Deep Learning-Based Sentiment Analysis and Video Recommendation System Using Convolutional Neural Networks

Abstract

In recent years, the exponential growth of user-generated content on social media
platforms has created a need for intelligent systems that can analyze textual data and
provide meaningful insights. Sentiment analysis, also known as opinion mining, plays a
crucial role in understanding user emotions, opinions, and behavioral patterns. This
research presents a deep learning-based sentiment analysis and recommendation system
that leverages Convolutional Neural Networks (CNN) to predict sentiment scores and
generate personalized video recommendations.
The proposed system integrates natural language processing (NLP) techniques with deep
learning models to classify user comments into sentiment categories ranging from highly
negative to highly positive. The system uses the VADER (Valence Aware Dictionary and
sEntiment Reasoner) sentiment analyzer to initially compute sentiment polarity scores,
which are further refined using a CNN model trained on a labeled dataset. The dataset
consists of user comments, sentiment ratings, and associated video identifiers. Data
preprocessing techniques such as normalization, missing value handling, and scaling are
applied to improve model performance.The CNN model is designed with multiple
convolutional layers, pooling layers, and dense layers to effectively capture patterns in
sentiment-related features. The trained model predicts sentiment ratings, which are then
used to recommend videos based on similarity in sentiment preferences. The system also
supports batch processing of comments through file uploads and visualizes sentiment
distribution using graphical representations.