A data type identification and model training method and apparatus, and a computer device, wherein said model training method comprises: acquiring a first sample data set, and using the first sample data set to train an exception detection model; detecting an exception sample data set from a second sample data set by means of the exception detection model, and using the exception sample data set to train a classification model. With the described model training method, the number of classification model scoring events may be reduced and a relatively balanced sample data set for training may also be provided so that a fairly accurate classification model is obtained. By first inputting data to be identified into the exception detection model, it is possible to quickly distinguish whether the data to be identified is a first class of data, while other data not identified as the first class of data by the exception detection model is inputted into the classification model for identification, the online identification of data thus being relatively fast.
展开▼
