A bottom-up approach to assess the interdisciplinarity of journals from a multidisciplinary corpus of bibliographical records

International audienceThis work is investigating the possibility of assessing the interdisciplinarity of scientific and technological journals without using any taxonomy or classification scheme as those usually adopted in bibliographical databases. For that, we start from a large corpus of bibliographic records from which we extract terms, either keywords already present or obtained by text mining techniques. With the help of a clustering method, that corpus is split into clusters defining a given number of scientific fields. Those fields and the keywords indexing each document are the basis of the calculation of an interdisciplinarity score applying the diffusion model approach. Then, we calculate an interdisciplinarity indicator for each journal by combining the scores obtained by its articles