[Adapt] [Seminar] Inferring Binary Relation Schemas for Open Information Extraction

Luo Kangqi luo.kangqi at qq.com
Tue Apr 21 22:54:34 CST 2015


Hi All,


Our next seminar will take place tomorrow at 5pm in room 3-528. This time I'll give you a talk about the paper "Inferring Binary Relation Schemas for Open Information Extraction", which is demo paper submitted to ACL 2015.


Open Information Extraction (IE) system extracts huge number of binary relations from web corpus, forming <arg1, relation, arg2> triples. For one relation, we care about the preferred types pairs representing these two arguments. The preferred type pairs are called relation schemas, which is an important information in natural language understanding, especially in the task of entity disambiguation and question answering. In this paper, we present a system to model the schemas of binary relations in ReVerb system, collecting schemas for more than 150,000 binary relations, and reaches a high accuracy. Hope you can enjoy this talk.


Thanks,
Kangqi
------------------
Luo Kangqi
ADAPT Lab, SEIEE 3-341
Shanghai Jiao Tong University
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://cs.sjtu.edu.cn/pipermail/adapt/attachments/20150421/d56b07bb/attachment.html>


More information about the Adapt mailing list