The Research Of Restaurant Recommendation Based On Check-in Data

Personalized recommendation services not only improving people's living experience but also increasing revenue for the service providers.For this reason personalized recommendation in social networks becomes a research subject with great theoretical and practical value.But,current research on location-based recommendation methods is often due to the sparsity of check-in data and the recommendation performance is bad.In this paper,we take the restaurant recommendation as the research example,the proposed approaches can solve the user cold start problem and improve the accuracy and the diversity of the recommendation system.Firstly,based on the restaurant attributes,an enhanced content-based restaurant recommending method is proposed.In this method,the characteristics of restaurants in the check-in data is used to establish a interest model for each user by TFIDF algorithm,then check-in possibility of each restaurant was quantified with the user interest model,and the candidate restaurants with high possibility would be recommended.The experimental results show that the precision and recall with cold start problem improves about 4.34% and 7.78%respectively.Secondly,in order to solve the data sparsity problem and improve the diversity of recommendation system,a hybrid restaurant recommending method based on collaborative filtering and user model is proposed.In this method,an enhance Collaborative Filtering algorithm is used to utilize temple information,then based on the restaurant category,we learn to predict users' potential interest,finally,the temple influence and the user potential interest is combined to make personalize restaurant recommendation.The experimental results shows that the precision,recall and diversity of the hybrid method improves about4.18%,14.51% and 4.85% respectively.In conclusion,this paper proposed two methods to provide high accurate and diversity for restaurant recommendation based on check-in data.With such a work,it is beneficial to give users more accurate and diverse restaurant service.