. Accessed March 08, (2023) https://www.iof2020.eu/
Internet of Food and Farm 2020. (2022) CORDIS: European Commission. Accessed March 08, 2023. https://cordis.europa.eu/project/id/731884
Isabelle D, Westerlund M, Mane M, Leminen S (2020) The Role of Analytics in Data-Driven Business Models of Multi-Sided Platforms: An Exploration in the Food Industry. Technol Innov Manage Rev 10(7):4–15. http://doi.org/10.22215/timreview/1371
Jones M (2019) What We Talk About When We Talk About (Big) Data. J Strategic Inform Syst 28:3–16. https://doi.org/10.1016/j.jsis.2018.10.005
Kelleher JD, Tierney B (2018) What is Data Science? Data Science, pp 1–38
Kshetri N (2014) Big Data’s Impact on Privacy, Security and Consumer Welfare. Telecomm Policy 38(11):1134–1145. https://doi.org/10.1016/j.telpol.2014.10.002
Lau RY, Zhao JL, Chen G, Guo X (2016) Big Data Commerce. Inform Manage 53(8):929–933. https://doi.org/10.1016/j.im.2016.07.008
Luckin R, Holmes W, Griffiths M, Forcier LB (2016) Intelligence Unleashed. An Argument for AI in Education. Pearson, London
Manyika J, Chui M, Brown B, Bughin J, Dobbs R, Roxburgh C, Byers AH (2011) Big Data: The Next Frontier for Innovation, Competition, and Productivity. McKinsey Global Institute Reports, 1– 156. McKinsey Global Institute
Margaritis I, Madas M, Vlachopoulou M (2022) Big Data Applications in Food Supply Chain Management: A Conceptual Framework. Sustainability 14:4035. https://doi.org/10.3390/su14074035
Marvin HJP, Janssen EM, Bouzembrak Y, Hendriksen PJM, Staats M (2017) Big Data in Food Safety: An Overview. Crit Rev Food Sci Nutr 57(11):2286–2295. http://dx.doi.org/10.1080/10408398.2016.1257481
McAfee A, Brynjolfsson E (2012) Big Data: The Management Revolution. Harvard Business Rev 90:3–9
McCurdy J (2022) How Data Analytics in the Food Industry Unlocks Value. Aptean. Accessed April 01, 2023. https://www.aptean.com/en-US/insights/blog/data-analytics-in-food-industry
Merriam-Webster (2022) Definition of Scientific method. Accessed January 19, 2023. https://www.merriam-webster.com/dictionary/scientific%20method
Mikalef P, Pappas IO, Krogstie J, Giannakos M (2018) Big Data Analytics Capabilities: A Systematic Literature Review and Research Agenda. IseB 16(3):547–578. https://doi.org/10.1007/s10257-017-0362-y
Misra NN, Dixit Y, Al-Mallahi A, Bhullar MS, Upadhyay R, Martynenko A (2022) IoT, Big Data, and Artificial Intelligence in Agriculture and Food Industry. IEEE Internet Things J 9(9):6305–6324
Mourad M (2016) Recycling, Recovering and Preventing Food Waste: Competing Solutions for Food Systems Sustainability in the United States and France. J Clean Prod 126:461–477. https://doi.org/10.1016/j.jclepro.2016.03.084
Murphy RR (2000) Introduction to AI Robotics. The MIT, London
Nagel L, Lycklama D (2021) Design Principles for Data Spaces (1.0). Zenodo, 70–73. https://doi.org/10.5281/zenodo.5244997
Nychas GE, Panagou EZ, Mohareb F (2016) Novel Approaches for Food Safety Management and Communication. Curr Opin Food Sci 12:13–20. http://dx.doi.org/10.1016/j.cofs.2016.06.005
Nychas GE, Sims E, Tsakanikas P, Mohareb F (2021) Data Science in the Food Industry. Annual Rev Biomedical Data Sci 4:341–367. https://doi.org/10.1146/annurev-biodatasci-020221-123602
Ogbuke NJ, Yusuf YY, Dharma K, Mercangoz BA (2022) Ethical, Privacy and Security Challenges Posed to Business, Industries and Societies. Prod Plann Control 33:123–137. https://doi.org/10.1080/09537287.2020.1810764. Big Data Supply Chain Analytics:
Parvin S, Venkatraman S, Souza-Daw T, Fahd K, Jackson J, Kaspi S, Cooley N, Saleem K, Gawanmeh A (2019) Smart Food Security System using IoT and Big Data Analytics. 16th International Conference on Information Technology-New Generations (ITNG 2019), 253 – 58. https://doi.org/10.1007/978-3-030-14070-0_35
