Identifying Short-Names for Place Entities from Social Networks

Abstract

Organizations can be identified by a myriad of terms apart from their official names. While abbreviations remain a common "short-name" to reference organizations, the prevalence of other short-names has risen in conjunction with social networks. When a user enters a short-name as a locational search query, it remains a challenge to infer the relationship between the short-name and the organization it ostensibly represents. For a number of organizations around the Washington D.C., Maryland, and Virginia area, we first generate a list of possible short-names for each of them. We then search through their tweets to build a corpus of short-names associated with each organization. By measuring our list against the corpus, we can identify potential short-names, and return the location of the organization.