Patent attributes
This application provides a retrieval item generation method. In the method, a machine learning model is used to determine discarding/keeping conditions of terms in a query, and the machine learning model is obtained by training by a query sample having a target term feature. Compared with an existing retrieval item generation method, the target term feature can reflect importance of a term for a query result, and the machine learning model obtained by training by using the target term feature has a more accurate selection result of terms in the query, to generate a more accurate retrieval item, thereby avoiding recall of an unrelated query result. In addition, this application further provides a related apparatus for generating a retrieval item, to ensure actual implementation and application of the foregoing method.
