| Summary: | In this essay, a search strategy for citation index is studied. The strategy, which is essentially a citation cycle, starts with documents retrieved by a subject search, wherefrom new documents are identified following the network of citations backwards and forwards in time. Based on the theory of polyrepresentation, the strategy exploits overlaps between cognitively different interpretations of the same documents in order to automatically select references seed documents used as a starting point for the forward chaining. The purpose of the investigation is to evaluate the retrieval effectiveness of the search strategy and to find out whether the strategy can be used to expand a subject search with the help of the network of references in order to retrieve new relevant documents. Two questions are formulated: 1a. What is the difference in precision between the initial subject search and the citation search? 1b. What is the novelty ratio for the citation search? 2. Is there a larger proportion of relevant documents in the overlaps emerging from the citation search, especially overlaps generated by several interpretations? The searches are performed in Science Citation Index and Social Science Citation Index. Results show that the citation search retrieves additional relevant documents, but that the average precision is low compared to the subject search. On the other hand the average novelty ratio for the citation search is rather high, with an average of 39 %. Very few overlaps on higher overlap levels emerge from the citation search in this investigation. The last question has therefore been difficult to answer. === Uppsatsnivå: D
|
|---|
