Aspect Based Sentiment Analysis for Bangla Newspaper Headlines
Md. Nasir Hossain Hridoy, Mohammad Mohitul Islam, Ayesha Khatun · 2021
Aspect-based sentiment analysis means recognizing an aspect in a given content and afterwards perform sentiment analysis of the content regarding that aspect. Sentiment analysis in the Bengali language advances and is also viewed as a significant examination interest. Because of shortage of assets like appropriately commented on the data-set, dictionary, for example, tagger and so on aspect-based sentiment analysis barely has been done in the Bengali language. This research paper aims to find the Bangla newspaper headlines sentiment based on the aspect. Everywhere in the world, people would like to read newspaper headlines first and set up their minds with a summary of the news content instead of reading the news article. This impact may be positive, negative, or neutral. There are many approaches to find the true sentiment among them. Here We use aspect-based sentiment analysis. However, using the small training classifiers data-set for this analysis in Multinomial and Bernoulli Naive Bayes, Logistic Regression (LR), SGD Classifier, SVM, Random Forest, MLP Classifier and Voting Classifier. In our experiment Bernoulli Naive Bayes performed better, resulting in the highest F1 score of 70.75.