Research on offline behavior similarity of consumers based on Spatio-temporal data set mining

Abstract

The Spatio-temporal data set can be used in business research, For example, The user's geolocation check-in data (POI) in social media can be used to trace back the user's behavior track, however, the analysis of the similarity of LBSN users is not involved in the user's geographical location track. As a result, a density clustering method based on partition hierarchy and different neighborhood radius by users' geographical location is proposed to help explore similar measurement based on Spatio-temporal data set mining. The method observes the number of times a user visits each cluster region at different spatial location scales, and then calculates the similarity of users at each level by taking advantage of vector space model (VSM). Finally, users' similarity in Spatio-temporal(geospatial) behavior is obtained by superimposing user similarity values at different levels with different weights. The experimental results based on real user data of a large-scale social networking site in China show that the proposed method can effectively identify those users when they visit similar geographical locations.