Extracting users' ideas in open innovation community using deep learning methods

The massive product experience and feedback in open innovation community could provide the inspiration of product innovation for enterprise. However, with the explosive growth of unstructured data, traditional opinion mining technology could no longer meet the real demand in terms of effectiveness and efficiency. To this end, this paper proposes a more efficient and timely method for idea extraction using deep learning algorithms. Specifically, firstly, this paper builds a multi-embedding layer CNN model with Dropout mechanism, called ME-CNN, to enhance the local feature extraction and to identify posts containing creative ideas. Then make full use of the feature of Transformer in capturing long-distance dependencies, and the strength of CNN in capturing local semantic information, we introduce a combined model, called TF-CNN, to achieve valuable sentences excluded non-creative texts. Finally, the hierarchical aggregation clustering method (HAC) is used to cluster creative ideas. Through experiments using real data, the results show that our proposed method performs well in extracting users' creativities, which can help enterprises obtain users' ideas from open innovation community more efficiently, and provide efficient decision support tools for product innovation. © 2021, Editorial Board of Journal of Systems Engineering Society of China. All right reserved.