Evaluating the competitiveness of enterprise's technology based on LDA topic model
Evaluating the competitiveness of enterprise's technology based on LDA topic model
复制标题
基于LDA主题模型的企业技术竞争力评价
DOI：
10.1080/09537325.2019.1648789
复制
发表时间：
2019-08-02
影响因子：
3.4
通讯作者：
Wang, Jiayun
中科院分区：
文献类型：
--
作者：
Wang, Xuefeng;Yang, Xuemei;Wang, Jiayun
随着知识经济的到来，技术已经成为许多企业前进的基础。企业要保持持久的竞争优势，必须准确判断特定技术的竞争力。然而，新兴创新正变得越来越复杂，跨学科趋势使问题变得更加复杂。在基于专利的竞争力评价中，传统的专利分类既松散又具有时间敏感性。本文基于潜在狄利克雷分配（LDA）主题模型生成的技术主题，构建了企业技术竞争力评价模型。 LDA主题模型能够将技术分为更窄的类别，因此可以提供有关某个领域竞争格局的丰富信息。模型中采用两个指标来衡量企业的技术竞争力——专业化指数和多元化指数。同时，我们通过相对份额、技术的吸引力以及技术可能给企业带来的竞争优势来探索不同技术主题的企业分布。智能网联汽车的实证研究验证了模型，结果为制定研发策略和/或做出投资决策提供理论支持。
With the advent of the knowledge economy, technology has become the foundation of advancement for many enterprises. To maintain a lasting competitive edge, enterprises must accurately determine the competitiveness of particular technologies. However, emerging innovations are becoming more and more complex, and interdisciplinary trends complicate matters even further. In the competitiveness evaluation based on patent, traditional patent classifications are both loose and time-sensitive. In this paper, we constructed the evaluation model of enterprise's technology competitiveness based on the technological topics generated by a latent Dirichlet allocation (LDA) topic model. LDA topic model is able to classify technologies into narrower categories and can, therefore, provide rich information on the competitive landscape of a field. Two indexes are used to determine the technological competitiveness of an enterprise in the model - a specialisation index and a diversification index. At the same time, we explore the distribution of enterprises with different technological topics through the relative share, the technology's appeal, and the competitive advantage that technology might give an enterprise. The empirical study on intelligent connected vehicles validates the model, and the results provide theoretical support for developing R&D strategies and/or making investment decisions.