Proceedings of the Workshop on Natural Language Processing in the 5th Information Systems Research Working Days (JISIC)

Abstract

Since, at the moment there is not a goldstandard annotated corpus for this objective, it is necessary to build one, to allow generation and testing of automatic systems for classifying the purpose or function of a citation referenced in an article.The development of this kind of corpus is subject to two conditions: the first one is to present a clear and unambiguous classification scheme.The second one is to secure an initial manual process of labeling to reach a sufficient inter-coder agreement among annotators to validate the annotation scheme and to be able to reproduce it even with coders who do not know in depth the topic of the analyzed articles.This paper proposes and validates a methodology for corpus annotation for citation classification in scientific literature that facilitate annotation and produces substantial interannotator agreement.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 2 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.