Multi stage sentiment analysis for product reviews on Twitter using optimized machine learning algorithm

Area of Science:

• Computational Linguistics
• Social Media Analytics
• Marketing Science

Background:

• Consumer feedback is increasingly found on social media platforms like Twitter.
• Analyzing this feedback provides valuable insights for product development and marketing.
• Traditional methods may not fully capture the nuances of social media sentiment.

Purpose of the Study:

• To explore the feasibility of using machine learning for sentiment classification of product-related tweets.
• To compare the effectiveness of conventional and deep learning models for this task.
• To identify the optimal sentiment classification framework for social media product discussions.

Main Methods:

• A multi-stage framework combining Support Vector Machines (SVM), Naive Bayes, Random Forest, and Long Short-Term Memory (LSTM) networks.
• Training and evaluation on a dataset of 5200 English tweets containing product opinions (positive, negative, neutral).
• Optimization and comparative analysis of the performance of different machine learning algorithms.

Main Results:

• Machine learning effectively extracts and analyzes unstructured social media text for sentiment.
• The study determined the most effective sentiment classification methods for product discussions.
• Sentiment analysis of social media data offers significant benefits for businesses.

Conclusions:

• Social media sentiment analysis is a viable and useful business strategy.
• Companies can enhance client approaches and marketing by analyzing consumer attitudes.
• Understanding customer opinions through social media leads to improved products, services, and customer loyalty.