Incrementally satisfying a searcher ’ s information need

1. Introduction The Internet has become the virtual reality of mankind – a world that we shape without many of the imperfections of reality. We can jump to literally every place in no time and reach every resource anywhere anytime. In particular this last promise of information at your fingertips is under siege. The growing complexity of information space overwhelms the wired consumer and the vast increase in information is outpacing the improvement of retrieval tools. Traditional information retrieval systems aim to satisfy a searcher's information need by matching some explicit formulation of the searcher's information need to the set of available information carriers (such as documents, web-pages, etc.) and then returning the set of information carriers that best match the formulation of the searcher's information need. From the perspective of the searcher, this traditional approach has two serious drawbacks: Need formulation: Searchers are presumed to have a very clear understanding of their information need, even though it is not likely that they will be able to articulate their precise information needs in terms of a query language. In the case of the Internet, this becomes even more apparent as the collection of available information is sheer endless. Need satisfaction: Even when the entire set of returned information carriers is indeed relevant to the searcher's information need, they are still required to manually wade through the result sets in search of the right combination of information carriers to cover their information need. In traditional information retrieval systems, searchers are not provided with an advice on an effective order to best read (a selection of) the information carriers to cover their information need. The word 'effectively' may refer to time, financial costs, low overhead, etc. Other drawbacks have to do with the quality of document characterization and limitations of matching algorithms. On the next few pages we report on ongoing research. This research extends on work as reported in [Proper and Bruza, 1999; Weide et al, 1998]. The current state of the research is that we have (formally) defined the information coverage problem area in the style of [Proper and Bruza, 1999]. As a next step, we are developing strategies and algorithms to actually develop an (idealised) information portal.