{"title":"Community Detection Based Clustering","authors":"Iveta Dirgová Luptáková, Jiri Pospíchal","doi":"10.1109/SISY.2018.8524662","DOIUrl":null,"url":null,"abstract":"Community detection is, in principle, a clustering of nodes based typically only on their topological properties that are derived from their positions in the network. Clustering generally uses non-topological information associated with nodes to group them. This paper uses a low-dimensional Euclidean distance of nodes to build a network (i.e. proximity or neighborhood graph) and applies community-based detection for clustering purposes. Nearest neighbors of nodes were connected by edges. Walktrap, edge betweenness, and fast greedy were used for community detection. The proposed approach generally proves superior to basic clustering methods, tested on popular 2D artificial benchmarks, and merits additional study. It also has lower computational complexity than other comparable approaches.","PeriodicalId":6647,"journal":{"name":"2018 IEEE 16th International Symposium on Intelligent Systems and Informatics (SISY)","volume":"31 1","pages":"000289-000294"},"PeriodicalIF":0.0000,"publicationDate":"2018-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"1","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"2018 IEEE 16th International Symposium on Intelligent Systems and Informatics (SISY)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/SISY.2018.8524662","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 1
Abstract
Community detection is, in principle, a clustering of nodes based typically only on their topological properties that are derived from their positions in the network. Clustering generally uses non-topological information associated with nodes to group them. This paper uses a low-dimensional Euclidean distance of nodes to build a network (i.e. proximity or neighborhood graph) and applies community-based detection for clustering purposes. Nearest neighbors of nodes were connected by edges. Walktrap, edge betweenness, and fast greedy were used for community detection. The proposed approach generally proves superior to basic clustering methods, tested on popular 2D artificial benchmarks, and merits additional study. It also has lower computational complexity than other comparable approaches.