Citation-reason Annotation Using Crowdsourcing

Abstract

We have proposed an idea to predict citation-reasons between scientific papers with machine-learning techniques, and try to narrow down the search range for relevant papers based on the citation-reasons. However, the machine-learning method seems not accurate enough according to a subject experiment. In this paper, as a substitution of the machine-learning method, we have proposed a strategy to annotate citation-reasons between papers in a crowdsourcing manner. An analysis on the result has shown the effectiveness of our strategy and some future tasks.