WSEAS Transactions on Business and Economics
Print ISSN: 1109-9526, E-ISSN: 2224-2899
Volume 16, 2019
Applying Co-mention Network Analysis for Event Detection
Authors: Vladimir A. Balash, Alexey R. Faizliev, Elena V. Korotkovskaya, Sergei V. Mironov, Fyodor M. Smolov, Sergei P. Sidorov, Daniil A. Volkov
Abstract: This paper studies some characteristics and features of the economic and finance news flow. We consider a company co-mentions network as a graph in which nodes serve as the world’s largest companies mentioned in financial and economic news flow. Two companies (nodes) are linked if they were mentioned in the same news item. First, we analyze the dynamics of the structural properties of the company co-mentions network over time. Then we propose new method for event detection based on the company co-mention network. The idea behind the method is that more significant news should attract more attention and lead to an increase in the intensity of the news flow. A change in the intensity of co-mentions can be interpreted as a signal or marker of unexpected phenomena that may affect a relatively narrow or a wide range of economic actors. The analysis performed in the paper suggests that the decomposition of the co-mention matrices can be used to separate news signals. News corresponding to the stable part of the graph appear more often; respectively, they carry less information. The unexpected news revealed by the method described this paper deserves special consideration when making financial and investment decisions. The proposed approach to the selection of the event part can be used in the development of algorithms for detecting new events in the financial and economic sphere.
Keywords: finance and economics networks, degree distribution, market graph, event detection
Pages: 18-24WSEAS Transactions on Business and Economics, ISSN / E-ISSN: 1109-9526 / 2224-2899, Volume 16, 2019, Art. #3