Restaurant based recommender system based on sentimental analysis

Abstract

Sentiment analysis is now widely used in the creation of recommender systems in a variety of areas, as well as the restaurant and food service industries. Most restaurant recommender systems based on sentiment analysis, on the other hand, depend only on static data such food quality, cost, and service quality. Personalised suggestions are generated by analysing user views and extracting their dietary preferences, which is a study need in the literature. A context-aware recommender system is suggested in this study, which extracts people's food preferences from their comments and proposes restaurants based on those choices. Finally, adjacent open eateries are suggested depending on how well they match the user's choices. The Kaggle platform was utilised for review, and feedback from various users was gathered. The system's In three scenarios, accuracy, recall, and f-measure are evaluated for the Wu-palmer module: top1, top3, and top5. The same dataset was used to measure the cosine similarity and the Jaccard similarity, and their accuracy and precision were calculated, and these three modules were compared. The findings show that the suggested systems can deliver suggestions with 92.8 percent (Wu-Palmer) accuracy, 87 percent for cosine similarity, and 82 percent for Jaccard similarity. Giving users a high level of precision, Wu-Palmer has the greatest accuracy compared to the other two Cosine and Jaccard Similarity algorithms. Furthermore, in these areas the system exceeds past studies.