Algorithms of Intelligent Language Recognition and English Discourse Analysis Based on Fisher's Three-dimensional Model

doi:10.21203/rs.3.rs-2445963/v1

Download PDF

Research Article

Algorithms of Intelligent Language Recognition and English Discourse Analysis Based on Fisher's Three-dimensional Model

https://doi.org/10.21203/rs.3.rs-2445963/v1

This work is licensed under a CC BY 4.0 License

You are reading this latest preprint version

As an important analysis theory in critical discourse analysis, Fischer's three-dimensional analysis model has been used more and more to analyze news discourse. The Fischer three-dimensional analysis model includes vocabulary and sentence descriptions in text exercises, as well as messages in discourse exercises, describing the source of events and discourse practices, and explaining international and domestic backgrounds. The development of intelligent language recognition technology is supported by people from all walks of life, and its development is very rapid. Many research results have been extended from theoretical research to daily applications. The convenience of intelligent speech recognition technology has greatly changed our way of life, and intelligent speech recognition technology has been able to recognize human voices by recognizing discrete symbols. At present, the research on classroom discussion is mainly based on the direct observation experience of researchers on classroom teaching. In the environment of rapid development of information, we will find more effective classroom discourse analysis methods, provide fast and convenient classroom discourse analysis skills, and make classroom discourse analysis more efficient. The innovation of classroom discourse research methods in an informed environment provides convenience for English language analysis in an intellectual classroom environment. The application of English language makes various applications of English in intelligent recognition technology systems more and more extensive.

Fisher's three-dimensional analysis model

Intelligent language recognition

English language

Discourse analysis

Fairclough's three-dimensional analysis model provides comprehensive analysis steps for discourse analysis in practic. Because of this, many foreign experts and scholars have used three-dimensional analysis models as the basis for critical analysis of news discourse [1–2]. For a period of time in China, "people pay little attention to -critical linguistics, or even understand very little. Foreign language teaching always uses language as a pure way of conveying information, no matter how much its understanding of social process and political interference is, how can we manage a stable society? Economic and cultural life is the top priority [3]. Therefore, the role of morphology has attracted the attention of foreign scholars and experts. "However, in recent years, critical linguistics has also attracted the attention of the language community in my country. Indigenous research on critical discourse analysis has shown encouraging achievements [4]. The start of all computer language systems using statistical techniques today, this is because language and discourse are the same to some extent. Recognition based on statistical models has high accuracy and semantic accuracy [5]. In intelligent computer recognition technology, T-type has a wide range of applications related to voice interaction, simultaneous translation, voice control, voice recognition, text and voice intelligence, and provides people with a variety of beautiful services [6]. In order to make better use of intelligent computer recognition technology, experts at home and abroad have worked tirelessly to find effective methods and implement them [7]. In recent years, continuous efforts and in-depth understanding of advanced recognition English have further improved the speech form and uncertainty analysis skills, which play an important role in language translation and effectively improve the level of speech English translation. In order to recognize text intelligent characters, it mainly involves three processes: pre-recognition processing, recognition processing and post-recognition processing [8]. In 1952, the American structural linguist Z.S. Harris published the first article on "discourse analysis", which unveiled the veil of discourse analysis. Since then, the term "discourse analysis" has also become a professional term in modern linguistics, and a series of studies on discourse analysis have also begun to rise and develop [9]. Due to the foundation of discourse analysis, classroom discourse analysis also developed rapidly in the 1960s. It examines the form and sequence of classroom language to determine how to improve the teaching effect of the entire classroom, the structure of teacher speech and teaching, so the analysis of classroom discourse is more comprehensive and effective [10].

