Hindawi
Computational Intelligence and Neuroscience
Volume 2022, Article ID 3490860, 9 pages
https://doi.org/10.1155/2022/3490860
Research Article
Homogeneous Decision Community Extraction Based on
End-User Mental Behavior on Social Media
Suneet Gupta ,1 Sumit Kumar ,2 Sunil L. Bangare ,3 Shibili Nuhmani ,4
Arnold C. Alguno ,5 and Issah Abubakari Samori 6
1Department of CSE, School of Engineering and Technology, Mody University, Lakshmangarh, Rajasthan, India
2Indian Institute of Management, Kozhikode, India
3Department of Information Technology, Sinhgad Academy of Engineering, Savitribai Phule Pune University, Pune, India
4Department of Physical *erapy, Imam Abdulrahman Bin Faisal University, Dammam, Saudi Arabia
5Department of Physics, Mindanao State University, Iligan Institute of Technology, Tibanga Highway, Iligan, Philippines
6School of Engineering Sciences, University of Ghana, Accra, Ghana
Correspondence should be addressed to Issah Abubakari Samori; iasamori@st.ug.edu.gh
Received 30 January 2022; Revised 9 February 2022; Accepted 21 February 2022; Published 8 March 2022
Academic Editor: Deepika Koundal
Copyright © 2022 Suneet Gupta et al.*is is an open access article distributed under the Creative Commons Attribution License,
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Aiming at the inadequacy of the group decision-making method with the current attribute value as interval language information,
an interval binary semantic decision-making method is proposed, which considers the decision maker’s psychological behavior.
*e scope of this research is that this paper is based on localized amplification method.*e localized amplification method used in
this research may amplify physiological movement after removing unwanted noise, allowing the movement trend to be seen with
the naked eye, improving the CNN network’s mental identification accuracy.*ese two algorithms analyze the input picture from
various perspectives, allowing the CNN network to extract more information and enhance identification accuracy. A new distance
formula with interval binary semantics closer to decision-makers thinking habits is defined; time degree is introduced. An
optimization model is established to solve the time series weights by considering the comprehensive consistency of expert
evaluation. Based on prospect theory, a prospect deviation value is constructed and minimized weight optimization model, using
the interactive multiple attribute decision community making (TODIM) method based on the new distance measure to calculate
the total overall dominance of the schemes to rank the schemes. Taking the selection and evaluation of supply chain collaboration
partners as an example, the effectiveness and rationality of the proposed method are verified.
1. Introduction not continuously pursue the maximum utility in behavior
but show reference dependence and loss aversion.*erefore,
With the increasing complexity of decision-making prob- it is necessary to consider the psychological behavior of
lems in various fields, individual decision-making has long decision community makers in the decision-making process.
been unable to meet the requirements of scientific decision- In addition, in the entire group decision-making process,
making, and group decision-making has attracted more and due to the limitations of gender and decision makers’
more attention and attention from experts and scholars [1, cognition, decision-makers are more inclined to give eval-
2]. For multiattribute group decision-making, most existing uations in natural language. [5] proposed a binary semantic
research is based on the classical expected utility theory, information evaluation model, which attracted the attention
which assumes that the decision-maker is entirely rational, of scholars at home and abroad. At present, the research
but this is not in line with reality. Prospect theory [3, 4] based on binary semantics mainly focuses on two aspects: on
believes that decision-makers have systematic perception the one hand, the research on binary semantics set counters
bias in the decision-making process. Decision-makers do [6–10]. However, in the actual decision-making process, due
2 Computational Intelligence and Neuroscience
to the ambiguity of decision-making information and the evaluation is used as an example to check the suggested
limitations of decision-makers cognition, decision-makers method’s efficacy and reasonableness. *e TODIM tech-
are often more willing to give evaluation information in the nique completely analyses managements’ risk aversion
form of interval language to reduce decision-making pres- perspective and may represent key stakeholders’ risk pref-
sure. In response to such problems, [11] showed the defi- erence by altering the dimensions, which is more in ac-
nition of interval binary semantics and several set counters cordance with the actual decision-making needs. Any two
and used the interval binary semantic possibility formula to possibilities, however, must be evaluated using the TODIM
sort the solutions; reference [12], based on the maximum approach, which has a significant computational cost.
