Automatic classification of web databases using domain-dictionaries

Abstract

The identification, classification and integration of databases on the Web (also called web databases) as information sources is still a great challenge due to their constantly growing and diversification. The classification of such web databases according to their application domain is an important step towards the integration of deep web sources. Moreover, given the design and content heterogeneity that exists among the different web databases, their automatic classification become a great challenge and a highly demanded task, requiring techniques that allow to cluster web databases according to the domains they belong to. In this paper we present a strategy for automatic classification of web databases based on a new supervised approach. This strategy uses the visible information available on a group of specific-domain Web Query Interfaces (WQIs) to construct a dictionary or lexicon that will allow to better describe a particular domain of interest. The dictionary is enriched with synonyms. In our experiments, the dictionary was built from a set of randomly selected specific-domain WQIs. The automatic WQI classification based on dictionaries generated in this way showed efficient and competitive results compared against related work. © 2013 Springer-Verlag.