Application of semantic and lexical analysis to technology forecasting by trend analysis - thematic clusters in separation processes

Abstract

Abstract The development trends in a given field of research. are very important factors influencing decisions on R&D funding. The analysis of the thematic clusters of the publications allows for the identification of the structure of given field of research. The number of the scientific publications and patents as well as their change in time are the measures of dynamics of development trends. The existing approach of structuring a research field, based on clustering of the sets of high co-occurrence words, does not take into account semantic and lexical similarities between words and requires an intervention by the experts in a given discipline. The paper presents an improved method for identification of thematic groups of papers, based on clustering of sets of high co-occurrence words, as well as application of financial analysis techniques for prediction of development trends. The method takes into account semantic and lexical similarity factors between words as well as clustering of the sets of words with the automatic rejection of the redundant groups of terms. The material presented in this paper is limited to the journal papers. It covers publications in the field of distillation presented in ISI Web of Science database for the period 1990-2010. The identified 26 thematic clusters of research have shown the diversified patterns of development, e.g. stagnation, revival, slow development or intensive growth.