Automated Natural Language Feedback Generation in an interactive Information Retrieval System

摘要

The Internet has tens of millions of sites at this point and the growth is exponential without any bibliographic control. To find the proverbial needle in this immense haystack new retrieval methods and search support strategies need to be proposed. As such, information retrieval systems should (1) bridge the semantic gap between the user's and the author's vocabulary and (2) give the user feedback to his search result in order to refine the search request. In this paper we present an automated natural language feedback generation formalism which integrates syntactical knowledge via supertags and domain knowledge via an ontology. The following tries to give a rough idea on how these different knowledge is combined and why we hope that this would help to build up a natural language driven information retrieval system.