In this paper we present the results of a study devoted to identification of longitudinal changes that occur in a given research field. The approach that is employed is based on full text analysis. Extraction of terms and their relations, along with thematic clustering are performed by the use of the freely distributed VOSViewer software. The latter allows to detect terms in the form of noun phrases and to cluster these terms with the help of a modularity based algorithm. Longitudinal development of the constructed thematic clusters is analyzed through the use of directed graphs that are built to reflect significant changes in their content at the level of their formation and development over successive subperiods. An alluvial diagram is employed to show the overall transformation of the thematic clusters. The utilized approach is applied to the proceedings of 'EuropaCat' catalysis conferences over a ten-year period. The conducted analysis shows that thematic clusters identified for the processed data are characterized by a low degree of stability. Even then, shifts of the researchers' interests from one theme to another can be clearly observed. Three most frequent types of cluster transformation are recognized: 1) continuance of a theme; 2) emergence of a new theme with its steady further growth; 3) emergence and discontinuance of a new theme. Main tendencies of temporal development of the detected thematic clusters are characterized in quantitative aspects.