A Research Of Weighted Link Prediction Algorithm Based On Heterogeneous Social Network

With the continuous development of artificial intelligence technology,online Internet technology has accelerated the popularity of social networking applications.The network contains a huge amount of data,it also contains valuable data information and network topology features,including personal information and individual individuals.Relationship is an important part of social networks.However,how to find useful information in massive data and predict implicit links in the network has become an important research topic.Therefore,information analysis for social networks has become a hot topic.The main task of link prediction is to predict the future possible links through existing links in the current network.The link prediction problem in social networks has great practical application value and attracts more and more researchers’ attention and research.Due to the existing link prediction algorithm,there is a problem that the link weight is single,and it cannot be applied to a complex type of heterogeneous social network.Therefore,in this paper,based on heterogeneous social networks with weighted types,the concept of node weights is introduced into the link weighting process between users and users on the basis of constructing the heterogeneous network structure,which quantifies the time impact of users on the location signing behavior.At the same time,the location and user links,as well as the location and location links are weighted.Finally,a mixed weighted link prediction algorithm for heterogeneous social networks based on user location is proposed.In addition,according to the user’s behavioral relationship to location access,the user’s preference for location check-in behavior is regarded as a latent property,and the user-user link is weighted by bias,and an offset-weighted user link algorithm is proposed.Finally,this paper selects the real data set and conducts experiments and analysis on mixed weighted link prediction algorithm and biased weighted user link algorithm in heterogeneous networks.The results show that the mixed weighted link prediction algorithm for heterogeneous social networks based on user location has obvious prediction effect,and the biased user link algorithm has obvious prediction effect in the case of sparse data.