Department of Computer Science and Engineering
A Deep Learning-Based Multi-Modal System for Emotion and
Sentiment Analysis
CS3811 Project Work
312822104068 MOHAMED SAMEER R
312822104094 SHARMILA S
312822104123 YAMUNA S
312822104124 YOGESH V
Supervisor
Mrs SINDHU G
Assistant Professor/CSE
Department of Aerospace Engineering
Department of Computer Science Engineering Agni College of Technology(ACT)
PROBLEM STATEMENT
In today’s digital era, social media platforms like Instagram, facebook, twitter have become primary channels
for users to express opinions, share experiences, and react to events. The vast amount of unstructured textual
data makes it challenging to understand public sentiment accurately. Misinterpretation of opinions can affect
business strategies, political campaigns, and social studies. Therefore, there is a need for an automated system
that can analyze user posts, comments, and hashtags to identify sentiments positive, negative, or neutral. This
project focuses on building a framework to extract and classify emotions from social media posts to understand
public perception effectively.
Department of Computer Science Engineering Agni College of Technology(ACT)
OBJECTIVES
• To collect social media posts and comments from specific users or hashtags using social media API.
• To preprocess textual data for noise removal and standardization.
• To analyze sentiments of user posts and categorize them as positive, negative, or neutral.
• To study the correlation between user behavior and public reactions to trending topics or events.
• To provide accurate sentiment insights for better understanding of social trends and public opinion.
Department of Computer Science Engineering Agni College of Technology(ACT)
ABSTRACT
In the modern era, social media has become a vital platform for sharing opinions, thoughts, and emotions. The
emotional state of individuals significantly influences their daily life, making sentiment analysis an essential
tool for understanding public opinion. The system collects social media posts, comments, and hashtags using the
social media API. The collected data is preprocessed to remove noise such as special characters, emojis, links,
and stop words. After preprocessing, sentiment analysis is performed using Natural Language Processing (NLP)
techniques and ML - XGBoost. The TextBlob library is used to calculate sentiment polarity and classify the text
into positive, negative, or neutral categories. The analyzed results help in understanding public opinion and
emotional trends related to specific users, hashtags, or events. The system also studies sentiment patterns to
identify how people react to trending topics. The final results are presented using simple visualizations for better
interpretation. This project provides an easy and effective method to analyze sentiments on social media and
helps in understanding public perception. It can be useful for businesses, researchers, and organizations to make
better decisions based on social media opinions.
Department of Computer Science Engineering Agni College of Technology(ACT)
METHODOLOGY
[Link] Collection:
Use PHP and social media API to fetch social media posts, comments, and hashtags from target
accounts or trending topics.
[Link] Preprocessing:
Clean text data by removing emojis, special characters, links, and stop words for accurate analysis.
[Link] Analysis:
Apply TextBlob library to calculate polarity and classify sentiments as positive, negative, or neutral.
Department of Computer Science Engineering Agni College of Technology(ACT)
METHODOLOGY
[Link] of Trends:
Evaluate the frequency of sentiments and correlate with events, hashtags, or user activities to identify
public opinion patterns.
[Link] and Reporting:
Present results using charts or tables to summarize sentiment distribution and insights for decision-
making.
Department of Computer Science Engineering Agni College of Technology(ACT)
METHODOLOGY
Credits: From Research paper “Aesthetic Experience and Popularity Ratings for Controversial and Non-Controversial Artworks
Using Machine Learning Ranking” by Sofia Vlachou and Michail Panagopoulos
Department of Computer Science Engineering Agni College of Technology(ACT)
REFERENCES
1. P. Durga & D. Godavarthi, “Deep-Sentiment: An Effective Deep Sentiment Analysis Using a Decision-Based
Recurrent Neural Network” (D-RNN), IEEE Access , [Link]
2. M. Alfreihat, O. Almousa, Y. Tashtoush, “Emo-SL Framework: Emoji Sentiment Lexicon Using Text-Based Features
and Machine Learning for Sentiment Analysis”, IEEE Access, 2024. [Link]
3. L. Ciechanowski, “Assessing Accuracy: A Study of Lexicon and Rule-Based Packages in R and Python for Sentiment
Analysis”, IEEE Access, 2024.[Link]
4. Dhariwal, N., Akunuri, S. C., Shivama, & K. Sharmila Banu (2023). “Audio and Text Sentiment Analysis of Radio
Broadcasts”. IEEE Access, 11, 126900–126916.
IEEE Xplore Link: [Link]
DOI: [Link]
5. Hua Zhao, Manyu Yang, Xueyang Bai & Han Liu (2024). A Survey on Multimodal Aspect-Based Sentiment Analysis.
IEEE Access, 12, 12039–12052.
IEEE Xplore Link: [Link]
Department of Computer Science Engineering Agni College of Technology(ACT)
Thank You
Department of Computer Science Engineering Agni College of Technology(ACT)
