Analysis of Social Media Trends Using Integrated Bi-LSTM and CNN

Analysis of Social Media Trends Using Integrated Bi-LSTM and CNN

In the digital age, social media platforms generate vast amounts of data that present both challenges and
opportunities for trend analysis. Traditional approaches, such as Bag-of-Words, N-grams, sentiment
lexicons, and rule-based systems, often struggle with ambiguity, sarcasm, domain specificity, and
language variations. To overcome these limitations, this study proposes an integrated deep learning
approach combining Bidirectional Long Short-Term Memory (Bi-LSTM) and Convolutional Neural
Networks (CNN). Bi-LSTM captures contextual understanding and sequential dependencies, while CNN
extracts key patterns, enhancing the robustness and accuracy of social media trend analysis.
Using a Twitter sentiment dataset, our methodology includes preprocessing steps such as noise removal,
tokenization, and vectorization to optimize model performance. The proposed model achieved a validation
accuracy of 99.40% and a recall value of 90%, demonstrating its effectiveness in analysing and predicting
social media trends. This research provides a data-driven framework that can assist businesses,
policymakers, and researchers in understanding digital communication patterns and forecasting emerging
trends