Get our free extension to see links to code for papers anywhere online!

Chrome logo Add to Chrome

Firefox logo Add to Firefox

Crowdsourcing Semantic Label Propagation in Relation Classification

Sep 03, 2018
Anca Dumitrache, Lora Aroyo, Chris Welty

Share this with someone who'll enjoy it:

Distant supervision is a popular method for performing relation extraction from text that is known to produce noisy labels. Most progress in relation extraction and classification has been made with crowdsourced corrections to distant-supervised labels, and there is evidence that indicates still more would be better. In this paper, we explore the problem of propagating human annotation signals gathered for open-domain relation classification through the CrowdTruth methodology for crowdsourcing, that captures ambiguity in annotations by measuring inter-annotator disagreement. Our approach propagates annotations to sentences that are similar in a low dimensional embedding space, expanding the number of labels by two orders of magnitude. Our experiments show significant improvement in a sentence-level multi-class relation classifier.

* In publication at the First Workshop on Fact Extraction and Verification (FeVer) at EMNLP 2018 

   Access Paper Source

Share this with someone who'll enjoy it: