Clustering the Results of Brainstorm Sessions: Applying Word Similarity Techniques to Cluster Dutch Nouns

This research addresses the problem of clustering the results of brainstorm sessions. Going through all ideas and clustering them can be a time consuming task. In this research we design a computer-aided approach that can help with clustering of these results. We have limited ourselves to looking at single words, while at the same time we identify the different factors that could influence the results. These factors are: (1) word similarity algorithm, (2) dimensionality, (3) cluster count, (4) clustering algorithm, and (5) the evaluation approach. In total we tested six word similarity algorithms, two clustering techniques and three evaluation methods, in order to see which configuration works best for the task. We found evidence that the clustering of these results is feasible, though the results are influenced by the subjective behaviour of human interpreters.