Internet browsing and searching: user evaluations of category map and concept space techniques

摘要: The Internet provides an exceptional testbed for developing algorithms that can improve browsing and searching large information spaces. Browsing tasks are susceptible to problems of overload vocabulary differences. Much the current research is aimed at development refinement by addressing these problems. Our was focused on discovering whether two our group has developed, a Kohonen algorithm category map browsing, automatically generated concept space searching, help and/or Internet. results indicate self-organizing (SOM)-based successfully categorize eclectic (the Entertainment subcategory Yahool) into manageable sub-spaces users navigate locate homepage interest them. SOM worked best with were very broad, in which subjects skipped around between categories. Subjects especially liked visual graphical aspects map. who tried do directed search, those wanted use more familiar mental models (alphabetic or hierarchical organization) found did not work well. from experiment encouraging. There no significant differences among precision measures set documents identified subject-suggested terms, thesaurus-suggested combination subject- terms. recall indicated terms exhibited significantly better than alone. Furthermore, analysis homepages there limited overlap retrieved Since most part different, this suggests user enhance keyword-based search using space. level control they could exert over fact suggested thesaurus real (i.e., originating homepages) therefore guaranteed have retrieval success.