Volume 6, Issue 6, June – 2021 International Journal of Innovative Science and Research Technology

ISSN No:-2456-2165

Emotion Detection Using Text

M. Lubna Yasmeen1, E. Sai Santoshi2, G. Varshika3, K. Pragnya4, K. Sahasraksha5

1 Assistant Professor in Sreyas institute of engineering and Technology, JNTUH, India,

2 [Link] in Sreyas insinstitute of engineering and Technology, 17VE1A0576.
3 [Link] in Sreyas insinstitute of engineering and Technology, 17VE1A0580
4 [Link] in Sreyas institute of engineering and Technology, 17VE1A0584.
5 [Link] in Sreyas institute of engineering and Technology, 17VE1A0589.

Abstract:- The Internet and communication technologies be done. Although humans are capable of identifying their
of today are extremely fast and dynamic. In the age of feelings and emotions, this is still a challenge because
mobile phones and computers, the use of different computers are not capable of identifying human feelings in
communication channels, such as cell phones and written form. Now here, the role of human computer
computers, is very common. This project develops an interaction in digitalization is crucial.
Emotion Detection Model that takes sentence-level
emotion into account. In natural language processing, The problem now is that we have so much data
content-based classification problems include concepts available via the internet that it is our job to turn that data
from both machine learning and natural language into digital form and to find feelings. A feeling can be
processing. Our technique employs direct emotional expressed in as few or as many words as necessary. Emotion
keywords in text as a means of identifying emotions. In is Social Physiology: A book with six Ekman emotions—
order to increase the accuracy of the detection, words happiness, anger, fear, happiness, surprise, and fear—is
and phrases containing emotion-affect were also titled "Emotion is Social Physiology." It has additional
considered. We have thought about emotions like feelings that can be classified as secondary and subject-
happiness, sadness, anger, and so on to help us recognise related. At sentence level, we will look for emotions in text.
emotion in text. Human beings actually use these When you're reading, you can track the writer's feelings by
expressions. They are important investigations, as their looking for clues in the text. To respond to “Are you
findings have the potential to better express human happy?” you can use a positive response that conveys
emotions and help facilitate interpersonal positive feelings. The concept could be described as
communication. frustrated, sad, or angry if “the person is not happy.” So,
from the text, the user's emotional state can be deduced.
Keywords:- SVM, Python, Emotion Detection, Testing. There are various applications like when an employee writes
an angry email to a colleague, he will see the point of the
I. INTRODUCTION email and protect his job that way. It's also important for the
marketer to be aware of the emotions that their customers
Our daily life presents us with different situations, and are feeling so that they can provide better relationship
we come to know how we feel about them. Emotion is a management, product service, and product delivery
strong feeling about how humans are situated with regard to strategies for their customers. While psychologists also can
each other or their relationships. Facial expression is used to benefit from being able to guess people's emotions based on
display feelings and Emotion. Love, joy, anger, sadness, the writing they produce which can be used to predict their
fear, and surprise are the primary emotions. A human being mental state, psychologists can also take advantage of
can display a wide range of emotion through various means, knowing other people's emotions from what they write.
including facial expression, speech, actions, and written text.
Most of the work here is about two concepts: written text II. PROBLEM DEFINITION AND EXISTING
and speech. Computers, tablets, and smartphones are now SYSTEM
commonly used to access the internet. Textual data is
generated in large quantities due to this process. Manually Content-based classification in text documents
analysing all the data for a specific purpose is now revolves around concepts from both Natural Language
impossibility. The development of new research directions Processing (NLP) and Machine Learning. This paper
has been enabled by advancements in automatic data discusses textual data and the techniques used in emotion
analysis, such as automatic emotion analysis. Researchers detection, as well as their effect on emotion recognition. We
have paid attention to the use of emotion analysis because of can identify and summarise the issues by posing different
its application in various fields. A simple example of this questions.
concept is that security agencies can monitor emails,
messages, and blogs to identify suspicious activities. Collection of Data:
Emotions can be any strong feelings related to a particular Which features should be extracted from the data? And
situation, a particular mood, or a particular relationship. how do you go about dealing with the constant shifts in the
Through text, feelings, and speech, exchanging emotion can types of words we use in our daily conversations?

IJISRT21JUN921 [Link] 1228

Volume 6, Issue 6, June – 2021 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
Features Choices:
How many different kinds of emotion signs can be
found in a speech? What information can be derived from
contextual data? In order to get a good result, how can these
features be combined? Firms are using emotion detection
technology to improve customer service, conduct interviews,
and better target advertising for their customers. However,
researchers in the field have warned that some software may
not always be trustworthy because of out-dated
psychological theories that are used.

