Community Structure Detection in Social Networking Data Using Text Mining Approach

Abstract

In data mining techniques some of the problems are resolved using the visualization techniques. Among them some of techniques are derived from the graph theory and transparent data modelling. The data structures such as decision trees and semantic graph representation is one of the key implementation of the graph based solution development. Among these technique one of the mathematical model termed as the community detection is a part of data mining solution discovery technique. Data mining techniques are used for finding the application centric patterns recovery from the raw set of data. Additionally the community detection technique is a visual technique for performing the unsupervised learning. During community detection the data objects are keeping connected to represent the bounding among them. Therefore in order to perform categorization task in automatic manner this technique can be employed in different nature of data. In this presented work the social media text is used for community detection. Communities are the group of objects that are highly similar in their properties. Therefore an algorithm is proposed in this work, that first refines the text content, then the text features are computed form raw text. In next the data is evaluated to find the number of possible communities in the data and finally the data is grouped in the communities and their visualization is performed. The proposed algorithm not only used to find the community structure from the data that also provides the relationship among two different communities. The experimental results in terms of precision, recall, f-measures demonstrate the proposed model is efficient and accurate as compared to traditional clustering algorithms namely the k-means clustering.