ERIC Number: ED047760
Record Type: Non-Journal
Publication Date: 1968
Reference Count: N/A
Examination of the Validity of the Conclusions Arrived at in the Aslib Cranfield Research Project.
te Nuyl, Th. W.
The Cranfield Project was done in a laboratory sphere. In a practical situation, it is impossible to determine how many documents are relevant to a certain question. Although in the Cranfield Project collections of different sizes have been used, these collections do not fulfill the requirements that the small collections are randomly taken from the larger one. On the contrary, all the relevant documents in the small collection are the same as those in the large collection. The research work done does not give information to what extent the sampling method gives practical results. A more attractive method would be to accept figures obtained in the Cranfield Project for recall ratios at various coordination levels. In the Cranfield Project, coordination level of 3 means that out of 7-10 search terms all possible combinations are taken, but are counted only once independent of which terms match. This is inherent to the scan-column index technique used in the project, which is not easy to imitate in operational systems using computers, machine punched cards or manual punched cards. (Author/MF)
Descriptors: Indexing, Information Retrieval, Library Research, Performance, Relevance (Information Retrieval), Reliability
FID/CR, Danmarks Tekniske Bibliotek, Oster Voldgade 10, Copenhagen K., Denmark ($1.50)
Publication Type: N/A
Education Level: N/A
Sponsor: International Federation for Documentation, The Hague (Netherlands). Committee on Classification Research.
Authoring Institution: Danish Centre for Documentation, Copenhagen.