A Spatial Maximal Co-location Pattern Mining Algorithm Based on Maximal Clique

• Published in: Proceedings (International Conference on Cyber-Enabled Distributed Computing and Knowledge Discovery. Online) pp. 18 - 26
• Main Authors: Qiaochen, Li, Xuguang, Bao
• Format: Conference Proceeding
• Language: English
• Published: IEEE 02.11.2023
• Subjects: Data mining; Feature extraction; Filtering algorithms; hash map; Indexes; Knowledge discovery; maximal clique; spatial data mining; Spatial databases; spatial maximal co-location pattern; Synthetic data
• ISSN: 2833-8898
• DOI: 10.1109/CyberC58899.2023.00014

Spatial co-location pattern mining refers to the discovery of a set of spatial features in large spatial data sets, and instances of these features frequently appear together in the geographic space. Spatial maximal co-location pattern is a lossless compression form of spatial co-location pattern, which solves the problem that users hard to extract effective information from too many spatial prevalent co-location patterns. Traditional methods for mining spatial maximal co-location patterns usually require verification of candidate patterns from low size subsets to high size subsets. At the same time, traditional methods need to spend a lot of time on the calculation of the participation index. Therefore, a spatial maximal co-location pattern mining algorithm based on maximal clique(CPPM-MC) is proposed to address the above shortcomings. CPPM-MC not only stores table instance information by introducing a hash table structure but also designs a variety of pruning methods to improve the efficiency of prevalent pattern judgment of candidate patterns. Firstly, a maximal clique enumeration algorithm is used to enumerate all the maximal cliques of an input spatial data set, and corresponding table instances are generated; then, a hash map-based mining framework is proposed to traverse each table Instances and filter out spatial maximal prevalent co-location patterns. Finally, the efficiency of the algorithm is verified by comparative experiments using synthetic datasets and real datasets respectively. Compared with the existing methods, the mining time of the CPPM-MC algorithm was reduced by more than 30%.