dispersion [13], defined a new interval binary semantic Psychological behavior is one of the main ways people
Bonferroni average operator and its corresponding express their emotions. Research on various expression
weighting method; reference [14] combined interval binary recognitions has made significant progress [1–6]. In recent
semantics and VIKOR method, and proposed an out- years, the recognition of spontaneous expressions has be-
sourcing supplier selection method; reference [15] proposed come a new research hotspot [7, 8], and psychological ex-
a subway door failure risk assessment method based on pressions are often generated when people want to suppress
interval binary semantics and failure mode. *e method their feelings, which can neither be faked nor suppressed
proposed in the literature [11–15] has the following [7–9]. *e complete expression usually lasts 0.5∼4 s [10],
shortcomings: first, the evaluation information is uniformly which is relatively easy to be recognized by people. However,
distributed in the interval by default, and the distribution of psychology believes that when a person tries to hide his
information in the gap that is more in line with the psy- genuine emotions, occasionally, there are emotions that leak
chology of decision-makers is not considered; second, most out. Psychological behaviors were first discovered in 1966
of the existing research in the single static stage, [16] it is not [20]. *ree years later, [21] used the term psychological
suitable for the situation that requires dynamic multistage behaviors when analyzing a video interview of a patient who
analysis; thirdly, the above methods all assume that the attempted suicide. Psychological behaviors usually change
decision-maker is entirely rational, [17] but in the actual uncontrollably between 1/25 and 1/2 s [22], and the fre-
decision-making, the decision-makers psychological be- quency of occurrence is low, and untrained individuals do
havior is often bounded rationality. *e impact of social not have high recognition ability [23]. *erefore, the results
media on mental health is stress, anxiety, and sadness which reported by different researchers also vary considerably [11,
have all been related to the social network use. Individuals 24]. After this, Ekman and Friesen proposed the Brief Affect
who often use digital platforms, according to latest studies Recognition Test (BART) in 1979 [12]. In subsequent ex-
cited by *e Young Mind Centre and *e National Centre periments, they found that the subjects’ psychological-ex-
for Health Research, are more sad and unhappy with life pression recognition ability was positively correlated with lie
than others who stay longer on nonscreen-related activities. recognition ability [13]. Afterwards, the Japanese and
To sum up, based on existing research, this paper proposes Caucasian Brief Affect Recognition Test (JACBART) [14, 15]
an interval binary semantic dynamic group decision-making was conducted, verifying that the subjects’ psychological-
that considers the decision maker’s psychological behavior expression recognition ability was positively correlated with
for the multiattribute group decision-making problem in lie recognition ability [20], it can be proved that psycho-
which the attribute value is interval language, and the at- logical-expressions can effectively help people identify lies.
tribute weight and time series weight are entirely unknown Face psychological expression recognition involves image
method. Considering the thinking habits of human beings, it processing and analysis, computer vision, artificial intelli-
is believed that the density of evaluation information in the gence, psychology, biology, and other directions.
interval is more similar to the normal distribution, and based
on this, a new interval binary semantic distance formula is 2. Literature Survey
proposed. Solve the time series weights based on the time
degree and entropy; [18, 19] by determining the positive In 2002, Authors [24] developed a psychological-expression
ideal scheme under each time series, establish a prospect recognition tool, METT (Psychological behavior Training
theoretical profit and loss matrix, and build a linear pro- Tool). Studies have shown that METT tools can improve an
gramming model to minimize the sum of the squares of the individual’s ability to recognize psychological expressions by
foreground deviation values, and determine the attribute 30% to 40% on average. In addition, an Action Coding
weights under each time series; then, Construct an ITL- System (FACS) [21] is also designed, according to the an-
TODIM method based on interval binary semantic distance atomical characteristics of the face, it is divided into several
measure to calculate the complete overall dominance of each independent and interconnected motion units (Action Unit,
candidate scheme, and determine the pros and cons of the AU),*emotion characteristics of these motor units and the
system according to the total general authority; the evalu- main areas they control and the expressions associated with
ation is used as an example to verify the effectiveness and them are analyzed, and a lot of photo descriptions are given.
rationality of the proposed method. *e ITL-TODIM ap- Although human emotions are complex and diverse, they
proach which is based on intermediate binary semantic can still be divided into 6 basic emotional categories [22, 23].
