Human-aided filtering can make the difference (by Bob Zeidman).
Bob presents in this article how human-aided filtering can improve filtering accuracy. At first the author introduces an information retrieval model, dividing the domain (D) into the relevant area (D_U) as specified by the query (Q). External (e.g. page rank or the Hub-Threshold Kleinberg algorithm) and/or internal algorithm determine the result's ranking. Afterwards negative query filtering (with queries specified by a human) is applied to reduce the number of results and to boost the obtained accuracy.