Effectively-Heterogeneous Information Extraction To Stimulate Divergent Thinking


Conflicts in different concepts are often useful in creating new ideas. We have proposed an outsider model in which an artificial agent provides ``effectively-heterogeneous'' information to support human divergent-oriented discussions. Subjective experiments using a prototype system based on the outsider model and a detailed analysis on results confirming that the outsider model can extract information containing hidden relevance, i.e., ``effective-heterogeneousness'', are presented.