Constructing Policy Domain Dictionary Generated by DTM-Embeddings to Identify Policy Response Features of Listed Companies in Electric Vehicle Industry

New technology industry policies play an important role in stimulating organizational transformation and motivating their strategic actions in responding to the policies. By developing an integrated method of dynamic topic modeling (DTM) and word embedding from BERT (referred to herein as DTM-EM), this study constructs a domain-specific dictionary to measure policy response in Electric Vehicle (EV) industry. The study initially employs dynamic topic modeling to extract key terms representing various themes in policy texts. Subsequently, utilizing similarity measures, it identifies synonymous terms corresponding to the key terms in the policy text, forming a word set for the policy text. Finally, utilizing similarity measures once again, it derives synonymous terms from the annual reports that is associated with the key words in the word set for the policy text to constructs a domain-specific dictionary. Furthermore, this study evaluates the accuracy of the dictionary for measure response dimension and features of Chinese publicly listed companies from 2009–2023.