A deep learning-based model using hybrid feature extraction approach for consumer sentiment analysis.

Area of Science:

• Natural Language Processing (NLP)
• Artificial Intelligence
• Data Science

Background:

• Daily exponential growth in textual data from diverse online sources (social media, e-commerce, news).
• Need for businesses to interpret this data for product/brand analysis and strategic decision-making.
• Existing methods may not fully capture nuanced consumer sentiment from reviews.

Purpose of the Study:

• To develop a consumer review summarization model for extracting business insights.
• To leverage Natural Language Processing (NLP) and Long short-term memory (LSTM) for sentiment analysis.
• To create a hybrid approach for effective feature extraction and sentiment classification.

Main Methods:

• Utilized NLP techniques for text pre-processing to remove noise.
• Implemented a hybrid feature extraction method combining review-related and aspect-related features.
• Employed a Long short-term memory (LSTM) deep learning classifier for sentiment classification.

Main Results:

• The proposed hybrid model achieved an average precision of 94.46%.
• The model demonstrated an average recall of 91.63%.
• An average F1-score of 92.81% was attained across three datasets.

Conclusions:

• The developed NLP and LSTM-based model effectively summarizes consumer reviews.
• The hybrid approach provides substantial insights into consumer behavior and choices.
• This research offers a valuable tool for businesses seeking to understand their customers better.