The tweet data learned in the literature comes from the Trump Twitter Archive, which mainly collects and classifies Trump's tweets. Trump signed a presidential memorandum on March 22, 2020, deciding to impose tariffs on Chinese imports and restrict investment and mergers with Chinese and American companies. As a result, the Sino-US trade crisis has deepened. Therefore, the research data set is from March 22, 2020 to September 4, 2020. Literature believes that the federal decentralized political system and the bicameral political system in the United States are democratic and free, and there are misunderstandings and concepts about the socialist system [11–12]. Political statements like this are in line with public opinion that attracts the people, anti-establishment and anti-establishment structures. Literature compared the choices of the two media on different topics and found that their positions in the Libyan government and the opposition are completely different [13]. Even if they are not directly elaborated, they are in three main languages. The quick use function can also implicitly reflect various prejudices in the local media. Literature fully shows the different positions of the United States and China in the Libyan war [14]. Natural language research based on computer cognitive technology has multiple applications in grammatical speech, simultaneous translation, voice control, speech recognition, text analysis and speech recognition, and provides people with various appropriate services and voice intelligence as an example presented in the application [15].

3.1. Principles of Fischer's three-dimensional analysis model

Fairclough's three-dimensional analysis model refers to the text analysis integration proposed by Fairelough, which focuses on language-related political considerations. Fairclough believes that speech is the product of verbal communication or the process of practice. This process is accompanied by the creation, dissemination and reception of speech. All of this depends on the specific circumstances of the community. Based on Fairclough, it outlines the three stages of speech analysis: (1) Describe the characteristics of the text form structure, analyze the text form structure, including words, grammar and text structure. (2) Explain the importance of speech and the process of producing, disseminating and receiving speech; (3) Describe the relationship between the social environment in the communication process. Figure 1 shows the Three-dimensional conception of discourse.

Fairclough's three-dimensional analysis model provides comprehensive analysis steps for critical analysis of speech in practice. It is precisely because of this concept that many foreign experts and scholars have used i-measurement analysis as an in-depth analysis method for news statements. For some time in the country, “people pay little or no attention to important linguistics.”

3.2. Lip-based language feature extraction

It can be seen from Fig. 1 that the curve library has a better effect than the curve on the right (the first paragraph and the third row, and the appropriate curve library has a better effect than its rectangle. (The second and fourth rows) Therefore, the upper lip is selected for two-dimensional placement. At present, the appropriate curve on the surface of the lips has been determined. The curve is separate: long lips from the upper left corner of the mouth to the center of the main point, from the upper left corner of the mouth to the center of the main point, use the appropriate curve, and from the bottom to the left lip, use the appropriate curve: time correction curve, The expressions of the two curves are as follows:

$$\{\begin{array}{c}y={a}_{1}{x}^{3}+{a}_{2}{x}^{2}+{a}_{3}x+{a}_{4}\\ y={b}_{1}{x}^{3}+{b}_{2}{x}^{2}+{b}_{3}x+{b}_{4}\\ y={c}_{1}{x}^{2}+{c}_{2}x+{c}_{3}\end{array}$$

Between them, a and b (i = 1,2,3,4) are the two corners of the long-lip curve c; (i = 1,2,3) are the boundaries of the lower lever cycle.

Figure 2 shows Lip model.

Compared with the direct use of the length and width of the lips in the previous literature, this article believes that the proportion of the side shape of the lips reflects the language information and will be louder when the lips move. This article uses the following 4 borders to introduce the basic shape of lips:

(1) The ratio of the height and width of the outer contour of the lips R1

$${R}_{1}=\frac{{H}_{o}}{{W}_{o}}$$

(2) The ratio of the upper and lower heights of the outer contour of the lips R2

$${R}_{2}=\frac{{H}_{ou}}{{H}_{od}}$$

(3) The ratio of the height and width of the inner contour of the lips R3

$${R}_{3}=\frac{{H}_{i}}{{W}_{i}}$$

(4) The ratio of the upper and lower heights of the inner contour of the lips R4

$${R}_{4}=\frac{{H}_{i\nu }}{{H}_{id}}$$

Among them, the meanings of H_o, W_o, H_ou, H_od, H_i, W_i, H_iu, and H_i are the height of the outer contour of the lips, the width of the outer contour of the lips, the upper height of the outer contour of the lips, the lower height of the outer contour of the lips, the height of the inner contour of the lips, and the inner contour of the lips. The width, the height of the upper part of the inner contour of the lips and the lower height of the inner contour of the lips.

