Analysis on restaurant reviews using domain characteristics of text for eating and drinking

Abstract

Reading reviews takes a lot of time to figure out the characteristics of each restaurant, and users have to find out their own favorite one for each demand. In the reviews, there are two types of information, ``explanations of foods and drinks'' and ``experiences of eating and drinking.'' This paper aims to help users to easily find the reviews including ``explanations of foods and drinks.'' The proposed method applies the different types of resources concerning explanation and experience for eating and drinking as a pseudo dataset for constructing the classification model. We confirmed that the likelihood for each type of resource could be available to find the reviews describing the explanation of foods and drinks.