distance measure to compute the total overall dominance of *erefore, researchers combine different motor units to
each candidate scheme and identify the system’s advantages form FACS codes to correspond to different expressions,
and disadvantages in terms of total general authority. *e mainly divided into happy, angry, fearful, sad, surprised, and
Computational Intelligence and Neuroscience 3
others. When people express their inner state and psycho- the spatial and temporal information of the video, LBP-TOP
logical needs, they will produce many psychological be- extends LBP. Compared to LBP, this method finds three types
haviors [24]. Still, because psychological expressions exist (XY, XT, YT) instead of one plane (spatial XY). Given a video
for too short a time and are not easily detected by the human sequence, it can be viewed as a stack of XY, XT, and YTplanes
eye, computers can be used to solve this problem. along the temporal T-axis, spatial Y-axis, and spatial X-axis,
respectively. *e three histograms come from three planes,
3. Psychological-Expression respectively, and are concatenated into one histogram as a
Recognition Technology dynamic video texture descriptor, as shown in Figure 2. LBP-
TOP extends the application of the LBP algorithm to a higher
Psychological expression is the tiny movement change of the dimension and can identify textures in time series. *e in-
human face, including texture change. *e movements of formation between the frames before and after is correlated
psychological expressions are too small and of short dura- through this method, making the LBP algorithm more widely
tion to be easily captured by the human eye. *e physical used. Local Binary Patterns (LBP) are frequently employed in
character of all substances is dictated by their real physical texture analysis, texture identification, and other domains and
composition, which is referred to as texture. Textures can have major benefits such as grayscale and rotation invariance
elicit behavioral processes in people. *is mental reaction degradation. *e texture knowledge of the pixel’s surrounding
enables us to feel something without really touching it. *e area is reflected in the central pixel’s LBP value. Any radius and
blood circulation in our faces alters as we experience various number of neighborhood pixels may be achieved by using a
emotions. *is causes tiny colour shifts that other persons circular neighborhood and subsection linear interpolation data
might see. People can properly determine someone’s sen- at noninteger pixel locations. *e complimentary comparison
timents from these visual shifts up to 75% of the time, measure might be the grey scale variation of the immediate
according to recent research. *erefore, the psychological region.
expressions of the face can be studied through machine *ey generate fairly extensive descriptive statistics,
vision. According to the above characteristics, several which slow down identification speed, particularly on huge
mainstream methods for psychological-expression recog- face databases. (2)*ey overlook the spatial patterns in some
nition are Convolution Neural Networks (CNN) [11], Op- circumstances because they do not examine the influence of
tical Flow (Optical Flow) method, and Local Binary Pattern the central pixel.
(LBP). *e properties of LBP can be found in LBP-TOP, and
vice versa. LBP-TOP is unaffected if the pixel value increases
3.1. Convolution Neural Networks. Convolution Neural or lowers by the same amount.*ismethod can be employed
Networks (CNN) [11] has been widely used in various fields in real life despite the interference to the image created by
such as machine vision and speech recognition since their the natural parallel light environment, when in, the ro-
birth. Usually, CNN is used as feature extraction and depth bustness is very high. However, because it is pixel-based, this
feature extraction for image class input, and the desired method requires more expertise in Computer Science and
output results can be obtained after analyzing the extracted Engineering. *e first use of LBP is to compare the centre
features. Relatively small movements characterize Psycho- pixel to P pixels in the vicinity of radius R. When R is 1, the
logical-expressions. If convolution neural networks are used, centre pixel and the surrounding 8 pixels have a size of 2P.
it is usually necessary to use other auxiliary methods to *e final size is 28× 256. R is no longer a single number, and
change the input of the web, or to change and optimize the the size of the neighborhood grows exponentially. *e LBP
network structure so that the network can extract more operator’s mode type is utilized for dimensionality reduction
valuable features, thereby improving the recognition of in [26]’s proposal to employ a “equivalent pattern” (Uniform
psychological-expressions accuracy. Pattern) as a solution to this problem. *ere are twotransitions from 0 to 1 or 0 to 1, according to Ojala et al., in
natural images. As a result, the “equivalent mode” is defined
3.2. Local Binary Patterns and Improvement Methods. as having at most two transitions from 0 to 1 or 1 to 0 in a
Local Binary Patterns (LBP) [12] can effectively deal with cyclic binary integer. An analogous Pattern class refers to the
illumination changes and are widely used in texture analysis, binary that corresponds to the LBP. It takes 256 to get down
texture recognition, and other fields, with grey and rotation to 58 using this strategy. In other words, the values are
invariance degeneration and other significant advantages. divided into 59 categories, with the 58 uniform patterns
*e LBP value of the central pixel reflects the texture in- constituting one and the remaining values constituting the
formation of the surrounding area of the pixel, as shown in 59th. Consequently, the histogram’s 256-dimensional di-
Figure 1. mension is reduced to 59 dimensions using this method,
*is feature is widely used because of its simplicity and ease rather than the original 2P’s P (P− 1) + 2. *is decreases the
of computation. However, the traditional LBP algorithm cannot influence of high-frequency noise on the eigenvectors by
be applied to video signals for video images, so some im- making them less dimensional. Using LBP-TOP, an image is
provements to the LBP algorithm are needed. Among them, divided into 59× 3 blocks, with each block generating an
[25] proposed a robust dynamic texture descriptor that per- array with a size of 3.59. *is is because LBP-TOP applies
forms local binary patterns from three orthogonal planes (LBP- LBP in the time dimension. *e finished feature’s dimen-
TOP), widely used for psychological behavior. To consider both sions are 4× 4× 59× 3� 2.832. LBP-TOP is a high-
4 Computational Intelligence and Neuroscience
36 63 60 0 0 1
42 56 93 1 1
82 23 54 1 0 1
Figure 1: LBP.
