Research On Personalized Place Recommendation In Location-based Social Networks

The tight integration of location services and social networking services has become a popular Internet trend,and location-based social network(LBSN)services have emerged.The LBSN provides users with a platform where the virtual world and the physical world have been combined.On this platform,massive data information can be used to discover potential interests of users and provide personalized recommendations for users.Among them,place recommendation becomes a research hot spot.However,existing place recommendation algorithms in LBSN are affected by data sparsity and cold start issues,resulting in unsatisfactory recommendation results.Different from the traditional recommendation,location information and social information are added in the LBSN place recommendation.We took advantage of the characteristics of the location social network,digging deeper into the user's social relationship and location information,and combined it with the user's interests with the goal of improving the accuracy of personalized place recommendations,the main research results are as follows:(1)In this paper,the intrinsic attributes of the place are considered.The similarity of place is calculated based on the traditional item-based collaborative filtering recommendation algorithm,and is integrated with the tag similarity anddistance similarity of location.So a place-based collaborative filtering(PB-CF)recommendation algorithm is proposed.(2)Based on the theory of six-degree segmentation in social networks,this paper analyzes and researches the potential influence of social factors on users,including trusting friends and trusting spreadable friends,and determining the scope of social influences more fine-grained.And integrated with user's preferences and similarity users so as to redefine the target user's neighbor users and generate a pre-selection recommendation list.Also it is necessary to incorporate the geographical location influencing factors in the personalized recommendation model in order to filter the pre-selection recommendation list.This paper proposes a social and location collaborative filtering(SL-CF)algorithm that combines social relationships and location effects.The effectiveness of the proposed method is verified by experiments on the real dataset Foursquare.The accuracy of the recommendation was improved significantly by the proposed algorithm compared to the original algorithm,and the negative effects of data sparse and clod start problem are mitigated to some extent.