It is the most straightforward to use keyword-based

emotion detection. The goal is to identify emotion-related
patterns, and then match them. Sentiment analysis focuses
Fig 1: SVM (a)
on identifying whether a subjective expression is positive,
negative, or neutral and then classifying documents based on
What Support Vector Machines do is consider a region
their objectivity or subjectivity.
around the line separating two classes, also referred to as a
"bounded search space".
III. PROPOSED SYSTEM

Based on facial expression, there are 6 basic emotions

that are broadly applied to describe humans: anger, disgust,
fear, happiness, sadness, and surprise. These are generally
related to feelings of anxiety, particularly “Surprise,” which
can be linked to either positive or negative emotions.
Interestingly, it has recently been determined that the
number of basic human emotions has been “reduced” to
four: happiness, sadness, fear or surprise, and anger or
disgust. Many people are surprised to learn that we only
have four basic emotional states. To make things simpler,
we will use the widely used 6 emotions instead. Still, it
remains unclear how much emotion we can get across with
text. This is especially interesting since the expression on
your face and the way you say things communicate about Fig 2: SVM (b)
70% of the emotions that you intend to convey in speech.
An inherent characteristic of any kind of recognition task is
that there are three major approaches, one of which is rule-
based, the other being statistical, and the third being a hybrid
that relies on factors such as the amount of data available,
domain expertise, and domain specificity. Lexicon-based
methods, machine learning, or a concept-level approach are
useful approaches for sentiment analysis. Today, we will be
looking into the many ways in which we can do this. We
will use machine learning in particular, using the deep
learning method.

Algorithms: SVM
A common approach to supervised machine learning in
this field is via support vector machines (SVMs), also Fig .3: SVM (c)
known as support vector networks. An SVM is a
discriminative classifier defined by a hyperplane that splits This is the intuition of support vector machines, which
minimises the perpendicular distance between the two
the class into two sections. Supervised learning yields an
datasets in order to get the most accurate linear discriminant
optimal hyperplane that classifies new examples.
model. Finally, let's use our training data to train the
classifier. It is necessary to first import cancer datasets as
csv files so that we have two distinct features to train on.
This next task is to use a Support Vector Machine (SVM) to
assign a class label to these locations. while the math
involved in the likelihood model is fascinating, we will let
someone else describe the math elsewhere. We are going to

IJISRT21JUN921 [Link] 1229

Volume 6, Issue 6, June – 2021 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
treat the scikit-learn algorithm as a black box and simply UML also uses activity diagrams, which are helpful for
implement the above procedure inside it. documenting dynamic aspects of the system. A simple
flowchart that is commonly referred to as an activity
diagram can be defined as a flowchart used to represent the
flow form one activity to another. The operation of the
system can be referred to as this activity. The control flow is
constructed in one operation and passes from one operation
to another. Flow can be sequential, branched, or concurrent.
Flow control is handled by using various elements like
forks, joins, and the likes in an activity diagram. A particular
operation of the system is called activity.

We use NLTK tools to implement this project. The

project is implemented through a step-by-step process,
which starts by cleaning the given text. To reach a
successful system, implementation is critical, and users'
Fig 4: Architecture diagram confidence in the new system is essential. The modification
of the application was implemented so that it could be used
This is the intuition of support vector machines, which in place of an existing application. Conversations of this
minimises the perpendicular distance between the two type are manageable if no major changes are implemented.
datasets in order to get the most accurate linear discriminant Each programme is tested individually during development,
model. Finally, let's use our training data to train the and the program's data is verified to ensure that the linked
classifier. It is necessary to first import cancer datasets as programme works correctly in combination with the
csv files so that we have two distinct features to train on. computer system and the user's environment. The system
This next task is to use a Support Vector Machine (SVM) to has been established, and users have found it suitable. And
assign a class label to these locations. while the math thus, the system will be put into place shortly. Included in
involved in the likelihood model is fascinating, we will let the simple operating procedure is a feature that aids the user
someone else describe the math elsewhere. We are going to in quickly understanding how the different functions work.
treat the scikit-learn algorithm as a black box and simply The executable version of the application will be developed
implement the above procedure inside it. and loaded on the central machine shared by all users and a
network connection will be established. Finally, the system
must be documented from beginning to end so that all
components and procedures can be found.

Technologies Used
Python is a general-purpose, interpreted, interactive,
and object-oriented programming language. Python was
created to be readable for human beings. English keywords
are frequently used whereas punctuation is used in other
languages, and English does not have the same number of
syntactic constructions as other languages.

Testing is the process of determining whether a system

or its components meet a requirement by applying them and
evaluating the results. In contrast to the actual requirements,
testing is used to identify any gaps, errors, or missing
requirements.

