Research Of Analysis And Quantization Algorithm For Reference Relationship Of Scientific Literature

In the field of information retrieval, the reference relationship of scientific literature is animportant information resource. Traditional citation analysis is applied widely in the fields ofcomputer science, information metrology and scientific metrology. But traditional citationanalysis mostly considers the cited count of literature, without considering the importance ofthe literature itself and the closeness between literatures, also does not have the quantitativework.The importance of a document can be measured by attention and value. Attention of aliterature is defined as the degree that it is quoted by other literatures of the field. Value of aliterature is defined as the degree of reference for other literatures of the field. Attention andvalue of a scientific literature are influenced by three factors. It includes the count of citing orcited documents, the importance of the document itself and the closeness between documents.Citation analysis can be visualized in the form of knowledge spectrum. We can use a weighteddirected acyclic graph to analyze the quotation in the form of logical structure. Noderepresents the literature. Directed edge represents the citation. Weight represents the closenessbetween literatures. The paper uses the index table to analyze in the form of the physicalstructure, and gives the index of referring literature, the index of referred literature and thecorresponding closeness. The paper introduces the concept and computing method ofontology correlation degree of terms and ontology correlation degree of documents, anddesigns an iterative algorithm to calculate attention and value.Original retrieval method uses the similarity computed by the vector model as the factor.The paper proposes an improved method, which uses similarity, attention and value as thebasis. It makes the average recall and precision and sort of the search results gotten by twomethods evaluate the results. Experimental results show that the improved method caneffectively improve the retrieval performance.