Sentimental analysis from imbalanced code-mixed data using machine learning approaches

Area of Science:

• Natural Language Processing
• Machine Learning
• Computational Linguistics

Background:

• Sentiment analysis is vital for knowledge discovery but faces challenges with imbalanced datasets.
• Code-mixed data, common in multilingual contexts, exacerbates class imbalance issues in sentiment analysis.
• Existing research has primarily focused on sentiment analysis in monolingual data, neglecting code-mixed environments.

Purpose of the Study:

• To address the class imbalance problem in sentiment analysis specifically for code-mixed data.
• To propose a novel solution combining sampling techniques and Levenshtein distance metrics for improved sentiment classification.
• To evaluate the performance of various machine learning classifiers on this challenging dataset.

Main Methods:

• A combination of sampling techniques to handle class imbalance.
• Application of Levenshtein distance metrics for enhanced feature representation.
• Implementation and comparison of machine learning models: Random Forest, Logistic Regression, XGBoost, Support Vector Machine, and Naïve Bayes.

Main Results:

• The proposed method effectively handles class imbalance in code-mixed sentiment analysis.
• Performance comparison using F1-Score reveals the efficacy of the implemented machine learning approaches.
• The study provides insights into the best-performing models for this specific task.

Conclusions:

• The developed approach offers a robust solution for sentiment analysis in class-imbalanced, code-mixed datasets.
• This work contributes to advancing sentiment analysis techniques for real-world multilingual communication.
• Further research can explore more sophisticated methods for handling code-mixing and imbalance.