The Impacts of Spanning Implicit Categories in Online Markets: Mapping Stylistic Categories from Keywords

• In Event: B.E.S.T. Session: Knowledge, Innovation, and Media

Abstract

Systems of categorization structure markets by aligning the expectations of producers and consumers. In numerous domains, however, formal data on market categories is not available. Our paper uses social tagging data to track the features that products include. We draw on methodologies from the fields of data mining and network analysis to group tags together into clusters of closely related elements. These implicit categories describe the unarticulated norms of style and content that shape buyer behavior in an online marketplace, even when these categories are not explicitly defined. As a predictor of sales, we demonstrate implicit categories operate just as formal categories do; more conventional combinations of tags lead to higher sales than those with more unusual, boundary-spanning sets of tags. This finding implies that the power of categories to structure markets and the negative impacts of spanning these categories are apparent even when the boundaries themselves are not.

Authors

• Matthew Sargent, RAND Corporation

• Joshua Andrew Clark

• Peter Monge, U of Southern California

• Janet Fulk, U of Southern California