Food review sentiment analysis by FastText and machine learning approach

Abstract

The purpose of this project is to investigate the use of sentiment analysis techniques to analyze food evaluations that are written in the Bengali language and posted on online delivery platforms in Bangladesh. On the basis of a dataset consisting of 5000 reviews, a number of different approaches, including fastText, boosting, and machine learning, were utilised. An accuracy rate of 95.92% was attained by the Random Forest model, which was the model that achieved the highest level of accuracy. The study improved the level of happiness experienced by customers, bolstered the performance of local firms, and made a significant contribution to the existing body of literature on sentiment analysis. This effort placed a significant emphasis on the processing of data in a manner that was ethical, transparent, and accountable. The conservation of resources, the encouragement of environmental consciousness, and the maintenance of research viability over the long term are all components of a sustainability strategy. The subsequent stages include doing indepth research on advanced algorithms, putting in place real-time monitoring systems, integrating many modes of analysis, building artificial intelligence that adheres to ethical standards, offering education, and carrying out worldwide comparison studies.