RI, R2, R3 and R4 constitute the first half of the geometric visual features, namely

$${V}_{1}=({R}_{1},{R}_{2},{R}_{3},{R}_{4})$$

The structure of the growth scale and length may have an inner part which is the main point of the lip information. In order to merge all important information, the boundaries of the three lip regions are selected as part of the geometric vector segment, namely:

$${V}_{2}=({a}_{1},{a}_{2},{a}_{3},{a}_{4},{b}_{1},{b}_{2},{b}_{3},{b}_{4},{c}_{1},{c}_{2},{c}_{3})$$

So far, the geometric visual feature vector is obtained as follows:

$${V}_{g}=({V}_{1},{V}_{2})$$

(1) Discrete Fourier transform

Discrete Fourier Transform can quickly and efficiently process and analyze images. By changing the image of the lip part in the repeated part, different information in the image is analyzed to obtain information about the lip image. Fourier disc structure:

$${F}_{u,v}=\frac{1}{MN}\sum _{x=0}^{M-1}\times \sum _{y=0}^{N-1}\times {P}_{x,y}{e}^{-j2\pi \left(\frac{ux}{M}+\frac{w}{N}\right)}$$

Among them, M and N are the height and width of the image, Pxy are the pixels of the image, x and y are the pixel coordinates, the value after Fuv Fourier transform, and u and v represent the frequency in the horizontal and vertical directions, respectively. The inverse transformation of the two-dimensional Fourier transform from the frequency domain back to the image is:

$${P}_{x,y}=\sum _{u=0}^{M}\pm \sum _{v=0}^{N}\times {F}_{u,v}{e}^{j2\pi \left(\frac{ux}{M}+\frac{v}{N}\right)}$$

The result of the Fourier transform of a real function is a complex number:

$${F}_{u,v}={R}_{u,v}+j{I}_{1,v}$$

(2) Discrete Hartley transform

The Hartley transformer is a so-called Fourier transform, but it does not require complicated calculations. The advantage is that the anterior and posterior eye circles perform the same arithmetic. The same face is defined as a series of repeated segments. Discrete Hartley is described as follows

$${H}_{u,v}=\frac{1}{N}\sum _{x=0}^{N-1}\times \sum _{y=0}^{N-1}\times {P}_{x,y}\times (\text{c}\text{o}\text{s}(\frac{2\pi }{N}\times (ux+vy))+\text{s}\text{i}\text{n}(\frac{2\pi }{N}\times (ux+vy)\left)\right)$$

The inverse Hartley transform is applied to perform the same processing on the transformed lips image:

$${P}_{x,y}=\frac{1}{N}\sum _{x=0}^{N-1}\times \sum _{y=0}^{N-1}\times {H}_{u,v}\times (\text{c}\text{o}\text{s}(\frac{2\pi }{N}\times (ux+vy))+\text{s}\text{i}\text{n}(\frac{2\pi }{N}\times (ux+vy)\left)\right)$$

Hartley changes rapidly. By comparing Hartley with Hu,v as a function, different Fourier changes can be calculated.

However, compared with Fourier's own changes, Hartley's changes did not transfer changes, but the problem can be solved in different ways.

(3) Discrete cosine transform

Discrete cosine transform is a real number transform. The two-dimensional DCT transform formula of an image is as follows:

$$F(u,v)=\frac{2C\left(u\right)C\left(v\right)}{N}\sum _{x=0}^{N-1}\times \sum _{y=0}^{N-1}\times f(x,y)\text{c}\text{o}\text{s}\frac{\left(2x+1\right)u\pi }{2N}\text{c}\text{o}\text{s}\frac{\left(2y+1\right)v\pi }{2N}$$

The corresponding inverse transformation formula is:

$$f(x,y)=\frac{2}{N}\sum _{u=0}^{N-1}\times \sum _{v=0}^{N-1}\times C\left(u\right)C\left(v\right)F(u,v)\text{c}\text{o}\text{s}\frac{\left(2x+1\right)u\pi }{2N}\text{c}\text{o}\text{s}\frac{\left(2y+1\right)v\pi }{2N}$$

Where u,v = 0,1,2,N-1; x,y = 0,1,2...N-1:

$$C\left(u\right)C\left(v\right)=\{\begin{array}{cc}\frac{1}{\sqrt{2}}& u,v=0\\ 1& u,v=\text{1,2},\dots ,N-1\end{array}$$

It can be seen from the DCT formula that DCT has many advantages in terms of energy concentration. After the image is modified by DCT, the energy is mainly concentrated in the low frequency part. In this way, DCT can always be used for the same visual quality requirements. It provides a very high push ratio. In addition, like FFT, DCT also has rapid changes.