L1 L2 …….. Ln
M1 M2 …… Mn
N1 N2 ……….. Nn
Figure 2: LBP-TOP.
dimensional characteristic that has a significant impact on (b) *ere are three main methods of feature screening:
calculation speed and accuracy. filter, wrapper, and embedded. *e filtering feature
selection method assigns weights to the features of
each dimension, and then sorts the features
4. Mental Behavior for Psychological Change according to the weight; the encapsulating feature
Psychological-expression recognition has made some selection method views subset selection as an opti-
progress at present, but there is still a lot of room for im- mization problem, generates different combinations,
provement in technology. *e following is an analysis and evaluates the combinations, and then compares them
prospect of the problems existing in different methods of with others. By comparing the two approaches, we
psychological expression: can determine which features are most critical for aspecific model to be trained with. *ere are three
(a) Convolution neural network used as a Traditional basic screening procedures, and each is designed for
method is widely used, and the current mainstream a certain scenario. *e statistical performance of all
idea is to use the optical flow method as input and training samples is directly used to evaluate the
combine it with LSTM to achieve better results. relevance of each feature in the filtering feature
However, there is a fatal problem in using convo- selection technique. Although its performance is
lution not high. For convolution neural networks, it excellent, this algorithm has a lot of advantages,
is a small sample problem. *erefore, data en- including the ability to remove a large number of
hancement is required during calculation, or Cross- unnecessary attributes, its universality, and the
domain experiments, while improving robustness. In ability to prescreen features. It is necessary to
the subsequent process, you can try to use transfer combine the feature selection approach with the
learning, which can alleviate the problem of over- subsequent classification algorithm, evaluate each
fitting to a certain extent. In addition, you can also feature’s significance based on the classifier’s accu-
use methods such as adversarial neural networks to racy, and select the ideal feature subset for a certain
generate some samples for training, which can al- classification algorithm. It is similar to the filtering
leviate the problem of small sample data. approach, but the integrated feature screening
Computational Intelligence and Neuroscience 5
method uses machine learning training rather than recognition. *ese two methods process the input image
relying solely on the statistical indications of a fea- from different angles so that the CNN network can better
ture to evaluate its advantages and disadvantages. extract features and improve the accuracy of recognition.
Machine learning is also used in the embedding *ey generate fairly extensive descriptive statistics, which
method, as opposed to the packaging method. With slow down identification speed, particularly on huge face
this approach, all of the features are used for training, databases. (2) *ey overlook the spatial patterns in some
rather than just a subset of them. By utilizing the circumstances because they do not examine the influence of
embedded feature selection method, both the feature the central pixel.
selection and learner training are done in the same At present, most of the psychological-expression feature
optimization process. Each of the three ways has its extraction based on CNN is combined with the optical flow
own pros and weaknesses; therefore they can all be method extraction, which is also one of the advantages of the
utilized together. An ideal subset of features is picked CNN framework. However, if CNN is used for deep
out based on preprocessing and classifier approach learning, it is still unrealistic for the current data set. *e
advantages in order to increase the accuracy of amount of data in the existing data sets cannot meet the
psychological-expression recognition. needs of the deep learning network, and it is easy to cause
overfitting, which leads to the huge size of CNN. *e ad-
5. Psychological-Expression vantage cannot be played.
Recognition Method
5.1. Method for Psychological-Expression Recognition Based 5.2. Psychological-Expression Recognition Method Based on
on CNN. CNN is a commonly used method when pro- Optical Flow Method. FACS requires motion records of
cessing image signals. Still, because the psychological-ex- various positions such as eyebrows and mouth corners as an
pression changes are relatively small, the effect of only using essential tool for recognizing psychological expressions.*is
CNN is not very good, so some literature choose to process method uses ROIS and HOOF features together, and the
the input image. *e traditional psychological-expression obtained results correspond to AU, which can identify
feature extraction needs to consider the spatial image fea- psychological expressions more accurately. Psychological-
tures and the temporal sequence. *e amount of pixels used expression detection is often independent, and a long-video
to generate a digital image is referred to as spatial resolution. expression automatic recognition method consisting of
Higher spatial resolution images include more pixels than macros and psychological expressions (time segmentation).
lower spatial resolution images. Object (matrix) and picture *is method utilizes the stress generated by nonrigid motion
(graph) modes are available for representing spatial objects. on the skin during the expression process. It uses the central
Each spatial object can be specified as a spot, line, or polygon difference method to calculate the solid and dense optical
in object mode. Each spatial object can be specified in picture flow field observed in several areas of each subject’s face
mode as a collection of adjacent cells called regions. Gen- (chin, mouth, cheeks, forehead). Strain level: *is method
erally, feature extraction is performed on all frames of the can successfully detect and distinguish psychological ex-
psychological expression from the beginning to the end, pressions and fast local psychological expressions, which is
while proposng a psychological expression. Stage classifier: of great help to the detection of psychological expressions in
First, the psychological expressions are divided into three complex scenes.
