Knowledge graph representation learning method friendly to long text

The invention discloses a knowledge graph representation learning method friendly to a long text. The knowledge graph representation learning method comprises the following steps: 1, extracting text information friendly to the long text; 2, establishing a text representation model of the BCRL; 3, ca...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: CUI KAIQIAN, WU WENFANG, LI XUEYU, LI LEILEI, HAN DONGHONG, WU GANG
Format: Patent
Sprache:chi ; eng
Schlagworte:
Online-Zugang:Volltext bestellen
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:The invention discloses a knowledge graph representation learning method friendly to a long text. The knowledge graph representation learning method comprises the following steps: 1, extracting text information friendly to the long text; 2, establishing a text representation model of the BCRL; 3, carrying out structured expression based on TransE; 4, carrying out structure-text joint identification; 5, carrying out model training. The invention provides a text-enhanced knowledge graph representation model BCRL, the model utilizes entity description and relation mention to enhance knowledge representation of a triple, the method solves the problems of non-uniform entity description, inaccurate relation mention representation and the like from the perspective of text-sentence, semantic information of a text can be captured more effectively, and the semantic information of the text can be extracted more effectively; and compared with a reference system, the method is remarkably improved in the aspect of a link p