A Method of Entity Linking Based on Graph Model

Abstract

The invention discloses a graph model-based entity linking method, which is mainly used for dealing with the entity linking problem of unstructured text. The present invention first constructs an entity association graph of the text from all entity reference items and corresponding candidate entity sets obtained in the same text, as the basis of the dynamic entity disambiguation algorithm. Then, using the dynamic entity disambiguation algorithm based on graph and PageRank, the undisambiguated candidate entity with the highest score is selected in each round as the target entity of the entity reference, and the disambiguation selection process of the entity reference corresponding to multiple candidate entities is gradually completed. Finally, use XGBoost in the field of machine learning to judge the target entity referred to by the entity, correctly link the registered target entities in the knowledge base, and correctly identify the unregistered target entities in the knowledge base.

Description

technical field [0001] The invention belongs to the field of entity linking and relates to an entity linking method based on a graph model. Background technique [0002] Since the concept of the Semantic Web was proposed, more and more open link data and user-generated content have been published on the Internet, and the Internet has gradually changed from a document that only contains hyperlinks between web pages to a World Wide Web that contains a large number of descriptions of various entities and The World Wide Web of Data rich in relationships between entities. While Internet webpages, such as news, blogs, etc., involve a large number of entities, most of the webpages themselves do not have relevant descriptions and background introductions about these entities. In order to help people better understand the content of the webpage, many websites or authors will establish a link relationship between the entities that appear in the webpage and the corresponding knowledge...

AI Technical Summary

Problems solved by technology

[0006] Although there have been a lot of researches on entity linking and graph-based entity disambiguation methods, the common problem of existing methods is that they do not give full play to the role of unambiguous entities, and the semantic information in entity association graphs is not enriched with the addition of unambiguous entities

Images