Neural typing entities in Chinese-pedia
You, Yongjian1; Zhang, Shaohua1; Lou, Jiong1; Zhang, Xinsong1; Jia, Weijia1,2
Conference Name2nd Asia Pacific Web and Web-Age Information Management Joint Conference on Web and Big Data, APWeb-WAIM 2018
Source PublicationLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume10987 LNCS
Conference Date7 23, 2018 - 7 25, 2018
Conference PlaceMacau, China
Author of SourceSpringer Verlag
AbstractTyping entities in structured sources such as Wikipedia has been well studied to construct English knowledge bases automatically. However, there still remain two tough challenges in typing entities in Chinese-pedia. The first one is that structured information from Chinese-pedia cannot assign entities fine-grained types due to its inaccuracy and coarseness. The other challenge is the incompletion of Chinese-pedia, which means we can only use limited attribute fields to type entities. In this paper, we propose a novel Hierarchical Neural System (HNS) to infer fine-grained types for entities in Chinese-pedia. The HNS contains three main models which are hierarchical attention model, feature fusion model and hierarchical classification model. The hierarchical attention model extracts features from entity description based on a bi-LSTM network with hierarchical attention mechanism to break the limitation of inaccurate Chinese-pedia. To deal with the incompletion of Chinese-pedia, the feature fusion model is presented to obtain type features from multi-source such as descriptions, info-boxes, and categories. Through this model, we fuse all the features from different sources together and reduce the features to low-dimensional and dense vectors. Finally, the hierarchical classification model is designed to infer fine-grained types for entities in Chinese-pedia with features obtained from the other two models. The experiments illustrate that HNS outperforms the start-of-art work by 15.6% on f1-score. © Springer International Publishing AG, part of Springer Nature 2018.
Fulltext Access
Citation statistics
Document TypeConference paper
CollectionUniversity of Macau
Affiliation1.Shanghai Jiao Tong University, Shanghai; 200240, China;
2.University of Macau, 999078, China
Recommended Citation
GB/T 7714
You, Yongjian,Zhang, Shaohua,Lou, Jiong,et al. Neural typing entities in Chinese-pedia[C]//Springer Verlag,2018:385-399.
Files in This Item:
There are no files associated with this item.
Related Services
Recommend this item
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[You, Yongjian]'s Articles
[Zhang, Shaohua]'s Articles
[Lou, Jiong]'s Articles
Baidu academic
Similar articles in Baidu academic
[You, Yongjian]'s Articles
[Zhang, Shaohua]'s Articles
[Lou, Jiong]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[You, Yongjian]'s Articles
[Zhang, Shaohua]'s Articles
[Lou, Jiong]'s Articles
Terms of Use
No data!
Social Bookmark/Share
All comments (0)
No comment.

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.