A Discrete Gravitational Search Algorithm for Effective Feature Selection Mechanism for Sentiment Classification

Sorting texts into categories based on their tone or emotional content is a crucial part of text mining, a process known as sentiment classification. Machine learning methods may be used to classify documents in sentiment classification by representing them as feature vectors. Feature vectors are unable to find without first doing the feature selection. This research develops a discrete nature-inspired feature selection mechanism for effective sentiment classification. The problem of feature selection has been addressed with discrete Gravitational Search Algorithm which is a well-known nature-inspired algorithm, by employing discrete operation on solution repairing process. Nine distinct document instances are utilized to figure out how the proposed method works. The operational model of the algorithm is quantified using five separate filter-based measures. The findings show that the models were far better than the state-of-the-art methods.