SALTS Lab

Argument Mining

A Response Centered Approach

Current members:

Faculty: Mark Aakhus, Smaranda Muresan, Nina Wacholder

Current Post-Doctoral Researchers: Elena Musi

Current PhD Students: Debanjan Ghosh, Shawon Sarkar

Prior students: Soon-Yeon Hwang, Joan Keegan, Matthew Mitsui, Alexander Pichugin

Overview

Differences inevitably arise in any human activity and the management of these differences is consequential for what can be achieved and known through human activity. This project develops a response-centered approach to argument mining that seeks to build methods for the computational modeling of the argumentative uses of language and reason in managing differences. A response-centered approach to argument mining contributes to explaining the of role argumentation in collective intelligence and advancing principles and uses of advanced computing and communication and information systems.

Response-centered methods are capable of identifying and characterizing the practices that shape the ebb and flow of argumentation in complex discussions and controversies in groups, communities, and societies. The computational models for argumentation mining, along with the theory behind them, promise a means to support systems that facilitate complex deliberations and mediate disagreements in large-scale collaborations, thereby providing valuable tools in an increasingly interconnected, citizen-participating social and political environment. Response-centered models will inform the design and development of argumentation analytics and support tools for interaction and reasoning in decision making, problem solving, and conflict management.

The project involves three main lines of research activity: 

NLP Models for Argumentation Mining Grounded in Argumentation Theory. This project offers a key reframing of prior natural language processing research related to argumentation mining by investigating novel computational models for a response-centered approach to argumentation mining that takes into account lexical, sentential, and discourse semantics. 

Ontology of Argumentation. This research develops an extensible response-centered ontology of argumentation that is sensitive to the interactive, polylogical nature of argumentation, that affords detection of pattern and variation in argumentative practices, and that scales to the variety of contexts where argumentation happens.

Efficient and Reliable Annotation. A key resource needed to empirically validate the proposed ontology and to develop computational models for argumentation mining grounded in this theoretical model of argumentation is an annotated corpus of online polylogues using the ontology as an annotation scheme.

Publications

Aakhus, M., Muresan, S., & Wacholder, N. (2016). An Argument-Ontology for a Response-Centered Approach to Argumentation Mining. In F. Bex, F. Grass, & N. Green (Eds.), The 16th Workshop on Computational Models of Natural Argument. New York: CMNA. [pdf]

Musi, E., Ghosh, D. & Muresan, S. (2016). Towards Feasible Guidelines for the Annotation of Argument Schemes. ACL 2016, 82. Chicago. [pdf]

Wacholder, N., Muresan, S., Ghosh, D., & Aakhus, M. (2014).  “Annotating Multiparty Discourse: Challenges for Agreement Metrics”. Proceedings of the 8th Linguistic Annotation Workshop. Dublin, Ireland: COLING, Aug. 24-25, 2014, pp. 120-128. [pdf] [appendix] [Annotation Guidelines] [ppt]

Ghosh, Debanjan,  Muresan,Smaranda, Wacholder, Nina, Aakhus, Mark & Mitsui, Matthew. (2014).“Analyzing Argumentative Discourse Units in Online Interactions”. In: Proceedings of the First Workshop on Argumentation Mining. Baltimore, Maryland: Association for Computational Linguistics, 2014, pp. 39–48. [pdf]  [ppt]

Aakhus, Mark, Muresan, Smaranda. & Wacholder, Nina. (2013). Integrating natural language processing and pragmatic argumentation theories for argumentation support. In D. Mohammed & M. Lewiński (Eds.), Virtues of Argumentation. Proceedings of the 10th International Conference of the Ontario Society for the Study of Argumentation (pp. 1-13). Windsor, ON: OSSA. [pdf]

Resources