Abstract
In order to ensure data mining analysis and knowledge discovery in the context of multi-disciplinary fields, big data analysis takes into account domain complexity, data analysis ease of use, and efficient execution. The domain-driven big data analysis process modeling is proposed to guide the construction and implementation of big data analysis process models. The big data analysis process is divided into a domain-oriented and platform-oriented two-layer model, and the transformation between the two is automatically done through the model-driven model mapping method. The domain-oriented analysis model is defined from the perspective of domain business, allowing users to focus on the analysis logic itself during the big data analysis process construction phase, without concern for the implementation details of specific algorithms. The platform-oriented analysis model is defined from the perspective of calculation and execution, so that the big data analysis process makes full use of the platform's computing resources, storage resources, and algorithm resources during the execution phase, the process execution efficiency and algorithm calculation speed are improved. Finally, the Hadoop platform as the underlying execution platform is taken as an example to illustrate the transformation process of the domain-oriented big data analysis process model.