This study aims at the systematic requirements of German external publicity terminology in cross-cultural communication scenarios and proposes a dynamic terminology database construction and governance framework based on big data technology. By integrating heterogeneous data sources and combining natural language processing technology with domain knowledge fusion methods, the dynamic update mechanism of term semantic representation is explored. The research results show that this framework can effectively enhance the domain adaptability of the terminology database and provide technical support for the accuracy and timeliness of the production of external publicity content.