AI/ML Model for Sentiment Analysis
Derive sentiments from the text that has emojis and characters
- Developed NLP and Ensemble Machine Learning-based Sentiment Analysis model for the Arabic language.
- Successfully developed ensemble-based machine learning model for sentiment analysis. Six different sentiments, very positive, positive, very-negative, negative, neutral, and mixed kinds of sentiment predicated from social network posts represented in Arabic.
- This entire solution is based on online social network post data.
- After collecting the social network post data for text-wise sentiment analysis we did all the pre-processing steps like filtration, tokenization, stop-word removal, stemming, spell checking.
- We used Stanford core NLP for POS-tagging and TF-IDF vectorization in feature extraction. Then we developed the model using Ensemble Machine Learning Approached.
- For emoji-based sentiment analysis we classified all emojis into six different sentiments, very-positive, positive, very-negative, negative, neutral, and mixed, and saved in a dictionary. Then we extracted all emojis from text and found the Unicode and matched it with the dictionary to find the final sentiment.
- After finding the sentiment from each module we applied the ensemble machine learning approach, found the weighted average, and based on the weighted average we found the final sentiment