Question and answer method based on knowledge map

Abstract

The invention provides a question and answer method based on a knowledge map. The question and answer method based on a knowledge map provided in the invention is realized by subject entity matching,relationship matching and answer determination. The subject entity matching mainly comprises naming entity identification and entity linking. The naming entity identification is aimed at identifying naming entities such as names of people, names of places, and names of organizations in natural language questions q. The entity linking corresponds the identified naming entity to a certain entity inthe knowledge base, that is, finding out an entity s in triples; Relationship matching is to understand the semantics expressed by question q through natural language understanding technology, and match the relationship p in the triples (s, p, o) in the search space in order to determine the semantics of the question and its corresponding relationship with the knowledge base. The candidate subjectentity is obtained through entity identification and entity linking, and the relationship matching can obtain the candidate relationship, thereby obtaining several candidate triples; the answer determination is to rank the candidate triples according to entity recognition score, relationship match score, etc. to determine the final answer.

Description

technical field [0001] The invention relates to a retrieval method, in particular to a question answering method based on a knowledge graph. Background technique [0002] Question Answering System (QA) is an advanced form of information retrieval system. It can answer questions raised by users in natural language in accurate and concise natural language. The main reason for the rise of its research is people's demand for fast and accurate access to information. Question answering system is a research direction that has attracted much attention and has broad development prospects in the field of artificial intelligence and natural language processing. [0003] In the early days of the birth of computers in the 1950s and 1960s, people have researched and attempted question answering systems. Representative ones include Baseball and Lunar, but the early systems were mostly designed for specific domains, with relatively small data scale and weak semantic understanding ability...

AI Technical Summary

Problems solved by technology

However, there are many limitations in such a system: First, the retrieved information is only links to thousands of related files, and the answers may or may not be in these files. There are too many related information, and users cannot quickly and accurately locate the required files. information

Secondly, the retrieval system cannot accurately understand the user's retrieval intention through shallow semantic analysis of the combination of several keywords.

Especially in the face of complex natural language, the retrieval system often leads to missing information, returning wrong information, and returning too much information

These limitations make the effect of the question answering system always unsatisfactory

Images