Food Items Prediction Using Sentimental Analysis

In today’s scenario, feedback, opinions and reviews provided on the web, social media etc., has proven to be an indirect promotion for the products, companies etc., which acts as an x-factor for increasing the sales of those items or gaining the popularity of the particular domains. This research work has proposed an autonomous feedback system to the food items available at the Chandigarh University using sentimental analysis, that will help the customer to judge the items as a good or bad based on the opinion score of the items. Here, a database has been created by using an initial survey on the previous customers for the food courts located at the University. For comparative analysis, five machine learning algorithms viz., Naıve Bayes, Naıve Bayes using TF-IDF techniques, Multi-nomialNB, XG Boost, and Cat Boost were selected. Naıve Bayes was selected as a baseline for the comparative sentimental analysis of the food items using the CUFC corpus. Experimental analysis shows the highest accuracy (80.40%) using the Cat-Boost algorithm, thus providing the improvement of 18.80% over the baseline feature. Further, this study will also help the customers to select the food items based on the opinions scores and will be also provided to the shopkeepers for improving the items as per feedback received. Thus, imparting and integrating a latest technology for food-oriented applications.