CASIA OpenIR  > 模式识别国家重点实验室  > 自然语言处理
Question Answering over Knowledge Bases
Liu Kang; Zhao Jun; He Shizhu; Zhang Yuanzhe
AbstractThe key of such translation is to understand the meaning of the question. The dominant methods usually convert a natural language question into a complete and formal meaning representation (FMR) first, such as logical form. Based on FMR, the structured query is then smoothly generated. However, completing this aim isn’t trivial. Four questions should be addressed: • How do we represent the meaning of questions grounded to a specific KB? This meaning representation should reflect the corresponding concepts in the KB and organize them according to their semantic relations in the question. The representation Deep Web search is on the cusp of a profound change, from simple document retrieval to natural language question answering (QA).1 Ultimately, search needs to precisely understand the meanings of users’ natural language questions, extract useful facts from all information on the Web, and select Natural L a n guage Processi n g september/october 2015 27 should be unambiguous and adapted to the machine for automated inference and processing. • How do we convert natural language questions to the predefined formal meanings? Text is usually ambiguous, which makes this conversion very difficult. Different formal meanings can be generated from the same text under different contexts, so efficient disambiguation models or rules should be constructed. • How do we scale the QA to a largescale, open domain KB? With the increase of Web data, the sizes of KBs inflate quickly. A KB could contain billions of entities, millions of relations, and thousands of domains, which makes the textual ambiguity more serious and also makes traditional rule-based or labeled datadependent methods infeasible. • How do we answer questions over multiple, interlinked KBs? In many scenarios, answering a question requires synthesizing multiple KBs. The contents from different KBs can make up each other, although their structures are heterogeneous. Many unknown alignments exist among entities, classes, and relations from various KBs. However, discovering such alignments and performing question conversion over multiple heterogeneous KBs isn’t easy. This article gives a rough sketch of question answering over KBs, looking at traditional approaches and attempting to answer these four questions.
KeywordQuestion Answering Knowledge Bases
WOS HeadingsScience & Technology ; Technology
Indexed BySCI
WOS Research AreaComputer Science ; Engineering
WOS SubjectComputer Science, Artificial Intelligence ; Engineering, Electrical & Electronic
WOS IDWOS:000361315900004
Citation statistics
Cited Times:3[WOS]   [WOS Record]     [Related Records in WOS]
Document Type期刊论文
Corresponding AuthorLiu Kang
AffiliationChinese Acad Sci, Inst Automat, Natl Lab Pattern Recognit, Beijing 100864, Peoples R China
Recommended Citation
GB/T 7714
Liu Kang,Zhao Jun,He Shizhu,et al. Question Answering over Knowledge Bases[J]. IEEE INTELLIGENT SYSTEMS,2015,30(5):26-35.
APA Liu Kang,Zhao Jun,He Shizhu,&Zhang Yuanzhe.(2015).Question Answering over Knowledge Bases.IEEE INTELLIGENT SYSTEMS,30(5),26-35.
MLA Liu Kang,et al."Question Answering over Knowledge Bases".IEEE INTELLIGENT SYSTEMS 30.5(2015):26-35.
Files in This Item: Download All
File Name/Size DocType Version Access License
Question Answering o(1292KB)期刊论文作者接受稿开放获取CC BY-NC-SAView Download
Related Services
Recommend this item
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[Liu Kang]'s Articles
[Zhao Jun]'s Articles
[He Shizhu]'s Articles
Baidu academic
Similar articles in Baidu academic
[Liu Kang]'s Articles
[Zhao Jun]'s Articles
[He Shizhu]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[Liu Kang]'s Articles
[Zhao Jun]'s Articles
[He Shizhu]'s Articles
Terms of Use
No data!
Social Bookmark/Share
File name: Question Answering over Knowledge Base.pdf
Format: Adobe PDF
All comments (0)
No comment.

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