An entity multi-classification method in combination with attribute information

Abstract

The invention discloses an entity multi-classification method combined with attribute information. The aim of the entity classification task is to endow an entity with an appropriate class label in combination with context information of the entity. and in the entity multi-classification task, only one category label of the entity is provided, the category label probability is predicted through the classification model, and the label with the maximum probability is used as the category of the entity. The method is based on a traditional entity classification method, attribute information of aknowledge base entity is combined, a deep learning method is used for conducting feature representation on entities, attributes and contexts, the entity representation comprises two features of vocabularies and character levels, and an attention mechanism of entity perception is introduced into the context feature representation. In addition, through joint training of entity and attribute characteristics, introduction of attribute information in the prediction stage is avoided. And finally, entity multi-classification is realized by comprehensively utilizing entities and context characteristics. According to the method, the attribute information is introduced to serve as additional features, and the entity multi-classification effect is improved.

Description

technical field [0001] The invention relates to entity multi-classification technology, in particular to an entity multi-classification method combined with attribute information. Background technique [0002] The purpose of the entity classification task is to assign an appropriate category label to an entity combined with its context information. In the entity multi-classification task, there is only one category label of the entity, and the category label probability is predicted through the classification model, and the label with the highest probability is used as the category of the entity. The category information of the entity can enhance the background information of the entity in the text, which is helpful for many natural language processing tasks, such as question answering and reading comprehension, knowledge base construction, entity linking and relationship extraction, etc. [0003] Traditional entity multi-classification methods use manual features, rely on ...

AI Technical Summary

Problems solved by technology

[0004] The present invention utilizes the text information of the entity and the attribute list information in the knowledge base, overcomes the defect that the existing method ignores the knowledge base information and the insufficient context information of the entity, improves the effect of multi-classification of entities, and provides a multi-classification of entities combined with attribute information method

Images