There are two ways to perform DCT on an image: treat the image as a whole for global DCT transformation and divide the image into several sub-regions for block DCT transformation, as shown in Fig. 3.

After obtaining the normalized lip image, divide the image into 6 equally, and the size of each small area is 10*10. Perform DCT transformation on each small area, as shown in Fig. 4.

$${V}_{p}=({v}_{s1},{v}_{s2},{v}_{s3},{v}_{s4},{v}_{s5},{v}_{s6})$$

Where v*=(rn > Cz,C3sSCq,Cis,Ci6,C;)i = 1,2,3,4,5,6,Ca(k = 1,2,3,4,5,6,7) Is the coefficient extracted from each subregion. In this way, a total of 7X6 = 42-dimensional data can be obtained as the feature vector of the lip area.

In the previous two subsections, the geometric model feature V and the pixel feature V0 of the lip area were obtained respectively. Combining V0 and V can get the mixed feature vector of the lips:

$${V}_{lip}=({V}_{g},{V}_{p})$$

Since the visual vector discovered later has many value differences in each dimension, and even different shapes, the detection accuracy may be damaged. Therefore, the value of the vector segment needs to be adjusted to understand the difference between the values of different segments. Since W, Y_i and V are actually very different shapes, the three different shapes are normalized as follows:

$${v}^{*}=\frac{v-{v}_{min}}{{v}_{max}-{v}_{min}}$$

$${V}_{1}={V}_{1}^{*}$$

$${V}_{2}={V}_{2}^{* }$$

$${V}_{p}={V}_{p}^{*}$$

Finally, the visual feature vector of the lips is as follows:

$${V}_{s}={V}_{lip}^{*}=\left({V}_{1}^{*},{V}_{2}^{*},{V}_{p}^{*}\right)$$

When speaking, each language has a variable form. If you can imagine a specific image of the lips, some information about the movement of the lips will be lost. The first part of the variance must be used to express this dynamic information, and its calculation formula is as follows:

$$d\left(n\right)=\frac{1}{\sqrt{\sum _{-k}^{k}\times {i}^{2}}}\sum _{-k}^{k}\times i\times c(n+i)$$

c(n) is the vector of the field of view of the specific lips, d(n) is the effective vector of the related image, and k is the effective change of the study.

$${V}_{d}={\Delta }{V}_{s}$$

Combine the single image feature of the lips with the dynamic feature, which is the total feature vector:

$$V=({V}_{d},{V}_{s})$$

3.3. Experimental results and analysis of intelligent language recognition

The experiments in this section prove that the appearance of the lips is reasonable and effective. The recognition range is 5 vowels. Because the pronunciation of the voice is very simple, the language of each voice is structured, and each mouth is repeated 10 times, so the geometric shape of the lips is used for recognition separately in statistics. (Experiment 1): Only the pixel symbols of the lips are used for recognition. (Experiment 2): And use the combination of the geometric shape of the lips and the pixel symbol for recognition. (Experiment 3): The identification method uses an active time adjustment algorithm. If there is a difference between speaking language and training protection, DTW can reduce the cumulative distance. It is usually used to solve the problem of misreading phrases due to different speech speeds. This is a simple and effective recognition algorithm. The test results are shown in Table 1.

