Sentiment analysis is concerned with a text's polarity (positive, negative, or neutral), but it may also detect specific moods and emotions (angry, joyful, sad, etc.), urgency (urgent, not urgent), and even intents (interested v. not interested). Social media sites such as Twitter, Facebook, and YouTube are becoming a valuable source of data known as social data (Manguri et al., 2020). Sentiment analysis can help individual or an organization to make decisions for further instructions. Below are types of sentiment analysis which are used to express opinions on issues such as policies, tax, etc.
2.1 Fine-grained Sentiment Analysis
Graded Sentiment is used to understand ratings. The ratings are expressed based how satisfied a person shows interest and/or dissatisfaction towards a policy. Ratings can be grouped as follows;
5 – Positive
4 –Very Positive
3 – neutral
2 – Very negative
1 – Negative
In fined grained sentiment analysis, 5 is expressed as highly positive (a person is more interested), 4 means a positive sentiment, 3 represents neutral (a person is neither against nor satisfied, and or have not understand what the content at hand means), 2 is not satisfied (very negative) with the content and 1 represents a negative sentiment. A fined grained sentiment analysis was carried out by Wang et al. (2017) described a social media analytics engine that employs a social adaptive fuzzy similarity-based classification method to automatically classify text messages into sentiment categories (positive, negative, neutral and mixed), with the ability to identify their prevailing emotion categories (e.g., satisfaction, happiness, excitement, anger, sadness, and anxiety). Their paper also embedded within an end to-end social media analysis system that has the capabilities to collect, filter, classify, and analyze social media text data and display a descriptive and predictive analytics dashboard for a given concept. In a research by Munikar et al. (2019) used a promising deep learning model called BERT to solve the fine-grained sentiment classification task. Their experiments showed that their model outperforms other popular models for this task without sophisticated architecture. They also demonstrated the effectiveness of transfer learning in natural language processing in the process.
2.2 Emotion detection
Emotion detection sentiment analysis deals with interpreting emotions like happiness, frustration, anger, and sadness. Emotion detection systems frequently employ lexicons, which are collections of words that express specific emotions. Robust machine learning (ML) algorithms are also used by some advanced classifiers. Researchers Sagum et al, (2021) experimented on emotion detection result to be used in sentiment analysis. The emotions that were included in their research are happiness, sadness, anger, and fear. Once emotion was detected the system will then use it to know the sentiment of the person on a particular movie. Their paper aimed to measure the accuracy in sentiment analysis enhanced by emotion detection and to know whether emotion detection plays a key role in reading sentiment analysis. Kusal et al. (2021) also developed an Ai based emotion detection on big textual data. They considered 827 Scopus and 83 Web of Science research papers from the years 2005–2020 for the analysis. Their qualitative review represented different emotion models, datasets, algorithms, and application domains of text-based emotion detection. Their quantitative bibliometric review of contributions presents research details such as publications, volume, co-authorship networks, citation analysis, and demographic research distribution. In the end, challenges and probable solutions were showcased, which can provide future research directions in the area.
2.3 Aspect-Based Sentiment Analysis
When analyzing text sentiments, you will usually want to determine the specific qualities or features people are referencing in a favorable, neutral, or negative light. Aspect-based analysis entails a more in-depth investigation. It assists you in determining which components of the conversation are being discussed. It takes into consideration the whole sentence or text. In a related work on aspect-based sentiment analysis, Alqaryouti et al. (2019) proposed an aspect-based sentiment analysis hybrid approach that integrates domain lexicons and rules to analyze the entities smart apps reviews. The proposed model aimed to extract the important aspects from the reviews and classify the corresponding sentiments. This approach adopted language processing techniques, rules, and lexicons to address several sentiment analyses challenges, and produce summarized results. According to the reported results, the aspect extraction accuracy improves significantly when the implicit aspects are considered. In another work, Hoang et al. (2019) showed the potential of using the contextual word representations from the pre-trained language model BERT, together with a fine-tuning method with additional generated text, in order to solve out-of-domain ABSA and outperform previous state-of-the-art results on SemEval-2015 (task 12, subtask 2) and SemEval-2016 (task 5). According to the researchers, no other existing work has been done on out-of-domain ABSA for aspect classification.
2.4 Intent Analysis
The intent analysis can assist you figure out whether a customer is looking to buy anything or is just looking around. If a customer is willing to make a purchase, you can monitor them and market to them. You can save time and money by not advertising to customers who aren't ready to buy. Lye & The (2021) analyzed data on customer Feedback in the form of Net Promoter Score (NPS) with a text box and demonstrated a hybrid representation that resulted in the accuracy improvement of the sentiment classification task and predicting customer intent. Their datasets were first trained using Word2Vec with the previous dataset and then fit into the Random Forest classifier, tested as the best configuration to prevent overfitting. The hybrid representation was compared against the baseline sentiment polarity tool through few experiments; the results showed that the hybrid model has improved accuracy for the sentiment classification task. Lastly, they performed customer intent prediction by using the Power BI influencer module.
2.5 Multilingual Sentiment Analysis
Helps detect language in texts automatically with a language classifier, then train a custom sentiment analysis model to classify texts in the language of your choice. Coding experience is required since it is difficult. Atiqah et al. (2021) in a study provided a systematic literature review on multilingual sentiment analysis, which summarized the common languages supported in multilingual sentiment analysis, pre-processing techniques, existing sentiment analysis approaches, and evaluation models that have been used for multilingual sentiment analysis. By following the systematic literature review, their findings revealed, most of the models supported two languages, and English is seen as the most used language in sentiment analysis studies. In a related study, Dashtipour et al. (2016) presented a state-of-the-art review on multilingual sentiment analysis. More importantly, they compared their implementation of existing approaches on common data. Precision observed in their experiments was typically lower than the one reported by the original authors, which they attributed to the lack of detail in the original presentation of those approaches.