CO-WORD-BASED SCIENCE MAPS OF CHEMICAL-ENGINEERING .2. REPRESENTATIONS BY COMBINED CLUSTERING AND MULTIDIMENSIONAL-SCALING

In this paper we present the results of the second part of our study on science mapping. We discuss an improvement of co-word analysis based on a combination of a clustering technique applied to the word co-occurrence data matrix and multidimensional scaling of the resulting word-clusters. As in the first of our study, we construct journal-based, author-based, and conference-based co-word maps of chemical engineering. We apply word-similarity analysis to compare these different maps, using the views of the experts (as discussed in part 1) as a frame of reference. Next, we apply word-similarity analysis to compare maps from subsequent periods of time, in order to identify developments over time. An important improvement in the practical applicability of our maps is attained by labelling the co-word clusters with data on journals, classification codes, authors and countries. Finally, we compare our author-based co-word maps with a ''hybrid'' author-keyword map and with a co-author map. Important developments in chemical engineering are indeed visualized by our mapping techniques. The usefulness of the various types of maps is strongly related with the questions one wants to answer.