NameIt: Extraction of product names

An important precondition for the semantic Web is to identify and annotate entities, their names, and their descriptions in the Web. In particular, the Web contains numerous Web pages describing various entities. In this paper we present a method for unsupervised generation of identities (i.e. product names) based on a set of concept instance describing Web pages. We exploit the redundancy of descriptions by statistical classification methods. We conducted an elaborated evaluation in order to identify the appropriate classification criteria and validated our system on two popular example domains. The result is a system for generating names which shows an F-measure of 0.9 in our experiments