Behavior Pattern Mining And Personalized Recommendation Research Of Indoor Mobile Objects

According to statistics,more than 80% of human life in daily life is located in indoor spaces such as houses,offices,and shopping malls.Understanding and mastering the time and space behavior of the crowd can help with public disaster protection,public facility optimization,and indoor personalized recommendation services.In recent years,with the continuous development of indoor positioning technology and mobile intelligent terminal equipment,the positioning data and trajectory information of the collected indoor moving objects have become more accurate.This lays the data foundation for analyzing the spatial distribution,behavior patterns and preference information of individuals and groups in indoor space.At the same time,the mining and application of human behavior patterns based on indoor location has become a hot topic in current GIS research.This paper is based on the indoor passenger flow positioning data of a large shopping mall,combined with the physical characteristics of the indoor space of the mall to study the indoor WIFI positioning data preprocessing method and trajectory reconstruction method;combining the attribute information of the space entity in the mall,researching the trajectory clustering of the customer group,and mining the spatiotemporal behavior pattern of the shopping mall passenger flow;based on the user's individual needs,a personalized recommendation framework based on trajectory similarity is constructed.The main work and contributions of this paper have the following three points:(1)Aiming at the characteristics of indoor WIFI positioning data and the complex spatial structure of indoors,this paper proposes a method of indoor data preprocessing that takes into account semantics.Firstly,the heuristic filtering method and the sliding window algorithm are used to layer and segment the WIFI positioning data and perform filtering processing to solve the problem of the abnormal value of the positioning point;secondly,the concept of position semantics and semantic trajectory is introduced in the data preprocessing process,and indoor positioning points are converted into semantic trajectory sequences,which facilitates further mining of indoor trajectory information.The whole preprocessing process uses Hadoop platform and Spark computing framework for data storage and processing,which realizes the accurate and efficient conversion of original positioning information into the semantic trajectory sequence of key information in the reserved trajectory,which lays a foundation for trajectory clustering and behavior patternmining.(2)This paper proposes the E-DBSCAN algorithm.The E-DBSCAN algorithm uses the weighted edit distance as a metric for the distance between the trajectory sequences.The weighted edit distance takes into account the duration of the semantic position,the type of the store and the distribution of the floor,which fully reflects the semantic value of the trajectory sequence.In view of the abstraction of the semantic trajectory sequence,the E-DBSCAN algorithm redefines the cluster set and rules based on the DBSCAN algorithm,which improves the difference of the result set.According to the clustering result information,the behavior patterns of the moving objects in the result set are researched and divided,and the reference indicators for indoor applications such as personalized recommendation are provided.(3)Construct a personalized recommendation framework based on trajectory similarity.The framework adopts the idea of collaborative filtering.Based on the semantic trajectory clustering,the similarity between the semantic trajectory of the target user and the trajectory of the result set is calculated,and it is divided into the result set with the highest similarity.According to the current semantic trajectory,the trajectory information in the result set,and the association rules between shops,the candidate shops are extracted,and the scoring mechanism is established to score the interest of the candidate shops.Finally,the shops with higher predictive interest are selected for recommendation,which improves the accuracy of personalized recommendation in the indoor environment,can effectively improve the shopping experience of customers,and increase the profitability of shops in shopping malls.