FCM vectorization for Twitter sentimental analysis using multi stacked BiLSTM.

Area of Science:

• Computational Linguistics
• Social Media Analytics
• Artificial Intelligence

Background:

• Twitter data offers insights into societal trends, consumer opinions, and brand reputation.
• Traditional sentiment analysis models struggle with the nuances of short, noisy Twitter text.
• Frequency-based vectorization methods fail to capture semantic relationships and contextual dependencies.

Purpose of the Study:

• To develop a robust sentiment analysis model for Twitter data.
• To address the limitations of existing frequency-based vectorization and word embedding techniques.
• To accurately classify tweets into positive, negative, or neutral sentiments.

Main Methods:

• Phase 1: Text pre-processing, vectorization using Frequency Co-occurrence Matrix, and feature selection via Fisher's score algorithm.
• Phase 2: Implementation of a Multi stacked BiLSTM (Bidirectional Long Short-Term Memory) network for classification.
• Utilizing advanced natural language processing and machine learning algorithms.

Main Results:

• The proposed model achieved a high accuracy rate of 98%.
• The mean squared error rate was significantly low at 0.01%.
• Demonstrated superior performance compared to previous sentiment analysis approaches on Twitter data.

Conclusions:

• The Frequency Co-occurrence Matrix combined with Multi stacked BiLSTM offers a powerful solution for Twitter sentiment analysis.
• The model effectively captures semantic nuances in short, noisy text data.
• This approach provides reliable insights for businesses, researchers, and social scientists.