Effects of Terms Recognition Mistakes on Requests Processing for Interactive Information Retrieval

In this work I propose to describe a new model of information retrieval. I study the effects of using weighted index terms recognition mistakes in a document indexing system and I evaluate indexing performance when short and long requests are used. The effect of weighting index terms in the document collection and in the requests is analyzed. Given the typical requests submitted to term indexing system, it seems easy to consider that the effects of term recognition mistakes in user requests must be severely destructive on the effectiveness of the system. The experimental study reported in this paper shows that the use of classical term Indexing technique for processing this kind of request is robust to considerably high levels of term recognition mistakes, in particular for long requests. Moreover, both standard pertinence feedback and pseudo pertinence feedback can be employed to improve the effectiveness of user request processing.