Research On Group Identification Method Based On Behavior Patterns

With the rapid development of modern information technology and computer technology,a large amount of data has been generated on the Internet for use.At the same time,with the rapid development of wireless communication technologies,a large amount of mobile object data has been spawned.These data portray individuals and groups.The space-time dynamics contain the behavioral information of mobile objects.By analyzing the mobile data of the target users,we can help us understand the behavioral rules and development trends of individuals and groups.Among them,group identification has always been a hot topic.Group identification in the broad sense refers to distinguishing the target objects into individual groups according to the specified characteristic information to identify the behavior of each group.The specific research focuses on group behavior identification and group abnormalities.Identification and other fields have different focuses and have important research value and broad application prospects,such as behavior prediction,target tracking,security protection,and traffic flow analysis.In this paper,the data mining method is used to extract the stay points according to the user's movement trajectory information,and then use a fast and simple clustering algorithm to mine the user's stay area.At the same time,it combines the POI information to mine the semantic information related to the stay area,and will stay in the area with the user.The relevant geographical location information and semantic information are used as the user's characteristic information,and the user similarity formula is customized,and the users are clustered by sharing the nearest neighbor clustering algorithm to mine the groups with strong relevance.Experiments show that the proposed method has a good effect on the mining of special users with strong relevance and provides good technical support for follow-up monitoring.The main research work of this paper is as follows:1.This paper proposes a method for hierarchically extracting location information of the stay region.Using the geographical position and time tags in the user's trajectory information,the frequency of their stay in a certain period of time and the frequency of going to each stop region is detected,and the fast and accurate mining of the geographic information of the target user's stop region is realized.2.This paper proposes a semantic information extraction method based on POI information and LDA topic model.The category information contained in the POI information can reflect the user behavior activity to a certain extent,use this information as the semantics,and use LDA to model the implicit topic of the information text and obtain the user semantic information probability distributions.3.This paper proposes a weighted user similarity measurement.In this paper,user's geographic location information,semantic information and the frequency of visits corresponding to each location are defined to define the user similarity formula,and the information contained in the user's trajectory data is fully utilized to embody user characteristics.4.This paper proposes a user clustering method based on shared nearest neighbors.Using such a clustering algorithm that emphasizes the degree of close relationship between users can effectively remove a large number of noise points and find close relationship groups.