ABSTRACT  With theboom of social media, it is a very popular trend for people to share what theyare doing with friends across various social networking platforms.

Nowadays, wehave a vast amount of descriptions, comments, and ratings for local services.The information is valuable for new users to judge whether the services meettheir requirements before partaking. In this paper, we propose a user-servicerating prediction approach by exploring social users’ rating behaviors. Inorder to predict user-service ratings, we focus on users’ rating behaviors. Inour opinion, the rating behavior in recommender system could be embodied inthese aspects: 1) whenuser rated the item, 2) whatthe rating is, 3) whatthe item is, 4) whatthe user interest that we could dig from his/her rating records is, and 5) how theuser’s rating behavior diffuses among his/her social friends. Therefore, wepropose a concept of the rating schedule to represent users’ daily ratingbehaviors.

In addition, we propose the factor of interpersonal rating behaviordiffusion to deep understand users’ rating behaviors. In the proposeduser-service rating prediction approach, we fuse four factors—user personalinterest (related to user and the item’s topics), interpersonal interestsimilarity (related to user interest), interpersonal rating behavior similarity(related to users’ rating behavior habits), and interpersonal rating behaviordiffusion (related to users’ behavior diffusions)—into a unifiedmatrix-factorized framework. We conduct a series of experiments in the Yelpdataset and Douban Movie dataset. Experimental results show the effectivenessof our approach.

 INTRODUCTION With the rapiddevelopment of mobile devices and ubiquitous Internet access, social networkservices, such as Facebook, Twitter, Yelp, Foursquare, Epinions, becomeprevalent. According to statistics, smart phone users have produced data volumeten times of a standard cellphone. In 2015, there were 1.9 billion smart phoneusers in the world, and half of them had accessed to social network services.Through mobile device or online location based social networks (LBSNs), we canshare our geographical position information or check-ins.

This service hasattracted millions of users. It also allows users to share their experiences,such as reviews, ratings, photos, check-ins and moods in LBSNs with theirfriends. Such information brings opportunities and challenges for recommendersystems. Especially, the geographical location information bridges the gapbetween the real world and online social network services. For example, when wesearch a restaurant considering convenience, we will never choose a farawayone.

Moreover, if the geographical location information and social networks canbe combined, it is not difficult to find that our mobility may be influenced byour social relationships as users may prefer to visit the places or consume theitems their friends visited or consumed before.In our opinion, whenusers take a long journey, they may keep a good emotion and try their best tohave a nice trip. Most of the services they consume are the local featuredthings. They will give high ratings more easily than the local. This can helpus to constrain rating prediction. In addition, when users take a long distancetravelling a far away new city as strangers. They may depend more on theirlocal friends. Therefore, users’ and their local friends’ ratings may besimilar.

It helps us to constrain rating prediction. Furthermore, if thegeographical location factor is ignored, when we search the Internet for atravel, recommender systems may recommend us a new scenic spot withoutconsidering whether there are local friends to help us to plan the trip or not.But if recommender systems consider geographical location factor, therecommendations may be more humanized and thoughtful. These are the motivationswhy we utilize geographical location information to make rating prediction. Themain contributions of this paper are summarized as follows:Ø  Wemine the relevance between ratings and user-item geographical locationdistances. It is discovered that users usually give high scores to the items(or services) which are very far away from their activity centers. It can helpus to understand users rating behaviors for recommendation.

Ø  Wemine the relevance between user rating differences and user-user geographicaldistances. It is discovered that users and their geographically far awayfriends usually give the similar scores to the same item. It can help us tounderstand users’ rating behaviors for recommendation.Ø  Weintegrate three factors: user-item geographical connection, user-usergeographical connection, and interpersonal interest similarity, into a LocationBased Rating Prediction (LBRP) model.

The proposed model is evaluated byextensive experiments based on Yelp dataset. Experimental results showsignificant improvement compared with existing approaches.


I'm Erica!

Would you like to get a custom essay? How about receiving a customized one?

Check it out