Intelligent question answering system construction method for air traffic control field knowledge graph

Abstract

The invention discloses an intelligent question answering system construction method for an air traffic control field knowledge graph, and the method comprises the steps: 1, carrying out entity recognition, which is used for recognizing an entity name appearing in a natural language question; 2, entity linking: establishing a linking relationship from entities in the question sentence to the entities in the knowledge graph by judging the semantic similarity of two entities in the question sentence and the knowledge graph; 3, predicting the relationship between the entity and the answer entity in the question sentence by using an encoder-decoder model, and further improving the model relationship prediction precision by using the additional information of the entity type; 4, further correcting an entity link result by utilizing the predicted relationship; and 5, retrieving a corresponding answer entity in the air traffic control knowledge graph by using the entity ID and the predicted relationship. The question and answer system is based on the knowledge graph and is different from a mode of performing template matching between the question and the answer, so that the answer has higher reliability.

Description

technical field [0001] The invention specifically relates to a method for constructing an intelligent question answering system of a knowledge map in the field of air traffic control. Background technique [0002] Google proposed the concept of knowledge graph in 2012 to enhance the function of its search engine. Traditional search engines screen and sort web pages based on the characters entered by users. Due to the lack of semantic relationship in their methods, they are increasingly unable to meet the needs of users under the current explosive growth of Internet knowledge. The emergence of knowledge graphs is to solve this problem. A feasible solution is provided. A knowledge graph is essentially a semantic network formed by entities with attributes linked through relationships. According to the oriented fields, the knowledge map is divided into general knowledge map and domain knowledge map. The general knowledge map contains a lot of common sense and covers a wide ran...

AI Technical Summary

Problems solved by technology

[0004] At present, with the rapid development of the national economy, the information data in the field of air traffic management continues to grow rapidly, and the use of air traffic management information tends to be complex. Under these circumstances, the traditional air traffic management information retrieval methods often cannot meet user requirements

Images