<i>Using Text Analysis to Determine Technology-intensive Firms from Technological Innovation Process Perspective</i>

Based on the perspective of technological innovation process, we reconstruct the indicator system for determining technology-intensive enterprises. The existing determination methods weaken the influence of underlying technology and overlook the influence of the technological innovation process. We use Computer-Assisted Text Analysis to generate characteristics of the technology-intensive industry. By analyzing the annual report texts, the technical intensity characteristics dimensions are summarized, and methods such as new word discovery and topic modelling are used to construct a technology intensity dictionary and a technology intensity index. Experiments show that the text-based technology intensity index is an extension of the existing output-based measurement index, capturing the technological innovation process of enterprises and having robust predictive performance for enterprise performance. Additional empirical validation shows that the text-based technology intensity measurement index can effectively identify the technology intensity characteristics of enterprises in different market sectors, different property rights, and different industries, ensuring the external validity of the index.本研究从技术创新流程视角出发，重构了技术密集型企业的判定指标体系。现有判定方法不仅弱化了底层技术的影响权重，还忽略了技术创新流程本身的作用。本研究采用计算机辅助文本分析（Computer-Assisted Text Analysis）方法挖掘技术密集型行业的特征：通过分析企业年报文本，归纳出技术强度特征维度，并借助新词发现、主题建模等手段构建了技术强度词典与技术强度指数。实验结果显示，基于文本的技术强度指数是现有基于产出的测度指标的延伸，能够捕捉企业的技术创新流程，且对企业绩效具备稳健的预测能力。进一步的实证验证表明，该基于文本的技术强度测度指数可有效识别不同市场板块、不同产权性质及不同行业的企业的技术强度特征，保障了该指数的外部有效性。