EMBRACE INDUSTRY 4.0. A DATA-DRIVEN METHOD ON DIGITAL TRANSFORMATION EVALUATION OF CHINA'S MANUFACTURING INDUSTRY

In the process of moving towards Industry 4.0, identifying and evaluating the strategies of enterprises is a noteworthy issue in the field of digital transformation. The advancement of natural language processing and deep learning has facilitated the extraction and analysis of strategies from enterprise disclosures. In this paper, we develop a data-driven evaluation method for digital transformation of enterprises using unstructured text data from annual reports. With topic generative natural language processing, we identify six strategic topics and related business environment topics for Chinese manufacturing companies in the context of digital transformation, and express the intensity of companies' attention to different strategies in terms of topic likelihood scores. Using topic score data, we train an artificial neural network based on deep learning methods to characterize the relationship between the business environment and strategy, which finally helps to achieve prediction and evaluation of enterprise decisions. This study extends the traditional case study approach through a data-driven method, we use bulk topic recognition of unstructured text and deep learning characterization instead of individual enterprise questionnaires or field researches. In addition, this research fills the gap in information mining of enterprises annual reports under the topic of digital transformation and broadens the utilization of deep learning in management. We deploy this data-driven approach on a dataset and compare it with traditional research methods in one enterprise case.