6. Ranking and Relevance Feedback
Society for Industrial and Applied Mathematics eBooks · 2005
Even though at first it seems that all search engines have the same basic features, one aspect that separates more full, functional search engines from their lesser counterparts is the ability not only to list the search results in a meaningful way (ranking) but also to allow the user a chance of using those results to resubmit another, more on-target query. This latter feature, referred to as relevance feedback, has been quite effective in helping users find more relevant documents with less effort [72]. Harman [36] has done extensive work in judging how relevance feedback positively affects system performance. Marchionini [58] points out another advantage of ranking and relevance feedback. As computer designers continue in their quest to make computers more receptive to users, ranking and relevance feedback can be considered as highly interactive information seeking. Along the same vein, Korfhage [44] refers to relevance feedback as a type of dialogue between user and system. Some commercial systems limit this feature and elect to forego building it into the system since relevance feedback can create computational burdens that can slow a system down. Also, it requires constant updating of the system's knowledge of itself because in order for the search engine to know if a certain document is the best it must know what every document in the system contains.