Table 1: Test Case

Fig 5: Sequence Diagram

IJISRT21JUN921 [Link] 1230

Volume 6, Issue 6, June – 2021 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
IV. RESULTS

Fig 10: Accuracy for decision tree

Fig 6: Home page

Fig 7: Input and Result

Fig 11: Accuracy for random forest

Fig 8: Input

Fig 12: Accuracy for multinominal navie bayes

Fig 9: Result

Fig 13: Accuracy for svm

IJISRT21JUN921 [Link] 1231

Volume 6, Issue 6, June – 2021 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
vol. 5, issue 2, Jun. 2006, pp. 165-183,
doi:10.1145/1165255.1165259.
[8]. Z. Teng, F. Ren, and S. Kuroiwa, “Recognition of
Emotion with SVMs,” in Lecture Notes of Artificial
Intelligence 4114, D.-S. Huang, K. Li, and G. W.
Irwin, Eds. Springer, Berlin Heidelberg, 2006, pp.
701-710, doi: 10.1007/11816171_87. Computer
Science & Information Technology (CS & IT) 377
[9]. C. Yang, K. H.-Y. Lin and H.-H. Chen, “Emotion
classification using web blog corpora,” Proc.
IEEE/WIC/ACM International Conference on Web
Intelligence. IEEE Computer Society, Nov. 2007, pp.
275-278, doi: 10.1109/WI.2007.50.
Fig 14: Graph [10]. C. M. Lee, S. S. Narayanan, and R. Pieraccini,
"Combining Acoustic and Language Information for
V. CONCLUSION Emotion Recognition," Proc. 7th International
Conference on Spoken Language Processing
Our project involved presenting different methods for (ICSLP02), 2002, pp.873-876.
text-based emotion classifications. Additionally, our system [11]. C.-H. Wu, Z.-J. Chuang and Y.-C. Lin, “Emotion
incorporates the surrounding words, and then depicts the Recognition from Text Using Semantic Labels and
results. Additionally, it incorporates user data with historical Separable Mixture Models,” ACM Transactions on
data to understand users' experiences. The future is Asian Language Information Processing (TALIP),
anticipated to include enhancing the textual emotional vol. 5, issue 2, Jun. 2006, pp. 165-183,
deduction modality's efficiency to the existing system. In doi:10.1145/1165255.1165259.
addition, to increase the intensity of other emotions we [12]. C. Elliott, “The affective reasoner: a process model of
haven't used in this paper, Other than feature extraction emotions in a multiagent system,” in Doctoral thesis,
techniques, the most effective classifiers may benefit from Northwestern University, Evanston, IL, May 1992
feature enhancement. [13]. Protégé tool, [Link]/,
[14]. Nicu Sebea, Ira Cohenb, Theo Geversa, and Thomas S.
REFERENCES Huangc “Multimodal Approaches for Emotion
Recognition: A Survey”, USA
[1]. R. Cowie, E. Douglas-Cowie, N. Tsapatsoulis, G.
Votsis, S. Kollias, “Emotion recognition in human-
computer interaction,” in IEEE Signal Processing
Magazine, vol. 18(1), Jan. 2001, pp. 32-80, doi:
10.1109/79.911197
[2]. Parrott, W.G, “Emotions in Social Psychology,” in
Psychology Press, Philadelphia 2001
[3]. C. Maaoui, A. Pruski, and F. Abdat, “Emotion
recognition for human machine communication”, Proc.
IEEE/RSJ International Conference on Intelligent
Robots and Systems (IROS 08), IEEE Computer
Society, Sep. 2008, pp. 1210-1215, doi:
10.1109/IROS.2008.4650870
[4]. Chun-Chieh Liu, Ting-Hao Yang, Chang-Tai Hsieh,
Von-Wun Soo, ”Towards Text-based Emotion
Detection: A Survey and Possible Improvements ”,in
International Conference on Information Management
and Engineering,2009.
[5]. N. Fragopanagos, J.G. Taylor, “Emotion recognition in
human–computer interaction”, Department of
Mathematics, King’s College, Strand, London WC2
R2LS, UK Neural Networks 18 (2005) 389–405 march
2005.
[6]. C. Elliott, “The affective reasoner: a process model of
emotions in a multiagent system,” in Doctoral thesis,
Northwestern University, Evanston, IL, May 1992.
[7]. C.-H. Wu, Z.-J. Chuang, and Y.-C. Lin, “Emotion
Recognition from Text Using Semantic Labels and
Separable Mixture Models,” ACM Transactions on
Asian Language Information Processing (TALIP),

IJISRT21JUN921 [Link] 1232