Mining social behavior ideas of Przewalski horses
Literature introduces idea mining as an approach for extracting interesting ideas from textual information. Related research focuses on extracting technological ideas as starting point for future technological research and development activities. Thus, it is limited to the technological domain. The algorithms standing behind idea mining also are optimized for the technological domain. In contrast to previous research, this work transfers idea mining to the social behavior domain by selecting and adapting parameters of the idea mining algorithm. Forward selection as main approach in stepwise regression is used to choose the predictive variables based on their statistical significance. Grid search is used to optimize the parameter values. A case study shows that these optimized idea mining parameters are successful in extracting social behavior ideas of animals in this case of Przewalski horses. Based on these findings, differences between technological ideas and social behavior ideas can be shown.