Abstract
The terminological profiles of specialists are drawn up based on publications from the digital library eLIBRARY. By averaging the individual terminological profiles, it is possible to draw up a generalized profile (“portrait”) of a small research team (a department, laboratory, or sector). Comparison of individual profiles with the use of proximity measures (for example, a cosine measure) makes it possible to group similar profiles and identify groups of employees who conduct research in the same subject area. This helps determine the research team specialization by means of Text Mining tools without using subjective expert assessments. The results obtained using the profile approach are confirmed by constructing graphs of co-authorship and a graph of terms in the Gephi computer program. The compilation of terminological profiles was also used in the development of personalized scientific activity support systems. This system is intended for helping the user (a specialist in a subject area) in choosing relevant scientific conferences and searching for useful (pertinent as far as possible) publications. For describing text documents, a vector model is used, and the weights of terms are determined by calculating the term occurrence frequency (or the tfc-weighting formula). At the preprocessing stage, the stop-words and rarely encountered words are removed, and lemmatization is carried out. The developed profile approach has been approbated on the example of a small research team specializing in computer science. The terminological profiles were constructed and analyzed, based on which the areas in which the team members specialize have been identified, and a personalized scientific activity support system has been developed, that tracks, in an automated mode, publications in the eLIBRARY in one of the relevant areas (Data Mining).
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have