Virtual confrontation-based knowledge graph question and answer method and device and storage medium

Abstract

The invention discloses a virtual confrontation-based knowledge graph question and answer method, system and device and a storage medium, and the method comprises the steps: obtaining to-be-queried text information of a target user, carrying out the entity recognition of the to-be-queried text information based on an entity extraction model, obtaining target entity information, and if the target entity information has a preset target type entity, carrying out the entity recognition of the to-be-queried text information; if yes, intention recognition is conducted on the to-be-queried text information on the basis of an intention recognition model and a preset recognition rule, a target intention recognition result is obtained, and the entity extraction model and the intention recognition model are both obtained by conducting training on the basis of pre-collected to-be-trained corpus information in combination with a virtual confrontation training algorithm. And based on the target entity information, the target intention recognition result and the to-be-queried text information, performing data query in a graph database to obtain target query information. The technical problem of low accuracy of model identification is solved.

Description

technical field [0001] The present application relates to the technical field based on machine learning, and in particular, to a question and answer method, system, device and storage medium based on knowledge graph based on virtual confrontation. Background technique [0002] With the development of the Internet, the question answering system has developed rapidly. The intelligent question answering system is to find the text information that can best meet the user's intention from a large amount of data. The classification and extraction models currently used in the question answering system field are generally based on convolutional neural networks or Recurrent neural network implementation, using a large amount of existing real data to train the model. When the amount of training data is small, the model can only learn features through a small number of current samples, lacking a large amount of prior knowledge, resulting in a low recognition accuracy of the model. SUM...

Images