No abstract

No abstract

No abstract

The retrieval effectiveness of five hierarchical clustering methods (single link, complete link, group average, Ward's method, and weighted average) is examined as a function of indexing exhaustivity with four test collections (CF, Cranfield, Medlars, and Time). Evaluations of retrieval effectiveness, based on three measures of optimal retrieval performance, confirm earlier findings that the performance of a retrieval system based on single‐link clustering varies as a function of indexing exhaustively but fail to find similar patterns for other clustering methods. The data also confirm earlier findings regarding the poor performance of single‐link clustering in a retrieval environment. The poor performance of single‐link clustering appears to derive from that method's tendency to produce a small number of large, ill‐defined document clusters. By contrast, the data examined here found the retrieval performance of the other clustering methods to be generally comparable. The data presented here also provide an opportunity to examine the theoretical limits of cluster‐based retrieval and to compare these theoretical limits to the effectiveness of operational implementations. Performance standards for the four document collections examined here were found to vary widely, and the effectiveness of operational implementations were found to be in the range defined as “unacceptable.” Further improvements in search strategies and document representations warrant investigation. © 1995 John Wiley & Sons, Inc.

An improved disambiguation mechanism, based on a recursive transition network, was used to enhance the automatic indexing system FASIT. FASIT with both the recursive transition network and alone showed improvement over single-term indexing in retrieval experiment with a medical test collection. However, FASIT with the enhanced disambiguation mechanism did not provide any noticeable improvement over FASIT alone. The failure of FASIT with the recursive transition network to improve upon the retrieval results of FASIT alone appeared to be due partly to the stepwise reduction of the effectiveness of further disambiguation in the FASIT environment. This factor was examined in a second experiment, which tested the effects of a perfect diambiguation mechanism on retrieval based on indexing representations provided by FASIT. Again, the difference in retrieval performance between FASIT alone and with a perfect disambiguation was minimal. The results corroborate the findings of the earlier experiment, which showed that retrieval performance is not noticeably affected by improvements to FASIT's disambiguation process. This experiment provides evidence that little improvement in retrieval performance can be expected, regardless of the degree of which FASIT's disambiguation mechanism alone is enhanced. Future research with FASIT should focus on other areas, such as refinements in concept selection and concept grouping and the incorporation of semantic information.