In this article, we report our implementation and comparison of two text clustering techniques. One is based on Ward's clustering and the other on Kohonen's Self-organizing Maps. We have evaluated how closely clusters produced by a computer resemble those created by human experts. We have also measured the time that it takes for an expert to "clean up" the automatically produced clusters. The technique based on Ward's clustering was found to be more precise. Both techniques have worked equally well in detecting associations between text documents. We used text messages obtained from group brainstorming meetings.

In this article, we report our implementation and comparison of two text clustering techniques. One is based on Ward's clustering and the other on Kohonen's Self-organizing Maps. We have evaluated how closely clusters produced by a computer resemble those created by human experts. We have also measured the time that it takes for an expert to "clean up" the automatically produced clusters. The technique based on Ward's clustering was found to be more precise. Both techniques have worked equally well in detecting associations between text documents. We used text messages obtained from group brainstorming meetings. q 1999 Elsevier Science B.V. All rights reserved.

Most comparative evaluations of Internet search engines provide soon outdated answers to a handful of pre-selected questions. Such an approach is too limited to be useful for practitioners wanting to evaluate information retrieval tools. Informed by an action research-oriented study of a large organisation conducting such an evaluation, this paper first discusses what questions to ask when analysing and evaluating intranet search tools. Applying Henderson and Cooprider’s approach and Saracevic’s theoretical foundation to the field data the author outlines a generic methodology for information system evaluation. The proposed methodology is refined through repeated interviews and discussions with people from a variety of organisational roles, thus taking a broader and more general approach. By suggesting that the selected features should be given different weights according to their relative importance and summarised in two dimensions, the methodology helps organisations detecting the most useful tool for their specific requirements.