Say Anything

As part of my thesis, I developed an interactive storytelling system that collaboratively writes a story with a human author. Unlike most interactive storytelling systems, Say Anything can respond to the user and continue the story in virtually any domain the user chooses. It does this by mining personal stories from weblogs that are about everyday life and experiences of the author.

Story Upgrade

Story Upgrade is an application that combines two technologies in order to provide an effective means of finding stories in Internet weblogs. First, it incorporates technology for separating stories from other textual genres that appear in Internet weblogs. About 14% and 17% of the text in weblogs consists of story content. In order to selectively retrieve only story content the Story Upgrade system incorporates statistical text classification technologies to identify story segments in weblogs. Second, it information retrieval techniques for matching the text of extracted stories to paragraph sized descriptions of activities. These descriptions are provided by the user and serve as the query to the retrieval engine.

Story Engine

While at Disney Research & Development I helped develop technologies for enabling persistent live-action roleplaying experiences that utilized the resources and cast members of the parks and resorts.

SIREN – Social Games for Conflict Resolution

SIREN was a large European Union grant with several international collaborators from the UK, Portugal, Denmark and Greece. The goal of this project was to develop a series of serious games to help adolescents improve their ability to resolve conflicts with their peers.


IMMERSE was a large DARPA project that investigated the ability to train military personal in general social skills useful for navigating social interactions regardless of the particular cultural or linguistic setting. The program looked at both the fundamental social science theory on what constitutes this type of fundamental skill set and how these theories can be modeled computationally. Our primary contribution was the development of an interactive virtual training simulation that incorporated the social and computation research.

Persuasion in Social Media

In this project we investigated the properties of discourse between multiple individuals and how these properties affect the structure of their communication. In 2012 Walker et al. introduced the internet argument corpus (IAC) to address these issues. The IAC is a large collection of online discussions from online debate sites such as, and These discussions unfold more like verbal dialogues where relatively small pieces of information are presented for others to respond. The mutual understanding between participants is updated incrementally as information is added, refined or rebutted. This project examined how various types of language and discourse actions, such as opinions, facts, sarcasm and insults, affect the ability to persuade the community participating in the discussion.

Interpretation of Personal Narratives

The primary goal of this project was to investigate if narrative discourse structure affects attitude change in a reader. Regardless of the content of the message it is possible that the way narratives are structured could play a role in how receptive a reader is to a particular narrative. We annotated stories from several different cultures in their native languages using a schema based on a theory developed by Gennette. We found that these stories were in fact structured differently and that the structure had a small but significant impact on the reader’s attitude toward the narrative.