> cis

Computer and Information Science

University of Pennsylvania

Nick Montfort

This section has material from my Ph.D. studies in computer and information science at the University of Pennsylvania. I defended my dissertation on 2007-06-20. My full dissertation, "Generating Narrative Variation in Interactive Fiction," is now available online. I was co-advised by:

My dissertation research was in interactive fiction, narratology, and computational linguistics — specifically, in natural language generation. My project involved modeling the content plane (story, mythos, fabula, narrated, what is told about) in IF separately from the expression plane (discourse, logos, sjuzet, narrating, telling).

Thesis Proposal Abstract


Nick Montfort
Mitchell P. Marcus and Gerald Prince
31 July 2006

A general method for narrating in natural language is described, one which allows the expression or narrative discourse to vary independently of the narrative's content, the underlying events and existents. Specifically, this variation is accomplished in a sort of literary dialog system which is based on an underlying simulation: an interactive fiction (IF) system.

IF systems, which have existed for about 30 years, are forms of text-based computer simulation, instances of dialog systems, and examples of literary art. Theorists of narrative have long distinguished between the level of underlying content (corresponding to the simulated world in interactive fiction) and that of expression (corresponding to the textual exchange between computer and user). While this distinction has been developed in the field of narratology beginning in the mid-1960s, IF systems have not yet distinguished between the telling and what is told.

The current project builds on computational linguistics techniques in natural language generation and on narratological concepts, contributing new techniques for automatic narration. First, types of narrative variation that are possible in IF are identified and a formalization of them suitable for use in a natural language generation system is developed. An architecture for an IF system is then described and implemented; the result is capable of running multiple works of interactive fiction and allows narration of the same events in different ways. Extensions to this system and ways of evaluating the system's ability to generate fluent text are proposed to complete the project.

The interactive fiction section of my site has much more about this topic, including several articles of mine from other disciplinary perspectives and my own creative work in interactive fiction.

Other research I started at Penn is on:

Previous research at Penn includes:

I have taught two sections of Fernando Pereira's intro to programming course. I co-taught an intensive four-week course in programming with Jean Griffin as part of the Pre-Freshman Program (PFP). I have been a teaching assistant for Michael Kearns in his Networked Life course.

I've also been invited to be a guest speaker in several classes at Penn:

Perform (Perceptron Classifier in Inform) was developed in one CIS seminar that I took.

In August 2004 I finished the report "Discovering Communities through Information Structure and Dynamics." This was for my WPE-II (Written Preliminary Exam-II). I did a presentation on this topic, too, as part of the exam. The slides are online. Since the Penn CIS Graduate Handbook requires that the report and slides be published on the Penn CIS server, they appear there, too: report, slides.