Table 1

Language recognition experiment under three visual characteristics
	Number of trials	Correct times	Number of errors	Accuracy
Experiment one	50	26	24	52%
Experiment two	50	34	16	68%
Experiment Three	50	48	12	76%

The experimental results can be seen from the visual shape of the lips, which corresponds to the detection accuracy from top to bottom: the geometric shape of the lip area and the shape pattern after connecting the pixel shapes, the pixel shape of the pixel, the lip part taken by DCT, and the lip area Geometric view. Visually, the process of extracting the mixed components is effective. In addition, the accuracy of pixel recognition is higher than that of geometric recognition for two reasons: First, the geometric shape is much higher than the pixel shape. Secondly, the size of the pixel image is better than the geometric shape, it can provide more language information.

4.1. Construction of English Discourse Analysis System

The use of Bayesian division realizes the automatic classification of class announcements, which automatically classify different languages and retrieved feedback questions. Due to the large number of companies, this article made a few modifications to the unconventional Bayesian classification method, and revised it based on the previous classification type, and used the classification results at this time. Figure 5 below shows a new training sample for training, so it needs to be updated regularly to change the classification mode and the concepts behind the classification:

Before classifying the class language (such as simple text steps), please advance the text that has been modified in the class language, including word sorting, word spelling, etc. They all represent the shape of the parasite space, and the image extraction is done in the way of extracting information. Use Bayesian practice text to create sorted categories. The next step is to implement classification, as shown in the figure above: by merging the new classification results into the training site, self-update during the modeling process. And update the existing "ranking table" so that only a few training examples are needed to determine the correct classification type.

4.2. English discourse analysis corpus design

The data used in the Bayesian classifier language library are all excellent teaching videos of the 5 main courses in the 9 classes in the required courses, as the test objects of the analysis. Each course has five types of courses: Chinese, Mathematics, Ethics, Technology and Art. Choose up to 2 teaching cases for each different course, and 40 teaching videos for every trillion research tools.

(1) Select excellent teaching videos.

(2) Convert classroom teaching videos into TXT files.

(3) Converted into XML documents.

(4) Java parses the xml file in eclipse, parses the teaching style and class utterance types and writes them into the corresponding training set corpus.

The allocation of teachers and the transfer of speech language assessment will determine the basic situation of the entire class, understand the full speech ability of the teacher in the classroom and the depth of student behavior, so as to realize the choice of the teacher. The classroom teaching method is shown in Table 1 below. For information about the classroom, the distribution of good class teachers’ speeches during class lectures accounted for 69–77% of the total number of class speeches, while student speeches accounted for 23%-31% of the total number of speeches. Research on most teachers and students' lectures shows that if there is a tendency to "deprive students of interest" and is only suitable for comparison when expressing themselves, then the teacher provides students with the basic content of the school content. Compared with regular courses, the proportion of teacher lectures in information courses has dropped by an average of 80%, which indicates that teachers have more opportunities to communicate with the classroom.

Table 2

Classroom Discourse Distribution Table
Types	Legislative	Integral	Radical	Judgmental
Teacher talk volume	0.71	0.77	0.75	0.69
Student speech	0.29	0.23	0.25	0.31

In the S method of teaching process analysis, the dialogue discussion between teachers and students studies the volume of speech and the mode of speech rotation. The model records the spoken language ratio and the spoken language cycle of teachers and students, and conducts a comprehensive analysis of the teaching process. The analysis steps are as follows:

4.3. Evaluation of the effect of English discourse analysis

For the test, please enter a total of 206 language libraries-Chinese "Thunderstorm" third grade, publish the EXCEL file search results, and write a program to compare the results. Using the correct level, memory level and value F1 test is a test to evaluate the fingerprint. Table 3 shows the experiments conducted using four different types of announcements.

Table 3

Classification result table after filtering the subject words of teaching content
Index	Type of discourse
Index	Classroom organization	Teaching content presentation	Interaction type	Evaluation feedback
Precision rate P (precision rate)	84.8%	95.0%	51.0%	74.0%
Recall rate R (recall rate)	52.0%	61.4%	98%	89.5%
F1 test value	64.5%	74.6%	67.5%	81.0%

Class discourse and speech guidance in internal speeches are of high value, while discussion and evaluation are of high value. It can be seen from the table that there is room for improvement, especially the low memory of classroom speeches and low accuracy of announcement integration. The reasons for these problems may be: (1) The discussion in the subject teaching is composed of different types, and the subject described is not accurate enough; (2) The degree of absolute freedom is between the two, but only when freedom is increased It may be resolved.