stages: onset, peak, and offset using temporal information,
and then spatial information is used to detect intensity 5.3. Method of Psychological-Expression Recognition.
changes. Compared with the traditional psychological-ex- According to the problems of LBP-TOP, STCLQP adds
pression feature extraction method, this method only uses a information such as amplitude and direction components,
psychological-expression (Apex) frame and an initial (On- extracts more helpful information, and fuses the extracted
set) frame to extract features, increasing the number of valid data into a feature vector, which improves the utilization of
frames reducing the amount of computation and outper- image features. Compared with LBPTOP, the STCLQP
forms other traditional methods. In addition, [30] proposed method extracts more information and thus results in higher
a video magnification method based on eye interference dimensionality. In this paper, the LQP technique is used.
elimination and used convolution neural network to realize Compared with the LBP, whose dimension increases ex-
the task of psychological-expression recognition. First, zoom ponentially with the neighborhood radius R, LQP maps the
in on the data to extract the eye position coordinates. After extracted features to the lookup table. When R increases, the
that, the original eye video will be replaced with the enlarged dimension no longer shows exponential growth. *erefore,
video for image fusion to eliminate eye interference. Finally, this method is more suitable when the R-value needs to be
the convolution neural network model network is designed improved, and because the LQP technology alleviates some
using the idea of VGG16 to realize emotion recognition.*e of the previously extracted more dimensional information.
local amplification technology applied in this paper can LBP-SIP reduces redundancy in LBP-TOP mode, pro-
amplify the psychological-expression movement after re- vides a more compact and lightweight representation, im-
ducing the noise interference so that the movement trend proves accuracy, and reduces computational complexity.
can be seen by the naked eye, which can increase the ac- Unlike LBP-TOP, this method only uses 4 pixels at the top,
curacy of the CNN network for psychological-expression bottom, left, and right of a pixel at time t and 6 pixels at the
6 Computational Intelligence and Neuroscience
same position at time t− 1 and t+ 1. Because the extracted on datasets. *e generalization ability of psychological-ex-
features are few, it is more suitable when there are many pression recognition, in reality, is not high and cannot be
images and a long time, and the operation is faster than LBP- applied. However, the required environment and application
TOP. Its speed is 2.8 times that of LBPTOP. *e feature scenarios can be estimated according to the characteristics of
extraction time of this method is 15.88 s, which is about 2.4 s existing methods. As far as the current development is
faster than that of the LBPTOP method, and the recognition concerned, if the multiple existing networks are integrated
time is 0.208 s, which is 0.3 s more quickly than the LBP- into the detection of psychological-expression recognition,
TOP method. Compared with the LBP-TOP method, LBP- complex scenes can be detected, such as shopping malls,
SIP reduces a considerable amount of time, but this time is streets with more critical locations, prisons, etc. Existing
still too long for subsequent practical applications to achieve psychological-expression recognition methods can be
real-time detection. But it provides an idea to reduce the combined with other networks. It can automatically detect
processing time. In addition, to reduce the dimension of the various features of pedestrians in the crowd to ensure the
features extracted by LBP-TOP and then use SVM for accuracy is improved.
