Ideas are essential for innovation and for the continuous renewal of a firm’s product offerings. Previous research has argued that online communities contain such ideas. Therefore, online communities such as forums, Facebook groups, blogs etc. are potential gold mines for innovative ideas that can be used for boosting the innovation performance of the firm. However, the nature of online community data makes idea detection labor intensive. As an answer to this problem, research has shown that it might be possible to detect ideas from online communities, automatically. Research is however, yet to provide an answer to what is it that makes such automatic idea detection possible?
Our study is based on two datasets from dialogue between members of two distinct online communities. The first community is related to beer. The second is related to Lego. We generate machine learning classifiers based on Support Vector Machines and Partial Least Squares that can detect ideas from each respective online community. We use partial least squares to investigate what are the words and expressions that allows for automatic classification of ideas. We conclude that ideas from the two online communities, contains suggestion/solution words and expressions and it is these that make automatic idea detection possible. In addition we conclude that the nature of the ideas in the beer community seems to be related to the brewing process. The nature of the ideas in the Lego community seems to be related to new products that consumers would want.