Cooperative disambiguation method based on deep semantic neighbor and multivariate entity association

Abstract

The invention provides a collaborative disambiguation method based on deep semantic neighbor and multivariate entity association. The collaborative disambiguation method comprises the following steps: generating an entity reference set and a candidate entity set; obtaining vector representations of candidate entities and entity references; constructing character string matching degree and context similarity local features between the entity reference and the candidate entity; extracting local consistency characteristics among the entity references to obtain adjacent references; constructing an initial entity semantic association graph based on the candidate entity sets of the local similarity features, the entity references and the adjacent references; dividing the whole entity reference set into a low-ambiguity part and a high-ambiguity part, and enriching and updating the entity semantic correlation graph based on the low-ambiguity part and the high-ambiguity part; and aggregating the local similarity features and the global features through a disambiguation model based on a graph attention network, and outputting a mapping entity corresponding to each entity reference. According to the method, the problems of high collaborative disambiguation calculation cost and local consistency of entities in a document-level entity disambiguation task can be solved.

Description

technical field [0001] The invention relates to the technical field of computer natural language processing, in particular to a collaborative disambiguation method based on deep semantic neighbor and multi-entity association. Background technique [0002] Entity Disambiguation (ED for short) is a key technology involved in many computer natural language processing tasks such as knowledge graph construction, information extraction, and knowledge question answering. Generally, the application of entity disambiguation in text is also called entity linking, which is used to accurately map entity mentions identified in unstructured text to specific entity entries in a specified knowledge base. The recognition of entity references in unstructured text is generally completed through the task of named entity recognition (NER); specifically, the task of entity disambiguation is divided into three modules: candidate entity generation, disambiguation and unlinkable prediction. However...

AI Technical Summary

Problems solved by technology

[0004] Aiming at the deficiencies in the existing technology, the present invention proposes a collaborative disambiguation method based on deep semantic neighbors and multi-entity associations to solve the document-level entity disambiguation task existing in the prior art, and there is a collaborative disambiguation algorithm using When calculating the entity semantic association graph, there are technical problems such as high calculation cost, weak anti-interference ability, and poor ability to identify entity references with high ambiguity

Images