Community search from multi-attributed large social graph

Abstract

With the constant evolvement of social network structure, complex data, as well as graph structure, has been growing with increasing importance to model the interconnection of various entities. Community spot is a method of detecting densely connected sub-graph within a large network, for the given set of query vertex in the graph. It has many uses in social networking for instance when a certain user wants to get connected with other people or groups that go with the personality the user possesses. The main purpose of this method is to plot a well-organized mechanism to track the most dominant nodes as well as the corresponding meaningful communities that the vertex belongs to in an online manner. The multi attributed graph contains the data and statistics as the properties of the nodes as well as the probable relationship among the nodes. These details are used to ensure accuracy and to figure out the target community. The present-day methods of working do not have enough features to allow the attributes or keyword information associated with a given edge for searching for the desired community. We have worked on developing a new multi attributed community search algorithm that takes all the attributes of the edge into account and uses modern weighted search algorithms to find communities for given nodes. These explored nodes are densely connected and share a lot of common features. Our study was conducted in two phases. In the first place, a weight was assigned to each of the attributes matching up their significance. Then an algorithm was applied to the weighted decision matrix to form a single-attributed graph from the initial multi-attributed graph. A sub-graph with the least required weight assigned as the community weight was used to get a strongly connected community that the query vertex belongs to. Our system was built using the tools and built-in libraries of Python programming language. Thus our experimental procedure was used in searching for communities from given data that resembles the real world more closely.