Link to the University of Pittsburgh Homepage
Link to the University Library System Homepage Link to the Contact Us Form

The Diagnosticity of Argument Diagrams

Lynch, Collin (2014) The Diagnosticity of Argument Diagrams. Doctoral Dissertation, University of Pittsburgh.

[img]
Preview
PDF
Primary Text

Download (9MB) | Preview

Abstract

Can argument diagrams be used to diagnose and predict argument performance?

Argumentation is a complex domain with robust and often contradictory theories about the structure and scope of valid arguments. Argumentation is central to advanced problem solving in many domains and is a core feature of day-to-day discourse. Argumentation is quite literally, all around us, and yet is rarely taught explicitly. Novices often have difficulty parsing and constructing arguments particularly in written and verbal form. Such formats obscure key argumentative moves and often mask the strengths and weaknesses of the argument structure with complicated phrasing or simple sophistry. Argument diagrams have a long history in the philosophy of argument and have been seen increased application as instructional tools. Argument diagrams reify important argument structures, avoid the serial limitations of text, and are amenable to automatic processing.

This thesis addresses the question posed above. In it I show that diagrammatic models of argument can be used to predict students' essay grades and that automatically-induced models can be competitive with human grades. In the course of this analysis I survey analytical tools such as Augmented Graph Grammars that can be applied to formalize argument analysis, and detail a novel Augmented Graph Grammar formalism and implementation used in the study. I also introduce novel machine learning algorithms for regression and tolerance reduction. This work makes contributions to research on Education, Intelligent Tutoring Systems, Machine Learning, Educational Datamining, Graph Analysis, and online grading.


Share

Citation/Export:
Social Networking:
Share |

Details

Item Type: University of Pittsburgh ETD
Status: Published
Creators/Authors:
CreatorsEmailPitt UsernameORCID
Lynch, Collincollinl@cs.pitt.eduCOLLINL
ETD Committee:
TitleMemberEmail AddressPitt UsernameORCID
Committee ChairAshley, Kevinashley@pitt.eduASHLEY
Committee MemberAleven, Vincentaleven@cs.cmu.edu
Committee MemberLitman, Dianelitman@cs.pitt.eduDLITMAN
Committee MemberSchunn, Chrisschunn@pitt.eduSCHUNN
Date: 29 May 2014
Date Type: Publication
Defense Date: 30 January 2014
Approval Date: 29 May 2014
Submission Date: 10 March 2014
Access Restriction: No restriction; Release the ETD for access worldwide immediately.
Number of Pages: 291
Institution: University of Pittsburgh
Schools and Programs: Dietrich School of Arts and Sciences > Intelligent Systems
Degree: PhD - Doctor of Philosophy
Thesis Type: Doctoral Dissertation
Refereed: Yes
Uncontrolled Keywords: Argumentation, Essay Writing, Argument Diagrams, Graph Analysis, Machine Learning, Ill-Defined Domains, Intelligent Tutoring Systems, Educational Datamining, Multiple Representations
Date Deposited: 29 May 2014 19:40
Last Modified: 15 Nov 2016 14:17
URI: http://d-scholarship.pitt.edu/id/eprint/20710

Metrics

Monthly Views for the past 3 years

Plum Analytics


Actions (login required)

View Item View Item