Processing method and device in question-answering system based on knowledge graph

Abstract

The embodiment of the invention discloses a processing method and device in a question-answering system based on a knowledge graph. The method comprises: after a question sentence is received, obtaining a main entity corresponding to the question sentence in a preset knowledge base; selecting at least two candidate main entities from the main entities; taking each candidate main entity as a root node, searching an adjacent edge and an adjacent node of the root node in a pre-stored knowledge graph, searching an adjacent edge and an adjacent node of the adjacent node in the next layer through the adjacent node, and so on, until a node of the last layer is searched, and obtaining a path corresponding to each candidate main entity; calculating the similarity between the text information corresponding to the path of each candidate main entity and the text information of the question sentence; selecting a path of which the similarity meets a preset highest similarity judgment condition as acandidate path of a candidate main entity; obtaining a final selection path of the main entity from the candidate paths of the candidate main entity; and determining text information corresponding tothe final selection path as an answer to the question.

Description

technical field [0001] The embodiment of the present application relates to the field of information processing, especially a processing method and device in a knowledge graph-based question answering system. Background technique [0002] Question answering systems are an advanced form of information retrieval systems. Knowledge-based Question Answering (KBQA, hereinafter referred to as "knowledge question answering") is to help people acquire knowledge from the knowledge base through natural language dialogue. Knowledge question and answer relies on a large knowledge base (such as knowledge graph or structured database, etc.), converts users' natural language questions into structured query statements (such as SPARQL, SQL, etc.), and directly derives the answers required by users from the knowledge base . [0003] The knowledge base stores knowledge in the format of RDF (Resource Description Framework), and each piece of knowledge is expressed as a triplet, namely subject...

AI Technical Summary

Problems solved by technology

[0005] There are two major difficulties in related technologies. One difficulty is that the existing natural language understanding technology is still relatively weak in dealing with the ambiguity and complexity of natural language.

For example, sometimes a sentence system can understand it, but another sentence cannot understand it; another difficulty is that the knowledge question answering system needs a lot of domain knowledge to understand natural language questions, which requires a lot of labor costs

Images