Onlinecustomer feedbacks are taken as an integral part for making a decision before purchasing any product. In this paper the customer's textual feedback has been taken into account for rating the restaurants available in Kolkata. To get the accuracy of a recommender system we have to first understand the users' preferences and then mine those data to valuable information so that it can benefit our society. The users' sentimental approach is measured first and then we have to calculate each user's sentiment on the food item. With that we can also consider an interpersonal sentimental influence. Then we have to consider the reputation of the restaurants from the sentimental distributions of a group of user's set. The performance evaluation of our experimental results is based on the concept of collaborative filtering and it is made on the real life dataset collected from the 'Zomato' rating website. This analysis can well denote the user's preferences which can upgrade our recommendation system. © 2017 IEEE.