A label architecture building method, adapted to a label architecture building system, the label architecture building system is configured to connect to an electronic device, wherein the electronic device is coupled to an enterprise system and a recommendation system, and access the enterprise system to obtain data of task and user information and data of user information, the label architecture building system comprises a processor and a storage device, the label architecture building method comprises:
- storing, through the storage device, a plurality of modules, wherein the storage device is coupled to the enterprise system and the recommendation system;
- and
- executing, through the processor, the plurality of modules, wherein the plurality of modules comprise a task profile label building module and a user profile label building module, wherein the task profile label building module is a first machine learning model respectively comprising a first classification model and an association analysis model, and the user profile label building module is a second machine learning model comprising a second classification model,
- wherein the step of executing the plurality of modules comprises:
- performing, through the processor inputs the data of task and user information into the task profile label building module, data mining to generate task label data;
- building, through the task profile label building module, a task profile label architecture according to the task label data by the first classification model and the association analysis model;
- performing, through the processor inputs the data of user information into the user profile label building module, data mining to generate user label data;
- and
- building, through the user profile label building module, a user profile label architecture according to the user label data by the second classification model, wherein the task profile label architecture is stored into a first database, and the user profile label architecture is stored into a second database, the first database and the second database are configured for access by the recommendation system, when the electronic device provides data of input information, the recommendation system generates recommendation information to the electronic device according to input information, the task profile label architecture, and the user profile label architecture,
- wherein the task and user information comprises task information and user behavior information, and the task label data comprises a first task feature data and second task feature data, wherein the step of performing data mining according to the task and user information in the enterprise system to generate task label data comprises:
- performing, through the task profile label building module, data extraction according to the task information to perform log mining to generate the first task feature data associated with a plurality of first features of the task information by applying unsupervised learning and calculation of natural language processing,
- wherein the step of performing data mining according to the task and user information in the enterprise system to generate task label data comprises:
- analyzing, through the task profile label building module, the task information according to the user behavior information to generate the second task feature data associated with a plurality of second features of the task information,
- wherein the user behavior information includes a task logfile,
- wherein the second task feature data includes an operating frequency with a task type and equipment requirements with process requirements,
- wherein the task profile label building module classifies the first task feature data and the second task feature data into a plurality of different categories, subdivides each classified category into a plurality of different subcategories, and accordingly builds these category data into a label architecture to be outputted as the task profile label architecture.