| Dietary is not only a basic condition for human survival and development,but also a topic often discussed in people's daily life.Through mining the dietary preferences of users,it can not only reveal the differences in eating habits among different groups,but also reflect the development and spread of regional food culture.The traditional research on dietary preferences is.·mostly based on questionnaire survey and interview,and many research productions have been achieved.However,these traditional methods have the defects of small scale of research objects and time-consuming of data collection.With the development of online social networking and the popularization of intelligent terminals,more and more people express and transmit information in the life through the Internet,resulting in massive user-generated contents.It provides sufficient data to study the user's dietary preferences.At the same time,the topics of massive user-generated contents are scattered,and the quality is uneven,which increases the degree of difficulty for users to obtain high quality dietary information to a certain extent.Therefore,this paper takes user reviews on“www.dianping.com”as data sources,and studies user preferences on two levels of region and individual.On this basis,we recommend restaurant information for network users.In the study of regional user dietary preferences,this paper utilizes the method for aspect-based sentiment classification to explore the dietary preferences of the regional users from three fields,including dietary concern,dietary similarity and dietary satisfaction.The results show that there are differences in user dietary preferences among different regions.And most of users are more concerned about their local cuisine.There is a significant correlation between geographical proximity and user dietary preferences.These results can provide references for restaurants in various regions to formulate menus and adjust business strategies.The main purpose of studying individual user's dietary preferences is to dig out the dietary needs or interests of individual users.In the study of individual user's dietary preferences,this paper divides user dietary interests into three categories:dish interest,dietary aspect interest and social interest.Then,the model of user dietary interest is represented by the vector space model based on keywords.The experimental results indicate that,based on the classification of user dietary interest,using the vector space model representation method based on keywords,we can obtain the dietary preferences of individual users to a certain extent.In addition,the model of user dietary interest based on the recent user-generated contents can be used to predict the future user dietary preferences better.In the application research of restaurant information recommendation,this paper constructs a representation model of restaurant information firstly,and then calculates the similarity between usr interest model and representation model of restaurant information based on the results of regional and individual user dietary interest model.Secondly,the method of measure user satisfaction in the study of regional user dietary preferences is utilized to measure users' expectations of restaurants.Finally,the recommended restaurants and their dietary information are determined for users based on the similarity of the similarity of model of user interest and restaurants information,and users' expectations of restaurants.Then,we evaluate the effect of the restaurant information recommendation.The experimental results confirm that the recommended method in this paper is effective,and it can provide users with interested and high-quality restaurant information to meet their personalized dietary needs to a certain extent.This study can not only provide a new research perspective for the study of food culture,but also provide reference for the practice of user information recommendation. |
PDF Full Text Request