Multi-Hop Reasoning for Question Answering with Knowledge Graph

Multi-hop Question Answering over Knowledge Graphs (KGQA) in previous studies has achieved remarkable results by exploiting the prediction property of Knowledge Graphs (KG) embedding. However, when facing Chinese sentences, its answer selection performs poorly. We improve the method for KGQA by comb...

Full description

Saved in:
Bibliographic Details
Published in2021 IEEE/ACIS 19th International Conference on Computer and Information Science (ICIS) pp. 121 - 125
Main Authors Zhang, Jiayuan, Cai, Yifei, Zhang, Qian, Cao, Zehao, Cheng, Zhenrong, Li, Dongmei, Meng, Xianghao
Format Conference Proceeding
LanguageEnglish
Published IEEE 23.06.2021
Subjects
CNN
Cognition
Computational modeling
Information science
Knowledge discovery
Knowledge Graph
Lattices
Optimization
Question Answering System
Online AccessGet full text

Cover

More Information
Summary:Multi-hop Question Answering over Knowledge Graphs (KGQA) in previous studies has achieved remarkable results by exploiting the prediction property of Knowledge Graphs (KG) embedding. However, when facing Chinese sentences, its answer selection performs poorly. We improve the method for KGQA by combining the traditional method for KGQA with a lattice based CNN (LCN) model. We refine the granularity of questions and answers to make its coverage more extensive and generalizable, and expand the answer set to improve the performance in single results.
DOI:10.1109/ICIS51600.2021.9516865