Utilizing the age of references to control the exhaustivity of the reference representation in information retrieval

摘要

The effectiveness of using the age of references to control the exhaustivity of the reference representation in information retrieval was investigated through analysis of optimal cluster-based retrieval results. The CF310 database, a subset of the CF database, was used. The two types of reference representations studied were the “foreground” representation, restricted to references with ages less than or equal to a specific age threshold, and the “background” representation, restricted to references with ages greater than this age threshold. It was assumed that the optimal level of exhaustivity for the foreground representation would be that age threshold at which the representation was restricted to references in the research front. It was also assumed that this representation would produced significantly better results than the exhaustive representation. The results show, as expected, that the foreground representation at its optimal level of exhaustivity is restricted to references of a relatively recent vintage—with ages less than or equal to seven. However, this representation does not produce significantly better results than the exhaustive representation. The background representation at its optimal level of exhaustivity is the exhaustive representation. Interestingly, the optimization of results for individual queries yields significantly better retrieval performance than the foreground representation at its optimal level of exhaustivity. Twenty-one out of 44 queries have optimal results with either the foreground or background representations, 15 have optimal results with just the foreground representation, and 8 have optimal results with just the background representation. The levels of exhaustivity for the optimal results vary considerably among the queries and, when the foreground representation is the optimal representation, span the complete range of age threshold values.