Research On Novelty Evaluation Of Crowdsourcing Idea Based On Text Mining

Nowadays,the development of enterprises is facing great competitive pressure.The independent innovation within the enterprise can no longer meet the needs of the public.More and more enterprises begin to use various crowdsourcing innovation platforms to obtain external innovation resources.Crowdsourcing breaks the traditional closed boundary of innovative enterprises,making enterprises more likely to produce more valuable heterogeneous innovation ideas.Efficiently identifying breakthrough ideas with high novelty in massive ideas is of great significance for improving the speed and efficiency of innovation.Firstly,a crowdsourcing idea novelty calculation model based on text mining,topic recognition and outlier detection algorithm is constructed.TF-IDF and Doc2 vec text vectorization are used to import LDA topic model for topic recognition.The generated document-topic distribution is imported into four outlier detection algorithms(cosine distance-based algorithm,KNN algorithm,LOF algorithm and IF algorithm)for outlier detection.Then,the algorithm combination of the model is tested with Dell Idea Storm and My Starbucks Idea datasets.Finally,the selected algorithm model with better experimental results is used to evaluate the novelty of Xiaomi MIUI community crowdsourcing ideas to identify the breakthrough innovations with high novelty.The results show that in the experimental study of the algorithm combination the TF-IDF-LDA-IF algorithm has better experimental results in identifying innovative ideas with high novelty in massive crowdsourcing ideas.In the empirical study of Xiaomi MIUI community crowdsourcing idea novelty evaluation,the breakthrough creative content is summarized into five aspects: common function,minority function,program operation,system update and system vulnerability.The effectiveness of the crowdsourcing idea novelty calculation model proposed is verified with Xiaomi dataset.