Commercial POI Recommendation In A Location-based Social Network Environment

Location-based social network(LBSN)is realized by positioning technology and intelligent technology.It has amazing research value.However,when the total amount of information raises,the problem of information overload has become increasingly serious.To solve this problem,Point-of-Interest(POI)recommendation has become an important research problem.It is undeniable that in order to achieve efficient recommendations,accuracy is an indicator that must be considered.Most current commercial POI recommendation research also focuses on the realization of high accuracy.Only pursuing accuracy will make the final recommendation result over-professional,and the long tail problem will be serious.Therefore,how to achieve accurate and diverse commercial POI recommendations has become a hot issue that can be studied.This paper summarizes the current relevant literature.It analyzes the current research status of the commercial POI recommendation field,and the research situation of the recommendation algorithm.This paper also tells the strengths and weaknesses of existing methods.In summary,it is found that the accuracy and diversity methods can be used in combination to achieve the goal of accuracy and diversity balance.Therefore,this paper believes that after using the weighted ripple net model to achieve high-accuracy recommendation,the diversity of recommendation results can be enriched by re-ranking diversity recommendation method considering user diversity preferences.First,in order to achieve high accuracy,this paper proposes a recommendation method based on the weighted ripple net model.The current mainstream focus of recommendation is to implement recommendations based on geographic information,and the influence of social relationships is not much reflected in recommendation methods.However,the focus of the method proposed in this paper is to use friend relationships in social relationships to recommendation.It establishes a user's preference network through friend relationships and interaction records,and uses water wave diffusion between network nodes to activate each node.Then,it uses the weight of the relationship between nodes to measure the influence of geographic information and time on the diffusion process which helps to calculate more accurate interaction possibilities.Finally,a highly accurate recommendation list is obtained.After that,in order to enrich the diversity of recommendation results,this paper proposes a re-ranking diversity method considering user diversity preferences.In order to achieve more accurate and diversity recommendation,this paper optimizes the calculation of user diversity preference by using hybrid similarity.Based on the existing accurate recommendation list,the hybrid similarity method is used to more accurately calculate the user's diversity preference degree.Then the degree is used as a replacement condition for candidate commercial POI to achieve the rich in diversity.Finally,this paper use the real data provided by the Yelp platform to perform verification experiments and verify the variable parameters involved.The results of contrast experiments have proven that using recommendation method this paper proposes not only has clear advantage in accuracy,but also facilitates the realization of diversity.It ensures the balance between accuracy and diversity,and effectively alleviates the problems of over-specialization and long tail.