Explorable Formal Models of Privacy Policies and Regulations (POEM students: Chinmaya Dabral). 2019--present

Supported by NSF CAREER Award #1846122.

Data collection and analysis enable great advancements in digital technology, but the stewards of this data have a responsibility to society to ensure that the practices of collection, storage, and user control abide by user expectations. Policies and regulations governing data privacy play a critical role in communicating privacy expectations to users and defining the bounds of permissible data use. However, in practice, there are severe mismatches between user expectations and the actual practices of software companies, even when those practices conform with their privacy policies. This project's goal is to enable automated reasoning over privacy policies and regulations that can assist users, policy developers, and regulators in understanding unintended consequences of data practices and policies.

Intelligent Support for Creative, Open-Ended Programming Projects (PI: Thomas Price, co-PI: Tiffany Barnes; POEM students: Alexander Card, Samuel Henderson). 2019--present

Supported by NSF Award #1917885.

Introductory computing curricula increasingly motivate students with creative, open-ended projects, such as making apps, games, and simulations. Advanced Placement (AP) exam results from the Computer Science Principles exam indicate that students need more support in solving open-ended problems. Current research on automated help systems can provide hints but cannot provide much information at higher abstraction levels. This project will provide technologies to support novices in open-ended program design and construction. The longer-term promise of this research is that it can be extended beyond novices' learning of computational thinking skills to a wider range of learning tasks to better prepare the workforce of the future. Additionally, the research contributes to computer science research by using novel data-driven technologies to identify the higher-level advice provided to the students.

Ceptre: Supporting Domain-Specific Inquiry with Rule-Based Modeling (POEM Students: Alexander Card, Jacky Hong, Kamai Guillory). 2015--present.

Supported by NSF Award #1755922.

Ceptre is a forward-chaining logic programming language based on linear logic, which can be used to model simulations of complex systems made from sets of rules. We have used this language to model games such as Minecraft and Overwatch, to model the dynamics of group conversations, and to generate stories with social simulation. The next generation of the Ceptre language is geared towards interdisciplinary communication; we aim to broaden participation in use of the tool to experts in other domains (such as chemistry, systems biology, and psychology). We are building a web-based structure editor with simulation analysis and debugging tools. This project will combine software engineering, human-computer interaction, and systems thinking disciplines such as game design, to create a rich modeling environment to bridge domains requiring system simulation.


Villanelle: Autonomous Character Authoring for Interactive Stories (POEM students: Jennifer Wellnitz, Owais Iqbal, Pascal Le, Rook Liu, Claire Christopher, Sasha Azad, Tony Mosolf, Maddie Ingling, Siyu Zhang, Emma McCamey, Janna Timmer). 2016--present

Project Page

Autonomous characters have the potential to transform game experiences as NPCs (non-player characters), interactive narratives, user interfaces, our understanding of human social interaction through formal models, and much more. Currently, authoring autonomous characters and generative storylines is seen as an esoteric and inaccessible practice for interactive narrative designers. We aim to provide accessible tools for character authoring.

We are building a web-based Javascript-interfacing framework called Villanelle for specifying world models, player actions, and non-player-character behavior.


Narrative Generation for Real World Locations (POEM students: Sasha Azad). 2019--present.

The goal of this project is to develop techniques for narrative generation that take into account real-world geographic constraints, where narratives emerge from the simulation of human populations. This project involves (1) simulating virtual populations of characters who respond believably to these constraints, to the affordances of built and natural environments, and to social and economic prerogatives such as family and work obligations; and (2) the usage of geographic data (such as maps) as meaningful simulation inputs. We aim to better support the development of playable, explorable story worlds that engage players with real-world geographical locations and our simulated populations.

Lyra: Social Simulation with Opinion Change (POEM students: Sasha Azad). 2017--present.

Creating believable simulations of large populations of virtual characters in virtual worlds represents a grand challenge for artificial intelligence. In this project we are developing models of opinion change and social influence among virtual characters, operationalizing a kind of social intelligence through the emergent formation of groups and identification of individuals with different groups. We use cognitive and social science theories to structure the flow of knowledge, opinions, and biases through social networks. This work provides a foundation for computational understanding of human social dynamics and, in the long run, may enable more effective human-machine collaboration.


Understanding Mental Model Formation through Puzzle Games (POEM students: Abhijeet Krishnan, Aaron Williams, and Ryan Alexander). 2018--present.

Puzzle game players and designers tend to agree on the ingredients of a satisfying and instructive puzzle: the player builds a mental model of the game's mechanics as she plays, and then the puzzle throws a wrench in her assumptions. After struggling, having an "aha" moment, and finally understanding and correcting the misleading assumption, the player's skill with the mechanics deepens. In this project, we're designing Laserverse, a set of game mechanics and levels in PuzzleScript, designed to compose modularly to form different test conditions for this hypothesis. In the long term, we're interested in generating puzzled toward specific experiential and educational ends.



Ostari: Modeling Belief and Intention in Virtual Agents (Markus Eger). 2016--2019.

Social reasoning, such as theory of mind, supposes that people have mental models of other people's mental models, and will act on those mental models intentionally. We examine contexts in which this form of social reasoning is important, such as board and card games that involve coöperation, bluffing, and partial information. We have created a programming framework called Ostari with explicit support for belief (epistemic reasoning) with authoring affordances for intentions that drive agents through planning.


Modeling Social and Cultural Norms through Interactive Narrative and Conversational Agents (Hannah Morrison and Louis Jacobowitz). 2016--2018

Conversational agents are typically limited to passive, 1:1 conversations with a single presumed-human interactor. We are exploring the space of possible agent simulations involving combinations of many simultaneous conversational participants, the upholding or breaking of social norms, and agents that exhibit intention and pursue goals. Possible applications include non-player characters for interactive stories, social skills training, therapeutic tools, interactive tutors, and collaborative systems. Currently we are working on adapting our models to a cultural competency training tool in collaboration with the Global Training Initiative.


Logic-Based Narrative Generation (Henry Mohr; Markus Eger; Rogelio Cardona-Rivera). 2015--2018.

We use logical formalisms such as linear logic to specify templates for narrative events, then use logic programming engines to produce stories from these specifications. We are interested in modeling properties such as causality and continuity through logical consistency.