( tilbake ) 

Nr 13 

Tord Høivik
Do the Web have the Answer?

The article is based on one hundred queries submitted during a fortnight - personally or by phone - to the main reference desk at Oslo Public Library in the early summer of 1997. We use the material to get a rough numerical estimate of the different types of questions asked - and, in particular, to judge to what extent answers are available on the Internet. In 1997 Internet had a success rate of 49%. When the same questions were tested again in 1999, the rate had increased to 61%. There was, however, a clear difference between questions that concerned Norway and Norwegian matters (47 questions) and the rest (53 questions). Norwegian questions had a success rate of 36% only. This tendency is even stronger for queries concerning specific facts, documents or images (29%) and for historical queries (25%). The Web was much more efficient for "non-Norwegian" questions (81%), for general subject queries (77%) and for queries on current affairs (70%). Success rates above 90% were realized for questions that combine several success factors.

Ragnar Nordlie
The user and the online catalog: a study of unassisted online search behavior

The article reports an investigation into the success of unassisted end-user online catalog searching in a public library. The data for the study comprise transaction logs of 175 search sessions where users pursue 350 search objectives through 1000 separate search requests.
 A somewhat rough generalization of findings from the study may be summed up as follows: Online searching by end users, particularly topical searching, is difficult, and does not get easier with the level of experience which average public library users are likely to achieve (Users remain "permanent novices"). Automatic error detection and search improvement functionality in the catalog system will probably not effect a substantial reduction of user problems.
 Users hardly ever apply advanced search facilities correctly; they state their search requests ambiguously and at a too low level of specificity; they fail to express their information need in the language required by the system; they have great difficulties with developing a search strategy based on previous search results. All these factors indicate serious communication problems between user and system. 
 The article suggests and motivates the employment of conversational techniques to improve system functionality and bridge the user/system gap.

Berit Irgens
A web-based interface to an OPAC may facilitate learning by exploration, but mismatch problems remain unsolved

This paper reports from an empirical study of how end-users utilize an online bibliographic information system available on the World Wide Web. The use of a union catalogue of information about the books and journals held by 70 academic libraries was investigated. Five doctoral students were interviewed about their use of the system, and log data of their use during a 7-month period were examined. One respondent never used the system. One respondent had integrated use of the system in reading and writing activities in a way not possible when use was restricted to the physical library. The last three respondents preferred to use the system in their workplace instead of at the library to save time.
 The log analysis indicated that visual cues provided by the graphical user interface - hypertext links and action buttons ­ appeared to encourage the user to explore the system, though most search sessions did not have explorative moves. However these features should be used more extensively to help users dealing with problematic situations. Conclusively the design of a new iteration of the study is sketched, and advice for system design based on the empirical observations is given.

Mildrid Ljosland
Evaluation of Web search engines

In this paper some evaluations of Web search engines is presented. After presenting related works, the author describes the results from two of her experiments on relevance. One is a comparison between three Web search engines, and the other is a study of the effect on the relevance of using different index sizes in a search engine. The paper also includes a discussion on evaluation methodology.

Maja Anjer og Are Egner Paulsen
Visual Dewey : Subject maps on Internet

The article describes a prototype of a system by which one can visualize the subject content in a library catalog, as a map image. The method uses dewey number as basis for clustering, or rearrangement, of the document multitude, and by this means exploits the cognitive work that is already part of the classification. The result is the application "Visual Dewey" - an interface that can be shown on the world wide web, and which is meant to partly substitute browsing in the physical library.

Kristin Røijen
Can documents on the World Wide Web be of relevance to the user?

The article investigates the relevance judgements of a researcher and a librarian confronted with the same WWW reference list. For the researcher, relevance judgement was an iterative process. Only negative judgements could be made on the basis of the reference list alone. Positive judgements required recourse to the documents themselves and their provenance. The librarian makes more ”partly relevant” judgements. The article discusses a number of features which might improve reference lists to facilitate relevance judgements.

 ( til toppen )