Project GenIE
Principal Investigator: Dr. Nancy Green, Dept. of Computer
Science, University of North Carolina at Greensboro
Sponsor: National Science Foundation (Start Date: June 1, 2002;
End Date with no-cost extension: May 31, 2010)
Award Title: CAREER: Intelligent Argument Generation in Text and Information
Graphics (Award Number 0132821).
Summary: Lay people are faced with increasing responsibility for
making vital decisions based on technical arguments, often involving quantitative
and statistical data. The goal of this type of argument is to explain
the reasoning and evidence leading to some conclusion, so that others
may evaluate its degree of plausibility and certainty. However,
lay people may have trouble understanding and evaluating arguments based
upon probability and statistics. Furthermore, for conciseness, supporting
data is often presented graphically, e.g. in line graphs and bar charts.
However, it is debatable whether people without technical training are
able to take full advantage of data graphics without additional assistance.
To complicate this problem, the data underlying an argument may change as
new information becomes available, requiring a new argument to be communicated.
This research will address these problems by developing techniques for
implementing computer systems to communicate technical arguments.
The arguments will be communicated using integrated text and data graphics.
Although the research is applicable to many different fields, for testing
purposes this project will implement the techniques in a prototype system
for use by genetics counselors. In current practice, a counselor meets
with a client to perform a risk analysis for genetic disorders and to
discuss the results; afterwards, the counselor writes a brief summary letter
about the client's case for the client to refer to later. The prototype
system will be designed to automatically produce a hypertext summary presentation
based upon information provided by the counselor, but expanded by the
system, e.g., to include references to relevant scientific data and methods.
Using a combination of text and graphics, the hypertext presentation
will be designed for a non-technical audience.
The project will integrate empirical methods from computational linguistics
and human-computer interaction. Empirical methods are necessary
to determine appropriate techniques for conveying technical arguments to
people with non-technical backgrounds. Initially, the project will
developanalytical models based upon a corpus of human-authored multimedia
presentations. For example, corpus studies might determine which features
of arguments suggest where supporting graphics should be placed. Next,
human-computer interaction experiments will be used to determine the effectiveness
of techniques based upon the analytical models; and to investigate alternatives
to the techniques that are shown not to be sufficient. The resulting
empirically-based techniques will be incorporated into computational models.
The computational models will extend current approaches in natural language
generation, automated graphic design, and intelligent multimedia presentation
systems. Although the computational models will be demonstrated
within a prototype system for genetics counseling patient education,
the models could be targeted for other technical domains and uses.
Other Funding: UNCG New Faculty Grant (Spring 2000), UNCG Summer
Excellence Research Grants (2000, 2001), UNCG Undergraduate Research Assistantships
(Fall 2002- Spring 2003, Fall 2004), DOE Graduate Research Assistantship
(Fall 2002)
GenIE involves research in all of these fields of computer science:
Selected publications and presentations on GenIE:
- Coding Manual
for corpus annotation.
- Nancy Green. An Empirical Study of Multimedia Argumentation.
Proceedings of the International Conference on Computational Systems,
Workshop on Computational Models of Natural Language Arguments (CMNA
2001), May 2001. [Available as Springer Lecture Notes in Computer
Science 2073, pp. 1009-18.]
- Nancy Green. Towards an Empirical Model of Argumentation in
Medical Genetics. Workshop Papers: Computer Models of Natural Argumentation
(CMNA 2003), workshop of International Joint Conference on Artificial
Intelligence (IJCAI-03), Aug. 9, 2003, Acapulco, Mexico, p. 39-44.
- Nancy Green. Using HCI Experiments to Validate Intelligent Multimedia
Cue Generation. Proceedings of Florida Artificial Intelligence Research
Symposium, FLAIRS 2004.
- Nancy Green. Analysis of Linguistic Features Associated with
Point of View for Generating Stylistically Appropriate Text. Papers from
the 2004 AAAI Spring Symposium: Exploring Attitude and Affect in
Text: Theories and Applications. AAAI Press. (Extended version in Shanahan,
Qu, and Wiebe (eds), Computing Attitude and Affect in Text: Theory and
Applications. Spring, 2005, 33-40.)
- N. Green, T. Britt, and K. Jirak. Communication of Uncertainty
in Clinical Genetics Patient Health Communication Systems. AAAI 2004
Fall Symposium on Dialogue Systems for Health Communication.
- N. Green. Graphics for Patient-Tailored Information in Clinical
Genetics. Papers from 2005 AAAI Spring Symposium: Reasoning with Mental and
External Diagrams.
- N. Green. A Bayesian Network Coding Scheme for Annotating Biomedical
Information Presented to Genetic Counseling Clients. Journal of Biomedical
Informatics 38 (2005) 130-144.
- N. Green. Affective Factors in Generation of Tailored Genomic
Information. User Modeling 2005, Workshop on Adapting the Interaction
Style to Affective Factors, July 2005, Edinburgh.
- N. Green, T. Britt, K. Jirak, D. Waizenegger, and X. Xin. User
Modeling for Tailored Genomic E-health Information. User Modeling 2005,
Workshop on Personalisation for eHealth, July 2005, Edinburgh.
- N. Green. Design of Information Graphics for Causal Arguments.
Computer Models of Natural Argument (CMNA 2005), July 2005, Edinburgh.
- N. Green. Representing Normative Arguments in Genetic Counseling.
Argumentation for Consumers of Healthcare: Papers from 2006 AAAI Spring
Symposium.
- N. Green. Generation of Biomedical Arguments for Lay Readers. Proc.
of Int. Conf. on Natural Language Generation (INLG 2006).
- N. Green. A Study of Argumentation in a Causal Probabilistic Domain:
Genetic Counseling. Int. J. of Intelligent Systems 22(1), Jan. 2007,
71-93.
- N. Green. Dialectical Argumentation in Causal Domains. CMNA 2008,
8th Int. Workshop on Computational Models of Natural Argument, 31-38.
- N. Green. Analysis of Communication of Uncertainty in Genetic Counseling
Patient Letters for Design of a Natural Language Generation System. Social
Semiotics 20(1), 2010, 77-86.
- N. Green. Representation of Argumentation in Text with Rhetorical
Structure Theory. Argumentation. 24(2), 2010, 181-196.
- N. Green. Towards Intelligent Learning Environments for Scientific
Argumentation. Proc. ITS 2010 Workshop on Intelligent Tutoring Technologies
for Ill-Defined Problems and Ill-Defined Domains, Pittsburgh, PA, June
18, 2010.
- N. Green, R. Dwight, K. Navoraphan, and B. Stadler. Natural Language
Generation of Transparent Arguments for Lay Audiences. Argument and Computation.
To appear (2010-11).
- Presentation
Genetics education interactive risk graphics developed by
undergraduate research assistants:
Contributors
Computer Science Graduate Students:
Tami Britt (Masters Thesis)
Jennifer Brooks (Masters Project)
Rachael Dwight (NCSU, MS Thesis)
Allison Fowlkes
Karen Jirak (Masters Project)
William Moates
Kanyamas (Jenny) Navoraphan (NCSU, MS Thesis)
Brian Stadler (Masters Thesis in progress)
Xuegong Xin (Masters Project)
Computer Science Undergraduate Research Assistants:
Darryl Keeter
Carolyn McCann
Carmen Navarro-Luzon
Timothy O'Dell
Zach Todd
David Waizenegger
UNCG Center for Biotechnology, Genomics,
and Health Research
Students and Faculty of the UNCG Master of Science Program in Genetic Counseling