Ranking Given Names: Algorithms and Evaluation Paradigms

As a result of the author´s need for help in finding a given name for the unborn baby, the name ling, a search engine for given names, based on data from the ``Social Web´´ was born. Within less than six months, more than 35,000 users accessed name ling with more than 300,000 search requests, underpinning the relevance of the underlying research questions. The present work compares different metrics for calculating similarities among given names, based on co-occurrences within wikipedia. In particular, the task of finding relevant names for a given search query is considered as a ranking task and the performance of different statical measures of relatedness among given names are evaluated with respect to name ling´s actual usage data. By publishing the considered usage data, the research community is stipulated for developing advanced recommendation systems and analyzing influencing factors for the choice of a given name.