Topical similarity of scientific articles can be studied using bibliographic metadata such as keywords, titles, descriptions, and citation distribution. For their processing and further arrangement, the researchers use traditional co-citation analysis, bibliographic coupling, or text analysis. Scientometric papers report that they put more and more effort into distinct combinations of the mentioned approaches to verify obtained results. It follows from these studies that the structure, volume, and quality of data strongly determine the chosen method. If we lack a valuable representative dataset, our attention can be directed at modeling the appropriate metadata space. Thus, the problem of topical organization of similar research articles can be considered in context as a construction of a proper semantic space. We propose the method applying two factors to help searching the most relevant articles among a set of publications dealing with similar problems. One measure is based on commonly used keywords, while the other takes articles that the selected person cites or are cited by. We apply these factors to create a thematic proximity space, showing some distance from the chosen publication. Moreover, we define a type of Takagi-Sugeno fuzzy inference system for the publications’ classification. The proposed model of thematic proximity space based on keywords and papers’ citations can be used as a basis of information retrieval interface dedicated to researchers. They can apply it for relevant articles searching and filtering. In modelling that space, the fuzzy logic rules were used, constituting the novelty in this kind of study.