09-05-2023 дата публикации
Номер: CN116092699A
Принадлежит:
The invention relates to a cancer question and answer interaction method based on a pre-training model, and the method comprises the steps: 1, collecting and sorting cancer and inducement data, and training a BERT model which is related to cancer and is combined with intention recognition and entity slot filling; 2, ERNIE model parameters are set, and ERNIE model training is carried out; step 3, collecting cancer medical data, preprocessing the data, and performing fine tuning training by using the Transform-XL pre-training model to obtain a natural language generation Transform-XL model of the cancer; and 4, inputting a question to be consulted, then inputting a text into the trained BERT model, performing cancer intention recognition and entity slot filling so as to calculate obtained entity and intention confidence, calculating an ERNIE model by using the trained text similarity, calculating the similarity of each item, and generating a Transform-XL model by using a natural language ...
Подробнее