classification. *is method first uses the LBP-TOP operator Based on the assumption of the consistent grey level of the
to extract psychological-expression features. It then pro- optical flow method, the visual flow method cannot detect
poses a feature selection algorithm based on Relief combined scenes or objects with changing brightness and has stricter
with a Locally Linear embedded (LLE) manifold learning requirements on the environment.*e camera cannot rotate too
algorithm, which can effectively reduce the number of fast. *erefore, psychological-expression recognition can be
postextraction features. Feature dimension: Finally, the SVM performed in interrogation rooms and negotiation rooms where
classifier with Radial Basis Function (RBF) kernel is used for the brightness is relatively fixed, and the target does not need to
classification, and good results are obtained. *is method is move. Psychological-expressions are based on the premise that
based on the LBP-TOPmethod for feature vector extraction, the target suppresses their expressions, but when faced with a
so the application environment is consistent with LBP-TOP. specific scene, the mark may have the problem of combining
*e filter method is used to exclude irrelevant features, and psychological-expressions and psychological-expressions. *e
then the wrapper method is used to filter out the features work proposed by Shreve et al. [36] proposed a solution for this
with greater influence, which avoids the dimensional di- type of problem by using the optical flow method, making
saster and reduces the amount of calculation accordingly. psychological-expression recognition more suitable for more
*is method gives a direction for future development. It is a stringent places such as interrogation rooms and optical flow
revelation that for the pixel-by-pixel feature extraction rooms. Compared with other methods, the extracted feature
method of LBP-TOP, the extracted features can be screened dimension of the optical flow method is smaller, and it is the
to improve the recognition accuracy during classification. most viable method for real-time detection. For local binary
For the redundant features of LBP-TOP, the above mode improvement methods, such as pixel-by-pixel analysis
methods adopt different ideas for dimensionality reduction, methods such as LBP-TOP, the overall brightness change of the
but it is far from enough for practical application in real life. environment has little effect on it, so it can be used as a
*e method of pixel-by-pixel feature extraction is easy to complement to the optical flow method, but its obvious dis-
cause dimensional disasters. *e selection of features and advantage is that in the extracted feature dimension, the number
feature extraction methods is particularly important before. is too high and the computational burden is too large for real-
It is only when the processing speed is reduced that there is time detection, so it can be used as a behind-the-scenes tool for
an opportunity to apply such methods to social life. psychological-expression calibration and recognition. With the
current research, after the feature dimension extracted by such
5.3.1. CNN (Convolution Neural Network). CNN is an ex- methods is reduced to real time, it may be necessary to extract
cellent method for dealing with image problems, but the specific parts of the target’s face, and it is necessary to minimize
network structure needs to be adjusted for different issues. the occlusion of the key parts of the target to be detected.
When using CNN, feature extraction is usually performed *erefore, suchmethods are used in it thatmay bemore suitable
first, and then further processing is performed after the for areas where the behavior and clothing of the target are strictly
features are extracted. However, the optical flowmethod and managed, such as detention centers or the military.
the improved local binary mode method use different re- LSTM for video-type inputs has a better effect. Rather
moval methods. For the feature map, CNN can continue to than using CNN alone, it is more appropriate to use CNN as
extract deep features, so the use of CNN and optical flow a framework, and CNN can be combined with various
method and local binary mode improvement method does methods to make the improved network recognition more
not conflict with itself. CNNs are often used in conjunction effective. An LSTM differs from a CNN in that it is often
with a popular form of recurrent neural network called the used to attempt to predict performance, and computational
Long Short-Term Memory (LSTM) module for video-type CNN, on the other hand, is meant to identify “spatial
files. patterns” in information and performs well on pictures andsounds. For example, the TV-L1 method extracts the optical
flow image and superimposes the original image as input
5.3.2. Applicable Scene Analysis of CNN. Optical flow while using the CNN+LSTM method and then sends the
method and improved local binary mode of the existing extracted features to the LSTM network for psychological-
psychological-expression recognition methods are all based expression recognition. After using the ROI method, it
Computational Intelligence and Neuroscience 7
Table 1: Performance comparison of CNN methods.
Method Accuracy F1-score
CNN+LSTM 60.98 65.85
ELRCN-SE 47.15 49.52
ELRCN-TE 52.44 55.63
Table 2: Communities comparison of CNN methods.
Method Modularity’s NMI
CNN+LSTM 56.63 85.63
ELRCN-SE 46.52 75.62
ELRCN-TE 53.23 80.56
Table 3: Performance comparison of improved local binary mode methods.
Method Accuracy F1-score
LBP-TOP 57.16 59.63
STLBP-IP 59.51 62.45
STLBP-IIP 62.75 66.48
Table 4: Performance of community extraction by local binary mode methods.
Method Modularity’s NMI
LBP-TOP 66.63 89.24
STLBP-IP 52.63 80.54
STLBP-IIP 56.62 82.26
70
60
50
40
30
20
10
0
CNN+LSTM ELRCN-SE ELRCN-TE
Accuracy
F1-Score
Figure 3: Performance comparison of CNN methods.
cooperates with the Flow Net 2.0 optical flow method to comparison of CNN methods is shown graphically in
identify the visual flow of a specific area and finally enables Figure 3.
the ROI +Revised HOOF to recognize different expressions Figure 4 shows the community features of the proposed
with FACS. *ese two methods use other optical flow binary method. *e improved local binary mode methods
methods. Still, the main body of the TVL1+LSTMmethod is performance is shown in Figure 5 and the community
CNN, and the visual flow method is only used as the features for local binary mode methods are shown in
standard input with the original image to increase the Figure 6.
recognition accuracy, from Tables 1–4. It can be seen that But the optical flow method is essential for the envi-
CNN, as a framework, is very inclusive and can be used in ronment. *e requirements of the database are more
combination with a variety of methods. *erefore, it is a stringent, and it needs to be modified on this basis before it
good tool for solving image problems. *e performance can be applied to parts outside the database.
