Research On Co - Location Model Mining Based On Sub - Graph And Maximal Ordered Tree

Currently, previous study has researches in how to mining spatial co-location patterns using data structures, such as tree and graph. Graph is a common data structure as well as builds complicated relationships among data. It also has the widespread use of web searching and social network. However, the process of mining maximal cliques from large graphs become very hard due to there are a lots of sub graphs. Thus this study adopted two methodologies, the method of using the tree of maximal ordered cliques and the method of using the graph of maximal cliques. Moreover, this study analyzed in-depth on algorithm, and modify the algorithm appropriately by analyzed results. Through add the mechanism of Candidate-test, the latter algorithm got better performances on speed and function. This study made some analysis in functions of the Candidate-test’s mechanism, and got the preliminary conclusion. Although the Co-location model algorithm which adopt the method of enumerating all maximal cliques in graph is not as good as other algorithms, this algorithm not only can digging simple Co-location model, but also can digging complicated Co-location model, its’ worth to made the research. In future, this study can focus on two directions:one way is to research the new participation index which can suit for the method of enumerating all maximal cliques in graph; another way is to prove functions of the performance of Candidate-test’s mechanism by mathematic methodologies.