Knowledge Extraction and Consolidation for Scientific Publications in the Educational Domain

Motivation

Digital Libraries have become an essential tool in research because they provide access to an enormous amount of information through distributed networks and from multiple locations. Today, the number of publications increases heavily which leads to a large effort for researchers to find all relevant information on a topic. Therefore we want develop methods and tools which are able to automatically extract and consolidate knowledge from scientific publications in the exemplar domain of educational research.

 

Goals

This project consists of two parts: Firstly methods for extracting arguments from scientific publications will be developed. Arguments are a common instrument in such papers to give reasons for statements. Usually they are made up of a conclusion and one or more premises substantiating the conclusion.

 

Secondly we will explore methods to calculate relations between the extracted arguments. As a baseline we will use semantic similarity measures to find arguments which address a similar topic and to build up an argument graph containing all arguments and their relations. Based on this, we will explore textual entailment methods to get further relation types such as argument support or contradiction.

 

Finally it should be possible to search the extracted arguments for a given topic and to retrieve related arguments for a given argument.

 

Methods

In the first part of the project, we will explore the structure of arguments in scientific publications in the educational domain. Concerning the extraction of arguments, we will build on the work of Palau and Moens (2009) who use different features like key words or sentence length to identify arguments from discussion fora, legal judgments, newspapers and further data sources. The analysis of the argument structure for publications in the educational domain will most likely reveal differences compared to the data sources used by Palau and Moens (2009) and should help to find suitable features for our domain.

 

In the second part of the project, we first will use the DKPro Similarity Framework (Bär et. al., 2013) to calculate relations (text similarity) between the extracted arguments and construct an argument graph. Given this, we will try to improve the graph by exploring textual entailment techniques (Stern and Dagan, 2011). A suitable framework for this purpose is the BIUTEE framework (Stern and Dagan, 2012).

 

For the evaluation, we will construct a gold standard with a limited number of arguments, manually extracted from scientific publications. Furthermore, we will construct an argument graph with relations between the arguments.

 

References

  • R. M. Palau and M.-F. Moens (2009). Argumentation mining: the detection, classification and structure of arguments in text. In Proceedings of the 12th International Conference on Artificial Intelligence and Law, ICAIL ’09, pp. 98–107. New York: ACM.
  • Daniel Bär, Torsten Zesch, and Iryna Gurevych (2013). DKPro Similarity: An Open Source Framework for Text Similarity. In Proceedings of the 51st Annual Meeting of the Association for Computational Linguistics. System Demonstrations, August 2013.

  • Asher Stern and Ido Dagan (2011). A Confidence Model for Syntactically-Motivated Entailment Proofs. In Proceedings of the International Conference on Recent Advances in Natural Language Processing, pages 455–462.

  • Asher Stern and Ido Dagan (2012). BIUTEE: A Modular Open-Source System for Recognizing Textual Entailment. In Proceedings of the ACL 2012 System Demonstrations, pages 7378, Jeju Island, Korea, July. Association for Computational Linguistics.

Publications

Displaying results 1 to 3 out of 3

Linking the Thoughts: Analysis of Argumentation Structures in Scientific Publications
Christian Kirschner,Judith Eckle-Kohler,Iryna Gurevych
In: Proceedings of the 2nd Workshop on Argumentation Mining held in conjunction with the 2015 Conference of the North American Chapter of the Association for Computational Linguistics – Human Language Technologies (NAACL HLT 2015), p. 1-11, June 2015.
https://www.ukp.tu-darmstadt.de/data/argumentation-mining/argument-annotated-scientific-articles/.

Knowledge Discovery in Scientific Literature
Jinseok Nam,Christian Kirschner,Zheng Ma,Nicolai Erbs,Susanne Neumann,Daniela Oelke,Steffen Remus,Chris Biemann,Judith Eckle-Kohler,Johannes Fürnkranz,Iryna Gurevych,Marc Rittberger,Karsten Weihe
In: Proceedings of the 12th Konferenz zur Verarbeitung natürlicher Sprache (KONVENS 2014), p. 66-76, October 2014. ISBN 978-3-934105-46-1.

Argumentation Mining in Persuasive Essays and Scientific Articles from the Discourse Structure Perspective
Christian Stab,Christian Kirschner,Judith Eckle-Kohler,Iryna Gurevych
In: Elena Cabrio and Serena Villata and Adam Wyner: Proceedings of the Workshop on Frontiers and Connections between Argumentation Theory and Natural Language Processing, p. 40--49, CEUR-WS, July 2014.
http://ceur-ws.org/Vol-1341/.

A A A | Drucken Print | Impressum Impressum | Sitemap Sitemap | Suche Search | Kontakt Contact | Webseitenanalyse: Mehr Informationen
zum Seitenanfangzum Seitenanfang