8 Computational Intelligence and Neuroscience
90 unaided eye. *e optical flow method is a popular method
80 for psychological-expression identification because it can
70 identify small moving targets, which is ideal for detecting
60 love. However, there are also evident issues. It is necessary to
50 refine and optimize the algorithm in order to ensure that the
40 brightness of the identified target remains consistent, which
30 is sometimes difficult to do in actual applications. Optical
20 flow, on the other hand, is a reliable method for detecting
10 face psychological expressions. Using three planes, LBP-
0 TOP can accurately detect videos. *e update of LBP-TOP
CNN+LSTM ELRCN-SE ELRCN-TE minimizes the number of redundant parameters, improves
the accuracy of recognizing texture features, and can do all
Modularity’s three at once. So, using the upgraded method of LBP for
NMI psychological-expression detection would be a good idea
Figure 4: Community feature of proposed methods. because it improves texture recognition.
Data Availability
68
66 *e data shall be made available upon request.
64
62
60 Conflicts of Interest
58
56 *e authors declare that they have no conflicts of interest.
54
52 F1-Score
References
LBP-TOP Accuracy
STLBP-IP
STLBP-IIP [1] H.-H. Shuai, “A comprehensive study on social network
mental disorders detection via online social media mining,”
Figure 5: Performance comparisons of improved local binary IEEE Transactions on Knowledge and Data Engineering,
mode methods. vol. 30, no. 7, pp. 1212–1225, 2018.
[2] H. Yazdavar, M. S. Mahdavinejad, G. Bajaj, K. *irunarayan,
95 J. Pathak, and A. Sheth, “Mental health analysis via social
90 media data,” in Proceedings of the 2018 IEEE International
85 Conference on Healthcare Informatics (ICHI), pp. 459-460,
80 New York, NY, USA, June 2018.
75 [3] Y. Chai, F. Wu, R. Sun et al., “Predicting future alleviation of
70 mental illness in social media: an empathy-based social network
65 perspective,” in Proceedings of the 2019 IEEE International
60 Conference on Parallel & Distributed Processing with Applica-
55 tions, Big Data & Cloud Computing, Sustainable Computing &
50
LBP-TOP STLBP-IP STLBP-IIP Communications, Social Computing & Networking (ISPA/BDCloud/SocialCom/SustainCom), pp. 1564–1571, Xiamen,
Modularity’s China, December 2019.
NMI [4] J. Bai, Q. Kong, L. Li, L. Wang, and D. Zeng, “Exploring
6: Community Feature of local binary mode methods. cognitive dissonance on social media,” in Proceedings of theFigure 2019 IEEE International Conference on Intelligence and Se-
curity Informatics (ISI), pp. 143–145, Shenzhen, China, July
2019.
6. Conclusion [5] M. E. Aragon, A. P. Lopez-Monroy, L.-C. G. Gonzalez-
Gurrola, and M. Montes, “Detecting mental disorders in
Convolution neural network and its improvement, optical social media through emotional patterns-the case of anorexia
flow method and its improvement, and local binary pattern and depression,” IEEE Transactions on Affective Computing,
and its improvement are discussed in this study. *e rec- 2021.
ognition of psychological-expression information can also [6] M. Hu and X. Wu, “Research on children’s network privacy
be improved through improvements to the convolution protection in mobile social media,” in Proceedings of the 2018
neural network itself. Convolution neural networks can be IEEE 20th International Conference on High PerformanceComputing and Communications; IEEE 16th International
useful for psychological-expression recognition when they Conference on Smart City; IEEE 4th International Conference
have been trained to extract visual features. After decreasing on Data Science and Systems (HPCC/SmartCity/DSS),
random noise, the local amplification method used in this pp. 1135–1138, Exeter, UK, June 2018.
study may magnify the psychological-expression movement [7] U. Iqbal and A. Hussain Mir, “Secure and practical access
such that the development trend can be perceived with the control mechanism for WSN with node privacy,” Journal of
Computational Intelligence and Neuroscience 9
King Saud University-Computer and Information Sciences, [21] A. Zahura and K. A. Mamun, “Intelligent system for pre-
2020. dicting suicidal behaviour from social media and health data,”
[8] S. Tariq, N. Akhtar, H. Afzal et al., “A novel Co-Training- in Proceedings of the 2020 2nd International Conference on
Based approach for the classification of mental illnesses using Advanced Information and Communication Technology
social media posts,” IEEE Access, vol. 7, pp. 166165–166172, (ICAICT), pp. 319–324, Dhaka, Bangladesh, November 2020.
