In wenigen Tagen ist es wieder soweit und die Mathe-Adventskalender im Internet öffnen ihre...
Namesakes for the HLRN-III complexes revealed
Kennziffer WA 40/13
Kennziffer: WA 39/13
Bibliometrics - Classification and Similarity analysis for mathematical publications
Project cooperation with the Institute for Research Information and Quality Assurance (iFQ), Competence Centre for Bibliometrics and University Bielefeld - Institute of Science and Technology Studies (IWT)
In contrast to previously followed article classification approaches at Competence Centre for Bibliometrics where only a portion of available metadata (for example, co-word and co-citation analysis) was used, we want to use all available metadata (such as title, author, journal with bibliographic data, abstract, references, author slogans, classification of the journal, etc.) and additional full texts. Different approaches of the analysis of document similaritiare will be combined and evaluated.
ZIB reaserch and experiences have shown that the concentration to only a portion of the available information will not produce useful results. The results of the different approaches and their combination will be compared to each other and to existing classifications or thesauri and then validated by experts.
Our goal is to develop an application for mathematics as a proof of concept. Based on a metadata graph, it can classify, bibliometrically analyze documents and find similarities among them.
The basis for the further apporach is the metadata graph which has to be build from the iFQ data. We develop algorithms that can calculate a mapping between publications and one or more classes of a given classification (for example, Mathematical Subject Classification MSC). In addition, the developed algorithms enable bibliometric analysis such as time courses, citations and co-authorship.
Other algorithms can use the metadata graph structure and the classification to define a similarity measure for publications.