ABSTRACT development of mobile devices and ubiquitous Internet




With the
boom of social media, it is a very popular trend for people to share what they
are doing with friends across various social networking platforms. Nowadays, we
have a vast amount of descriptions, comments, and ratings for local services.
The information is valuable for new users to judge whether the services meet
their requirements before partaking. In this paper, we propose a user-service
rating prediction approach by exploring social users’ rating behaviors. In
order to predict user-service ratings, we focus on users’ rating behaviors. In
our opinion, the rating behavior in recommender system could be embodied in
these aspects:

1) when
user rated the item,

2) what
the rating is,

3) what
the item is,

4) what
the user interest that we could dig from his/her rating records is, and

5) how the
user’s rating behavior diffuses among his/her social friends. Therefore, we
propose a concept of the rating schedule to represent users’ daily rating
behaviors. In addition, we propose the factor of interpersonal rating behavior
diffusion to deep understand users’ rating behaviors. In the proposed
user-service rating prediction approach, we fuse four factors—user personal
interest (related to user and the item’s topics), interpersonal interest
similarity (related to user interest), interpersonal rating behavior similarity
(related to users’ rating behavior habits), and interpersonal rating behavior
diffusion (related to users’ behavior diffusions)—into a unified
matrix-factorized framework. We conduct a series of experiments in the Yelp
dataset and Douban Movie dataset. Experimental results show the effectiveness
of our approach.




With the rapid
development of mobile devices and ubiquitous Internet access, social network
services, such as Facebook, Twitter, Yelp, Foursquare, Epinions, become
prevalent. According to statistics, smart phone users have produced data volume
ten times of a standard cellphone. In 2015, there were 1.9 billion smart phone
users in the world, and half of them had accessed to social network services.
Through mobile device or online location based social networks (LBSNs), we can
share our geographical position information or check-ins. This service has
attracted millions of users. It also allows users to share their experiences,
such as reviews, ratings, photos, check-ins and moods in LBSNs with their
friends. Such information brings opportunities and challenges for recommender
systems. Especially, the geographical location information bridges the gap
between the real world and online social network services. For example, when we
search a restaurant considering convenience, we will never choose a faraway
one. Moreover, if the geographical location information and social networks can
be combined, it is not difficult to find that our mobility may be influenced by
our social relationships as users may prefer to visit the places or consume the
items their friends visited or consumed before.

In our opinion, when
users take a long journey, they may keep a good emotion and try their best to
have a nice trip. Most of the services they consume are the local featured
things. They will give high ratings more easily than the local. This can help
us to constrain rating prediction. In addition, when users take a long distance
travelling a far away new city as strangers. They may depend more on their
local friends. Therefore, users’ and their local friends’ ratings may be
similar. It helps us to constrain rating prediction. Furthermore, if the
geographical location factor is ignored, when we search the Internet for a
travel, recommender systems may recommend us a new scenic spot without
considering whether there are local friends to help us to plan the trip or not.
But if recommender systems consider geographical location factor, the
recommendations may be more humanized and thoughtful. These are the motivations
why we utilize geographical location information to make rating prediction. The
main contributions of this paper are summarized as follows:

Ø  We
mine the relevance between ratings and user-item geographical location
distances. 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 help
us to understand users rating behaviors for recommendation.

Ø  We
mine the relevance between user rating differences and user-user geographical
distances. It is discovered that users and their geographically far away
friends usually give the similar scores to the same item. It can help us to
understand users’ rating behaviors for recommendation.

Ø  We
integrate three factors: user-item geographical connection, user-user
geographical connection, and interpersonal interest similarity, into a Location
Based Rating Prediction (LBRP) model. The proposed model is evaluated by
extensive experiments based on Yelp dataset. Experimental results show
significant improvement compared with existing approaches.