Authors : K.P. Swaraj and D. Manjula
Abstract: As a consequence of advanced research occurring in various domains there is always a hunt for emerging hot topics and expert authors of such trending topics. An interesting phenomenon noticed is that the topics of publications of a group of researchers at a particular time interval have similarity and therefore exhibit similar multi hot topic interests. Efficiently finding such expert author communities having similar multi hot topic interests which are closed and complete on a temporal basis is the major contribution of this study. The novel framework proposed in this study and the maximal Biclique based SMTAC_Find algorithm effectively detects such communities with multi hot topic interests. At first top-k hot-topics of a time interval are discovered utilizing article clusters created by ontology based domain classification system. Subsequently for a chosen time slot, the novel maximal biclique based algorithm is employed to detect the expert author communities having maximal similar multi hot-topic interests. DBLP dataset and its associated metadata of computer science articles were used for getting bibliographic information. A synthetic dataset of varying number of authors and their hot topic interests was used to verify Proposed SMTAC_Find algorithm. Analysis and performance of the algorithm depict that it can efficiently find expert communities in polynomial time for a specified number of hot topics ‘k’ under scoring its effectiveness.
K.P. Swaraj and D. Manjula, 2016. Discovering Expert Communities from Ontology Based Document Classification: An Efficient Maximal Biclique Approach. Asian Journal of Information Technology, 15: 1223-1231.