2019. [22] X. Huang, L. Xing, J. R. Brubaker, and M. J. Paul, “Exploring
[9] S. T. Sadasivuni and Y. Zhang, “Finding a depressive twitter timelines of confirmed suicide incidents through social me-
user by analyzing depress and antidepressant tweets,” in dia,” in Proceedings of the 2017 IEEE International Conference
Proceedings of the 2020 IEEE India Council International on Healthcare Informatics (ICHI), pp. 470–477, Park City, UT,
Subsections Conference (INDISCON), pp. 142–145, Visakha- USA, August 2017.
patnam, India, October 2020. [23] A. Suphan, M. Feuls, C. Fieseler, and M. Meckel, “*e sup-
[10] V. Vajre, M. Naylor, U. Kamath, and A. Shehu, “PsychBERT: portive role of social media networks for those out of work,” in
amental health languagemodel for social mediamental health Proceedings of the 2013 46th Hawaii International Conference
behavioral analysis,” in Proceedings of the 2021 IEEE Inter- on System Sciences, pp. 3312–3321, Wailea, HI, USA, January
national Conference on Bioinformatics and Biomedicine 2013.
(BIBM), pp. 1077–1082, Houston, TX, USA, December 2021. [24] K. Katchapakirin, K. Wongpatikaseree, P. Yomaboot, and
[11] K. Jairath, N. Singh, V. Jagota, and M. Shabaz, “Compact Y. Kaewpitakkun, “Facebook social media for depression
ultrawide band metamaterial-inspired split ring resonator detection in the *ai community,” in Proceedings of the 2018
structure loaded band notched antenna,” Mathematical 15th International Joint Conference on Computer Science and
Problems in Engineering, vol. 2021, Article ID 5174455, Software Engineering (JCSSE), July 2018.
12 pages, 2021. [25] M. Gulhane and T. Sajana, “Human behavior prediction and
[12] L. Chen, J. Chen, and C. Xia, “Social network behavior and analysis using machine learning-A review,” Turkish Journal of
public opinion manipulation,” Journal of Information Security Computer and Mathematics Education (Turcomat), vol. 12,
and Applications, vol. 64, Article ID 103060, 2022. no. 5, pp. 870–876, 2021.
[13] J. Bhola, M. Shabaz, G. Dhiman, S. Vimal, P. Subbulakshmi, [26] T. Hu, E. Bigelow, J. Luo, and H. Kautz, “Tales of two cities:
and S. K. Soni, “Performance evaluation of multilayer clus- using social media to understand idiosyncratic lifestyles in
tering network using distributed energy efficient clustering distinctive metropolitan areas,” IEEE Transactions on Big
with enhanced threshold protocol,” in Wireless Personal Data, vol. 3, no. 1, pp. 55–66, 2017.
CommunicationsSpringer Science and Business Media LLC,
Berlin, Germany, 2021.
[14] J. Tsujino, M. Oyama-Higa, M. Tanabiki, and H. Kihara,
“Longitudinal study on mental interactions between difficult
children and their nursery teacher based on DUAL mea-
surements of finger pulse waves,” in Proceedings of the 2018
International Joint Conference on Information, Media and
Engineering (ICIME), pp. 326–330, 2018.
[15] S. Deshmukh, K. *irupathi Rao, and M. Shabaz, “Collabo-
rative learning based straggler prevention in large-scale dis-
tributed computing framework,” Security and
Communication Networks, vol. 2021, Article ID 8340925,
9 pages, 2021.
[16] H. Hong, G. D. Abowd, and R. I. Arriaga, “Towards designing
social question-and-answer systems for behavioral support of
individuals with autism,” in Proceedings of the 2015 9th In-
ternational Conference on Pervasive Computing Technologies
for Healthcare (PervasiveHealth), pp. 17–24, Istanbul, Turkey,
May 2015.
[17] A. Sharma, S. Kaur, N. Memon, A. Jainul Fathima, S. Ray, and
M. W. Bhatt, “Alzheimer’s patients detection using support
vector machine (SVM) with quantitative analysis,” Neuro-
science Informatics, vol. 1, no. 3, Article ID 100012, 2021.
[18] M. P. Priya, M. K. Devi, S. U. Kamali, S. Akshayaa, and
P. R. Amutha, “Personality prediction using digital footprints
to analyze kids activities,” in Proceedings of the 2020 Inter-
national Conference on Power, Energy, Control and Trans-
mission Systems (ICPECTS), pp. 1–3, Chennai, India,
December 2020.
[19] J. Godara, R. Aron, and M. Shabaz, “Sentiment analysis and
sarcasm detection from social network to train health-care
professionals,” World Journal of Engineering, 2021.
[20] M. T. Ghozali, S. Satibi, Z. Ikawati, and L. Lazuardi, “Asthma
self-management app for Indonesian asthmatics: a patient-
centered design,” Computer Methods and Programs in Bio-
medicine, vol. 211, Article ID 106392, 2021.