Synonymous relationship forest construction method based on relationship similarity

The invention relates to a synonym relationship forest construction method based on relationship similarity, and belongs to the field of natural language processing and knowledge maps. The method comprises the steps: inputting all triples, acquiring an entity pair probability set C of relations in a...

Full description

Saved in:
Bibliographic Details
Main Authors ZHANG YINGJIE, FANG YIQIU, YU DEHUA, LI CHUNJIANG, LIN SONGCE, GE NA
Format Patent
LanguageChinese
English
Published 07.08.2020
Subjects
CALCULATING
COMPUTING
COUNTING
ELECTRIC DIGITAL DATA PROCESSING
PHYSICS
Online AccessGet full text

Cover

More Information
Summary:The invention relates to a synonym relationship forest construction method based on relationship similarity, and belongs to the field of natural language processing and knowledge maps. The method comprises the steps: inputting all triples, acquiring an entity pair probability set C of relations in all the triples through a traditional multilayer perceptron, performing a Cartesian operation on theC, calculating a Cartesian product C * C through a formula to obtain a similarity measurement value (matrix S), performing filtering through a threshold value ThresholdA, and screening out a relationsimilarity set; then, in combination with an open domain entity data set, predicting the number of entities corresponding to relationships in open domain entities on an improved multi-layer perceptionmodel by utilizing the trained similarity relationship, and performing filtering by using a threshold value ThresholdB to obtain a Result matrix; and finally, constructing a relationship forest according to the Result matrix a
Bibliography:Application Number: CN202010296825