Towards a language-independent solution: Knowledge base completion by searching the Web and deriving language pattern

详细信息    查看全文

• 作者：Lidong Bing ; ^a ; ^d ; ^{lyndonbing@tencent.com" class="auth_mail" title="E-mail the corresponding author} ; ^{binglidong@gmail.com" class="auth_mail" title="E-mail the corresponding author} ; Zhiming Zhang^b ; ^{zhangzhiming@baidu.com" class="auth_mail" title="E-mail the corresponding author} ; Wai Lam^c ; ^{wlam@se.cuhk.edu.hk" class="auth_mail" title="E-mail the corresponding author} ; William W. Cohen^d ; ^{wcohen@cs.cmu.edu" class="auth_mail" title="E-mail the corresponding author}
• 关键词：Knowledge base completion ; Language pattern ; Language-independent solution
• 刊名：Knowledge-Based Systems
• 出版年：2017
• 出版时间：1 January 2017
• 年：2017
• 卷：115
• 期：Complete
• 页码：80-86
• 全文大小：1567 K

文摘

Knowledge bases (KBs) such as Freebase and Yago are rather incomplete, and the situation is more serious in non-English KBs, such as Chinese KBs. In this paper, we present a language-independent framework to tackle the slot-filling task by searching the Web with high-precision queries, and deriving lightweight extraction patterns. The patterns are based on string matching, and since they make no use of complex NLP resources, which may be unavailable in some languages, they are very language-independent.

We use a traditional bootstrapping approach for extraction, but also use a novel approach to suppress the noise associated with distant supervision: in particular, we use a pseudo-testing method to validate the patterns derived from different sentences. Experiments show that our framework achieves very encouraging results.