Leveraging document-specific information for classifying relations in scientific articles

Qin Dai, Naoya Inoue, Paul Reisert, Kentaro Inui

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review


Tremendous amount of knowledge is present in the ever-growing scientific literature. In order to grasp this massive amount knowledge, various computational tasks are proposed for training computers to read and analyze scientific documents. As one of these task, semantic relationship classification aims at automatically analyzing semantic relationships in scientific documents. Conventionally, only a limited number of commonly used knowledge bases such as Wikipedia are used for collecting background information for this task. In this work, we hypothesize that scientific papers also could be utilized as a source of background information for semantic relationship classification. Based on the hypothesis, we propose the model that is capable of extracting background information from unannotated scientific papers. Preliminary experiments on the RANIS dataset [1] proves the effectiveness of the proposed model on relationship classification in scientific articles.

Original languageEnglish
Title of host publicationNew Frontiers in Artificial Intelligence - JSAI-isAI Workshops, JURISIN, SKL, AI-Biz, LENLS, AAA, SCIDOCA, kNeXI, Revised Selected Papers
EditorsKoji Mineshima, Kazuhiro Kojima, Ken Satoh, Sachiyo Arai, Daisuke Bekki, Yuiko Ohta
PublisherSpringer Verlag
Number of pages16
ISBN (Print)9783319937939
Publication statusPublished - 2018
Event9th JSAI International Symposium on Artificial Intelligence, JSAI-isAI 2017 - Tsukuba, Japan
Duration: 2017 Nov 132017 Nov 15

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume10838 LNAI
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349


Conference9th JSAI International Symposium on Artificial Intelligence, JSAI-isAI 2017


  • Lexical chain
  • Scientific document
  • Semantic relationship


Dive into the research topics of 'Leveraging document-specific information for classifying relations in scientific articles'. Together they form a unique fingerprint.

Cite this