Tracing Knowledge Transfer from Universities to Industry: A Text Mining Approach

This paper identifies transferred knowledge between universities and the industry by proposing the use of a computational linguistic method. Current research on university-industry knowledge exchange relies often on formal databases and indicators such as patents, collaborative publications and license agreements, to assess the contribution to the socioeconomic surrounding of universities. We, on the other hand, use the texts from university abstracts to identify university knowledge and compare them with texts from firm webpages. We use these text data to identify common key words and thereby identify overlapping contents among the texts. As method we use a well-established word ranking method from the field of information retrieval term frequency–inverse document frequency (TFIDF) to identify commonalities between texts from university. In examining the outcomes of the TFIDF statistic we find that several websites contain very related and partly even traceable content from the university. The results show that university research is represented in the websites of industrial partners. We propose further improvements to enhance the results and potential areas for future
implementation. This paper is the first step to enable the identification of common knowledge and knowledge transfer via text mining to increase its measurability.