4.4. Examples of English discourse analysis based on Fisher's three-dimensional analysis model

Halliday's working grammar system considers issues through the description or definition of specific events or applicable grammatical characters. Word selection is the most effective tool to implement a classification system. Different vocabulary choices reflect different behaviors and attitudes. An analysis of Trump’s vocabulary of Chinese tweets may provide a reasonable explanation for his views and prejudices about China. Use the Antconc3.2.4 program to search for word combinations related to Chinese, and the words found belong to different categories. The distribution and usage of Chinese-related messages are shown in Table 4.

Table 4 shows that Trump's China-related tweets used a lot of negative negative words, such as "vicious", "bad", "worse", "steal", "weaken", "worst", and "ripoff". The words are simple and emotional. The description of China’s economy, market, employment, and agriculture uses words such as "slowest", "recession", "drop", "collapse", "lost" and "attack", shaping an economic downturn, high unemployment, and damage to agriculture. The hostile country image, and other descriptions lack objective evidence. What is surprising is that China has suffered economic losses for a long time due to trade disputes and has benefited from Sino-US trade for a long time. This has caused psychological problems to the American people and has been regarded as being against China. Threats to development. Tweets involving intellectual property actually use the word "steal", believing that China is "stealing," and a large number of descriptions are seriously inconsistent with the facts. Tweets related to the US economy and market use a large number of positive words such as "best", "stronger", "win", "speedup" and "great".

Table 4

Trump's use of China-related tweets classification
Vocabulary trial range	Chinese vocabulary	American vocabulary
Country	hard,notaccep table; vicious; bad, worse;lose;badly;weaken;steal;target;develop nation ;npofT	First great potential; nice; srsuoca; we;exem richer; victory;good;growth;freed an
Economic	I great economic power; muchger; slowed;slowdown ;untarre cession;hun; marupulatc	best; more rapidly; speedup; much larger; free beat
Market	drop;collapse	Stroden go up dramaically; surge
Employment	lost;flee	Loucstune mplovment
Agriculture	pirchase;attack;buy	Fines; dcancs; grea

Example 1 China’s economic growth in the second quarter was the slowest in 27 years.

Example 2 Our economy is much larger than China, and it has continued to grow since the 2020 election.

Examples 1 and 2 show through the relationship process that China's economic growth rate in the second quarter was the slowest in 27 years, and the economic strength of the United States was even higher than that of China. This type of dialogue shows that Trump has defined purposeful behavior. It combines economic themes and communication processes, and describes uncertain information as complex facts in a convincing way, leading to the wrong ideas from the audience.

This research attempts to apply Faircraft's theory to the analysis of brief political posts on Twitter, which have their own creativity and reference points. However, due to keyword mining, the physical collection has not been completely completed. The transition process of sorting changes is purposeful, not completely purposeless, and even still requires a different perspective. In recent years, the combination of natural language technology and intelligent recognition has made unprecedented progress, and basic technology has also achieved certain success. Extensive research based on natural language research technologies (such as text messaging and language processing) Despite these trends, it can be seen that with the development of technology, new research results based on natural language working language technology will continue to change its way of existence. People's lives have many applications that are worth looking forward to.

Compliance with Ethical Standards

Funding

The study was supported by “Youth Research Fund of Humanities and Social Sciences of the Ministry of Education, China” (Grant No.15YJC740104) and “The Major Teaching Research Project of Education Department of Anhui Province, China” (Grant No. 2016jyxm0759).

Conflict of interest

The authors declare that they have no conflict of interests

Ethical approval

This article does not contain any studies with human participants performed by any of the authors.

Data Availability

Data will be made available on request.

