Research Interests and Affiliation
To date, my research has focused on computational and cognitive models of narrative understanding and generation. Narrative is a fundamental element of human cognition, a method of cognitive sense-making, often used for entertainment, communication, training, and education. My research approaches narrative from the standpoint of cognition, and attempts to build models of the ways in which narrative is created to be understood. My work has been applied towards the generation of narrative within interactive systems such as computer games, educational software, and training simulations. I seek to establish a broad research program that will focus on the design of computer systems that generate and understand narrative in the ways that humans do and on the use of this technology in entertainment, training, and education. I recently defended my dissertation with Dr. Michael Young at NCSU, and working with Dr. Mark Riedl at Georgia Tech as a Postdoctoral Fellow.
Papers
REFEREED JOURNALS
[2007] KERNELPOP, a Spatially Explicit Population Genetic Simulation Engine , Strand, Allan and Niehaus, James, Molecular Ecology Notes, v7, pg. 969-973, 2007.
REFEREED CONFERENCES AND WORKSHOPS
[2009] Scenario Adaptation: An Approach to Customizing Computer-Based Training Games and Simulations. Niehaus, James and Riedl, Mark Working Notes of the Workshop on Intelligent Educational Games at the Sixteenth International Conference on Artificial Intelligence and Education. Brighton, England, July 2009. (In press)
[2009] Toward Scenario Adaptation for Learning, Niehaus, James and Riedl, Mark, Poster presentation, Sixteenth International Conference on Artificial Intelligence and Education. Brighton, England, July 2009. (In press)
[2009] A Computational Model for Inferencing in Narrative, Niehaus, James and Young, R. Michael AAAI Spring 2009 Symposium on Interactive Narrative Technologies II. Palo Alto, CA; March 23-25, 2009.
[2008] Automatically Predicting Causal Inferences in the Modified Event Indexing Model, Niehaus, James and Young, R. Michael Eighteenth Annual Meeting of the Society for Text and Discourse. Memphis, Tennessee; July 12-15, 2008.
[2005] Toward improving recall and comprehension in automatically generated narratives, Niehaus, James and Young, R. Michael, Working Notes of the Workshop on Narrative Learning Environments at the Twelfth International Conference on Artificial Intelligence and Education. Amsterdam, The Netherlands, July 2005.
OTHER
[2008] A Computational Model for Focus and Inferencing in Narrative Comprehension, Niehaus, James Oral Examination, North Carolina State University, January 2008.
[2006] Employing Cognitive Theories of Discourse Comprehension to Improve Learning in Interactive Narratives, Niehaus, James and Young, R. Michael, Poster Presentation, The Twenty-First National Conference on Artificial Intelligence Doctoral Consortium (AAAI-DC-06), Boston, Massachusetts, July 2006.
[2006] Achieving Authorial Goals in Computational Narrative Generation, Niehaus, James Written Examination, North Carolina State University, March 2006.
Projects
-
INFER
INFER is a work in progress implementation for my dissertation. The INFER system (INferences For Extending Recall), employs the low level constructs of narrative focus, causal inferences, and intentional inferences to increase recall and comprehension of elements within the narrative. Specifically, I formalize aspects of the focusing and inferencing of the Event Indexing (EI) model (Zwaan, Langston, and Grasser 1995), a cognitive model of narrative comprehension, to measure each construct in the generated discourses. INFER uses these constructs to generate a discourse from a set of events, attempting to achieve desired comprehension effects, either shifting focus to a particular event or character or prompting desirable inferences.
-
Mimesis and Zocalo
Mimesis (Young 2001) and it's Web Service sibling, Zocalo (Vernieri 2006) are the integration of several software components related to planning. I have worked to integrate various tools and theories into the planning architectures. The liquid narrative lab has used these systems in interactive narrative research to drive educational virtual environments.
-
Longbow
Longbow (Young 1994) is the first extension to the DPOCL planning system and is targeted specifically for discourse planning applications. Longbow is a discourse planning system built on a theory of communication as planned, intentional action. I have worked to bugfix and integrate Longbow into Zocalo and INFER.
-
Virtuoso
Virtuoso is an undertaking by the HI-FIVES project at NCSU. Virtuoso is a program designed to help people who are not experienced programmers build 3D computer games. Virtuoso's user interface makes the process of creating a 3D world, populating it with interactive objects and specifying their behaviors much simpler than in typical game development environments. Using Virtuoso, groups of users work together within a 3D world to build games containing many of the conventions of commercial games, such as HUDs, multi-user, networked and team play, end-game conditions, splash screens and more.
-
Ultimate Duck Hunting
Ultimate Duck Hunting is a 3D duck hunting game featuring realistic duck hunting action in career and online multiplayer modes. It was sold for a time at your local Walmart. I was a full 50% of the programming staff!
-
WideAsleep
Unreal Tournament 2004 Mod developed with students in the Game Development course at NCSU. A small boy is tasked with escaping his nightmares, literally. Indy Week picked up the story of the class showcase.
-
CLforJava, Writing a Common-Lisp Compiler for the Java Virtual Machine (past)
The parent project, Common Lisp for Java, was started Fall 2003 by Prof. Jerry Boetje and is slated to last four years. The project is coded, documented, and maintained by students completing independent studies, such as myself, as well as the 400 level Software Engineering class at the College of Charleston. I led the compiler team to complete a bootstrap compiler for compiling Common-Lisp to Java bytecode with all but two of Lisp's special operators.
-
Rmetasim, An Individual-Based Population Genetic Simulation Environment (past)
A project with Dr. Strand of the College of Charleston Biology Department. We used Dr. Strand's individual-based stochastic population engine, Metasim, to test theories on the effectiveness of many genetic estimates for single-species inter-population migration. We have wrote a clean, simple, and well documented interface to Metasim for the R statistical language, Rmetasim.
