INFORMATION SEEKING IN ONLINE CATALOGUES

Three different interfaces for online public access catalogues were evaluated using transaction logging as a methodology. In two cases exercises were set and run against each interface so that results could be critically evaluated to indicate areas of weakness. In all cases a random transaction log was derived and this was used to diagnose system problems and identify patterns of behaviour. Results showed that searches were often difficult to categorise, users employing different approaches to the same end. Users also were prepared to undertake sustained searching until something was retrieved; however they then frequently stopped. Search input problems are also identified and possible areas for further research suggested.