Entity linking method based on RoBERTa and heuristic algorithm

Abstract

The invention discloses an entity linking method based on RoBERTa and a heuristic algorithm. The method comprises the following steps that a sequence labeling model based on a pre-trained language model RoBERTa is used for labeling a problem, and the RoBERTa model stacks 12 layers of transcriber structures to obtain multi-level grammatical semantic information in the problem; dynamic representation of each word in the question based on context is obtained through a multi-head attention mechanism in transformator, and then the entity mention range in the question is obtained; and after the entity mention range is obtained, directly matching the entity mention with the knowledge base entity by using a heuristic algorithm to complete entity linking. The method can be applied to various knowledge base questioning and answering scenes, and provides underlying basic services for many advanced applications.

Description

technical field [0001] The invention relates to the field of knowledge base questions and answers, in particular to an entity linking method based on RoBERTa and a heuristic algorithm. Background technique [0002] Entity linking is a subtask of knowledge base question answering, which aims to extract knowledge base entity texts that appear in questions and link them to entity objects in knowledge base, so as to obtain candidate answers for knowledge base question answering tasks. In recent years, with the continuous development of large-scale knowledge bases such as YAGO, Freebase, and Dbpedia, knowledge base question answering tasks have also begun to attract people's attention, and how to establish a connection between natural language questions and structured knowledge base entities is A burning problem that is also the goal of entity linking tasks. [0003] For this task, the early research mainly used the n-gram method to traverse the phrases in the question, and then...

AI Technical Summary

Problems solved by technology

This method has the following disadvantages: (1) There are many n-gram word combinations in the question, resulting in too much time-consuming query knowledge base; (2) Too many irrelevant entities are introduced in the candidate entities

Most of the current entity linking models based on sequence annotations model the problem on the basis of word-embedding, which also limits the performance of current entity linking methods to a certain extent.

Images