A Novel Topic Extraction Model for Science and Technology Demand Data
Haiyan Cui, Zhe Xue, Junping Du, Xin Xu, Junqing Xi · 2021
There are few studies focus on enterprise science and technology demand data, which is very important for enterprise development and innovation. These data are scattered on several websites and contain a lot of noise, which make it difficult to accurately analyze their topic. In this paper, the topic extraction algorithm based on deep learning is proposed to obtain the topic of demand in various industries. We adopt topic features clustering method to refine the classification of science and technology demand data. Keyword extraction method is proposed to filter the extracted theme words. The extracted topics are combined with time series to analyze the evolution of the topics and show the applicability of the extracted results of the science and technology demand data. A lot of experiments are conducted to verify the effectiveness of our algorithm. The optimal parameters and the number of topics are also analyzed in the experiments.