A Bipartite Graph Based Competitiveness Degrees Analysis with Query Logs on Search Engine

Competitive intelligence analysis based on user generated contents (UGCs) shows advantages on possessing the benefit of the wisdom of crowds, evading cognitive biases and timely updating. This paper investigates and constructs a bipartite graph model by extracting joint co-occurrence of competitors from query logs, i.e., an important type of UGCs. The model represents the inherent competitiveness among competitors, based on which their mutual competitiveness degrees can be measured. Moreover, the BGComp algorithm is designed as well. Finally, an analysis on forecasting automobiles’ sales demonstrates the better forecasting power of the regression models integrating the competitiveness degrees than the benchmark model.