T. Foeckerer, M. F. Zaeh, and O. B. Zhang, “A three-dimensional analytical model to predict the thermo-metallurgical effects within the surface layer during grinding and grind-hardening,” International Journal of Heat and Mass Transfer, vol. 56, no. 1–2, pp. 223–237, 2013.
J. Lee, Y. Kim, and S. Jeong, “Three-dimensional analysis of bearing behavior of piled raft on soft clay,” Computers and Geotechnics, vol. 37, no. 1–2, pp. 103–114, 2010.
N. J. Albury, “Mother tongues and languaging in Malaysia: Critical linguistics under critical examination,” Language in Society, vol. 46, no. 4, pp. 567–589, 2017.
S. Hadigheh, and M. S. Khaghaninezhad, “How to apply critical linguistics in English/Persian literary works' criticism,” International Journal of Linguistics, vol. 5, no. 4, p. 90, 2013.
O. Banos, M. Damas, H. Pomares, et al., “Daily living activity recognition based on statistical feature quality group selection,” Expert Systems with Applications, vol. 39, no. 9, pp. 8013–8021, 2012.
W. Cai, X. Wen, Q. Tu, and X. Guo, “Research on image processing of intelligent building environment based on pattern recognition technology,” Journal of Visual Communication and Image Representation, vol. 61, pp. 141–148, 2019.
W. Liu, C. Qin, K. Gao, et al., “Research on medical data feature extraction and intelligent recognition technology based on convolutional neural network,” IEEE Access, vol. 7, pp. 150157–150167, 2019.
R. Hussain, H. Gao, and R. A. Shaikh, “Segmentation of connected characters in text-based CAPTCHAs for intelligent character recognition,” Multimedia Tools and Applications, vol. 76, no. 24, pp. 25547–25561, 2017.
A. Brock, “Critical technocultural discourse analysis,” New Media & Society, vol. 20, no. 3, pp. 1012–1030, 2018.
N. M. Noor, I. Aman, R. Mustaffa, and T. K. Seong, “Teacher's verbal feedback on students’ response: A Malaysian esl classroom discourse analysis,” Procedia-Social and Behavioral Sciences, vol. 7, pp. 398–405, 2010.
D. E. White, N. D. Oelke, and S. Friesen, “Management of a large qualitative data set: Establishing trustworthiness of the data,” International journal of qualitative methods, vol. 11, no. 3, pp. 244–258, 2012.
A. Neundorf, “Democracy in transition: A micro perspective on system change in post-socialist societies,” The Journal of Politics, vol. 72, no. 4, pp. 1096–1108, 2010.
Q. T. N. Nguyen, “The Vietnamese Values System: A Blend of Oriental, Western and Socialist Values,” International Education Studies, vol. 9, no. 12, pp. 32–40, 2016.
S. Xie, and O. Boyd-Barrett, “External-National TV news networks’ way to America: Is the United States losing the global “information war”?,” International journal of communication, vol. 9, p. 18, 2015.
N. Dethlefs, M. Milders, H. Cuayáhuitl, et al., “A natural language-based presentation of cognitive stimulation to people with dementia in assistive technology: A pilot study,” Informatics for Health and Social Care, vol. 42, no. 4, pp. 349–360, 2017.

Download PDF

Editorial decision: Major Revision
08 Mar, 2023
Reviewers agreed at journal
01 Mar, 2023
Reviewers invited by journal
23 Jan, 2023
Editor assigned by journal
19 Jan, 2023
First submitted to journal
18 Jan, 2023

You are reading this latest preprint version

Algorithms of Intelligent Language Recognition and English Discourse Analysis Based on Fisher's Three-dimensional Model

Status:

Version 1

Abstract

Figures

1. Introduction

2. Related Work

3. Fischer'S Three-dimensional Analysis Model And The Theoretical Basis Of Intelligent Language Recognition

3.1. Principles of Fischer's three-dimensional analysis model

3.2. Lip-based language feature extraction

3.3. Experimental results and analysis of intelligent language recognition

4. Design And Application Analysis Of English Discourse Analysis System

4.1. Construction of English Discourse Analysis System

4.2. English discourse analysis corpus design

4.3. Evaluation of the effect of English discourse analysis

4.4. Examples of English discourse analysis based on Fisher's three-dimensional analysis model

5. Conclusion

Declarations

References

Status:

Version 1