Application Analysis of Overlapping Community Detection Algorithms for Multidimensional Network Big Data and IoT
Xin Yang,Liping Cui,Yan Liu
DOI: https://doi.org/10.1155/2022/1172186
2022-01-01
Wireless Communications and Mobile Computing
Abstract:Capture the design and elements of these layers. Each layer corresponds to an alternative connection type between hubs in the normal world and requires tracking down communities in multidimensional networks. Most community disclosure approaches for multidimensional networks, then again, may ignore the transaction between layers or a layer’s unmistakable topological construction. Moreover, most of them are just equipped for distinguishing nonoverlapping communities. In this exploration, we offer another multidimensional network community disclosure strategy that exploits the connection among layers and the extraordinary geography of each layer to track down overlapping communities. First, use an overall assessment of edge behavior within and between layers to calculate the similarity of edges from similar layers and cross layers. You can then use these similarities to build a dendrogram of a multidimensional network that takes into account both characteristic topology structures and basic transactions. Finally, you can remove the overlapping communities in these layers by splitting the dendrogram and adding another community thickness metric for the multidimensional network. We show that our strategy is precise in recognizing overlapping communities in multidimensional networks by applying it to both manufactured and genuine world datasets. In chart and enormous information examination, community detection is a commonplace issue. It involves finding groups of firmly associated hubs with little associations with hubs outside the bunch. Distinguishing communities in huge scope networks, specifically, is a basic errand in numerous logical fields. In the writing, community detection techniques have been demonstrated to be wasteful, bringing about the improvement of communities with uproarious communications. To defeat this requirement, a framework that decides the best community among multifaceted networks in light of important determination standards and substance dimensionality should be created, eliminating loud communications in a continuous setting. Our outcomes likewise show that it is vital to utilize integral measurements to assess the exhibition of overlapping community detection calculations. Performance metrics, such as the NMI or the Omega Index, only measure the overall quality of a detected cover, whereas complementary metrics give us more information about the behavior of each algorithm in detecting overlapping communities. Finally, while some algorithms perform well on synthetic networks, none of the algorithms can detect the community structure in real networks. This is due to the detected communities of the algorithms being substantially different from the communities defined by the meta-data.