Sentiment analysis using long short term memory and amended dwarf mongoose optimization algorithm.

Area of Science:

• Natural Language Processing
• Machine Learning
• Artificial Intelligence

Background:

• Sentiment analysis is crucial but challenging due to text complexities like abbreviations and errors.
• Traditional methods struggle with nuanced sentiment detection in diverse text data.

Purpose of the Study:

• To enhance sentiment analysis performance using advanced machine learning techniques.
• To investigate the effectiveness of different word embedding models combined with an optimized LSTM architecture.

Main Methods:

• Utilized GloVe and Word2Vec for text vectorization.
• Employed an optimized Long Short-Term Memory (LSTM) model with the amended dwarf mongoose optimization (ADMO) algorithm for hyperparameter tuning.
• Evaluated the model on the IMDB and SST-2 datasets.

Main Results:

• The LSTM-ADMO model achieved high accuracy, reaching 97.84% on SST-2 with Word2Vec.
• Performance differences between Word2Vec and GloVe were minimal.
• The proposed model demonstrated superior performance compared to existing methods.

Conclusions:

• The optimized LSTM-ADMO model is highly effective for sentiment analysis, even with complex text data.
• Both Word2Vec and GloVe are suitable word embedding techniques for this approach.
• The study highlights the potential of metaheuristic optimization for deep learning models in NLP.