Paul B, Agnihotri S, Kavya B, Tripathi P, Babu CN (2022) Sustainable Smart Aquaponics Farming Using IoT and Data Analytics. J Inform Technol Res 15(1):1–27. https://doi.org/10.4018/jitr.299914
Peters K, Silva S, Wolter TS, Anjos L, Ettekoven N, Combette E, Melchiori A, Fleuren H, Hertog D, Ergun Ö (2022) UN World Food Programme: Toward Zero Hunger with Analytics. Informs J Appl Analytics 52(1):8–26. https://doi.org/10.1287/inte.2021.1097
Pitts J, Gopal S, Ma Y, Koch M, Boumans RM, Kaufman L (2020) Leveraging Big Data and Analytics to Improve Food, Energy, and Water System Sustainability. Front Big Data 3(13). https://doi.org/10.3389/fdata.2020.00013
Rizk A, Elragal A (2020) Data Science: Developing Theoretical Contributions in Information Systems via Text Analytics. J Big Data 7(1). https://doi.org/10.1186/s40537-019-0280-6
Russom P (2011) Big Data Analytics. TDWI 4th Quarter, 1–38
Sadiku MNO, Ashaolu TJ, Ajayi-Majebi A, Musa SM (2020) Big Data in Food Industry. Int J Sci Adv 1(3):148–152
Saggi MK, Jain S (2018) A Survey Towards an Integration of Big Data Analytics to Big Insights for Value-creation. Inf Process Manag 54(5):758–790. https://doi.org/10.1016/j. ipm.2018.01.010
Sharma S, Gahlawat VK, Rahul K, Mor RS, Malik M (2021) Sustainable Innovations in the Food Industry through Artificial Intelligence and Big Data Analytics. Logistics 5:66. https://doi.org/10.3390/logistics5040066
Shelhamer E, Long J, Darrell T (2017) Fully Convolutional Networks for Semantic Segmentation. IEEE Trans Pattern Anal Mach Intell 39(4):640–651. https://doi.org/10.1109/tpami.2016.2572683
Smith N (2016) Food Policy. Reference Module in Food Sciences. https://doi.org/10.1016/b978-08-100596-5.03428-4
Tao Q, Ding H, Wang H, Cui X (2021) Application Research: Big Data in Food Industry. Foods 10(9):2203. https://doi.org/10.3390/foods10092203
The 17 Goals | Sustainable Development. (n.d.). United Nations. https://sdgs.un.org/goals
Tirgul C, Naik M (2016) Artificial Intelligence and Robotics. Int J Adv Res Comput Eng Technol 5:1787–1793
Trends M (2021) Numbers and Orders: The Role of Data Analytics in the Food Industry. Analytics Insight. Accessed April 01, 2023. https://www.analyticsinsight.net/numbers-and-orders-the-role-of-data-analytics-in-the-food-industry/
Wamba SF, Gunasekaran A, Akter S, Ren SJ, Dubey R, Childe SJ (2016) Big Data Analytics and Firm Performance: Effects of Dynamic Capabilities. J Bus Res 70:356–365. https://doi.org/10.1016/j.jbusres.2016.08.009
What is the Internet of Things (IoT)? (n.d.). Accessed February 03, (2023) https://www.oracle.com/internet-of-things/what-is-iot/
Zhou L, Zhang C, Liu F, Qiu Z, He Y (2019) Application of Deep Learning in Food: A Review. Compr Rev Food Sci Food Saf 18(6):1793–1811. https://doi.org/10.1111/1541-4337.12492

No competing interests reported.

AppendixA.docx

Download PDF

Version 1

posted

You are reading this latest preprint version

A Literature Review of Food Analytics

Status:

Version 1

Abstract

Figures

1. Introduction

2. Method

3. Theoretical Background

3.1 Big Data

3.2 Data Science

3.3 Artificial Intelligence and Machine Learning

3.4 Internet of things (IoT)

4. Food System

4.1 The Food Context

4.2 Food Ecosystem

4.3 Food Shortage

4.4 Food Safety

4.5 Institutions & Food

4.6 Policy-making in Food

5. Food Analytics

5.1 Big Data in Food Analytics

5.2 Data Science in Food Analytics

5.3 Artificial Intelligence & Machine Learning in Food Analytics

5.4 IoT in Food Analytics

6. Research Gaps

7. Research Agenda

Declarations

Author Contribution

Acknowledgments

References

Additional Declarations

Supplementary Files

Status:

Version 1