The NTNU System at SemEval-2017 Task 10: Extracting Keyphrases and Relations from Scientific Publications Using Multiple Conditional Random Fields

Proceedings of the 11th International Workshop on Semantic Evaluations (SemEval-2017), pages 951–955, Vancouver, Canada, August 3 - 4, 2017. c⃝2017 Association for Computational Linguistics The NTNU System at SemEval-2017 Task 10: Extracting Keyphrases and Relations from Scientific Publications Using Multiple Conditional Random Fields Lung-Hao Lee1, Kuei-Ching Lee2, Yuen-Hsien Tseng1 1Graduate Institute of Library and Information Studies, National Taiwan Normal University 2China Development Lab, IBM 1No. 162, Sec. 1, Heping East Road, Taipei 10610, Taiwan 2No. 13, Sanchong Road, Taipei 11501, Taiwan lhlee@ntnu.edu.tw, jklee@tw.ibm.com, samtseng@ntnu.edu.tw Abstract This study describes the design of the NTNU system for the ScienceIE task at the SemEval 2017 workshop. We use self-defined feature templates and multi- ple conditional random fields with ex- tracted features to identify keyphrases along with categorized labels and their relations from scientific publications. A total of 16 teams participated in evalua- tion scenario 1 (subtasks A, B, and C), with only 7 teams competing in all sub- tasks. Our best micro-averaging F1 across the three subtasks is 0.23, ranking in the middle among all 16 submissions. 1 Introduction Keyphrases are usually regarded as phrases that capture the main topics mentioned in a given text. Automatically extracting keyphrases and deter- mining their relations from scientific articles has various applications, such as recommending arti- cles to readers, matching reviewers to submis- sions, facilitating the exploration of huge docu- ment collections, and so on. An adapted nominal group chunker and a supervised ranking method based on support vector machines have previous- ly been used to extract keyphrase candidates (Eichler and Neumann, 2010). The conditional random field based keyphrase extraction method has been presented (Bhaskar et al., 2012). A na- ïve approach has been proposed to investigate characteristics of keyphrases with section infor- mation from well-structured scientific articles (Park et al., 2010). Features broadly used for the supervised approaches in scientific articles have been assessed in the compilation of a compre- hensive feature list (Kim and Kan, 2009). Maxi- mal sequences and page ranking have been com- bined to discover latent keyphrases within scien- tific articles (Ortiz et al., 2010). Noun phrases containing multiple modifiers have been extract- ed from earth science publications and general- ized by matching tree patterns to the syntax trees of the sources texts (Marsi and Öztürk, 2015). Keyphrase boundary classification has been re- garded as a multi-task learning problem using deep recurrent neural network (Augenstein and Søgaard, 2017). The ScienceIE task seeks solutions to auto- matically identify keyphrases within scientific publications, label them, and determine their re- lationships. Specifically, the ScienceIE task con- tains three subtasks: (A) Identification of keyphrases: to identify all the keyphrases within a given scientific publication; (B) Classification of identified keyphrases: to label each keyphrase as Process, Task, or Material; (C) Extraction of relationships between two identified keyphrases: to label keyphrases as Hyponym-of or Synonym- of. The ScienceIE task presents three evaluation scenarios. In Scenario 1, only plain text is given for subtasks A, B, and C; in Scenario 2, plain text with manually annotated keyphrase boundaries …

Content truncated to 500 words