shop’s first iteration held in 2011, asked participants
to reflect on how to provide real-time and interactive
feedback to designers during the design process. In
response, participants shared systems and position
papers that aligned to four main themes: (a) real-time
visual feedback during design, ( 2) exploratory tools
for designers, ( 3) using declarative knowledge representation and constraints, and ( 4) player modeling
We saw a system that made emergent behavior visible on an interactively editable map, and heard a
proposal for generic infrastructure for continuous
game-play trace sampling that would keep these de-sign-time views up to date. We saw a design exploration tool for a commercial puzzle game that leveraged exhaustive search techniques as well as a
simulation-based player behavior preview tool integrated into the widely used Unity game development
environment. We heard about how to use formal ontologies to validate game implementations, and saw
the use of grammars and other structural constraints
for controllably shaping the space of a map generator’s output. We saw and discussed the use of machine-learning techniques, in contact with game analytics, for modeling player behavior and using that
to provide data-informed feedback on novel designs.
Following the morning paper presentations, the afternoon session for the workshop was devoted to a
lively group discussion on a variety of topics of interest to the community, which converged upon
three main discussion groups addressing: evaluation
and generalization, player modeling, and automated
QA as a service. Outcomes from this discussion included a set of future research directions in how to
evaluate creative systems, what success means for
such systems, and how to design new AI systems
whose role in the design process is that of a colleague
with the human designer. There were also several
suggestions for concrete systems that could be built,
including a dynamic matchmaking system that
would reason over how individuals cooperate in mul-tiplayer online battle arena (MOBA) games in order
to ensure that teams will cooperate and have fun together. Finally, there was a suggestion for creating a
cloud-based automated testing service that would
make computationally intensive, AI-based design automation accessible to a wider audience.
Adam M. Smith, Gillian Smith, and Mark J. Nelson
cochaired the workshop. Papers from the workshop
were published as AAAI Press Technical Report WS-
The Sixth Workshop on Intelligent
Narratives are a pervasive aspect of human culture
and are one of the core frameworks people use to
view and understand the world. People use narratives
for social interaction, education, and entertainment.
Narratives as forms of knowledge representations
have long held interest among AI and cognitive science researchers alike. Imbuing technologies with
narrative intelligence to represent and reason on narratives holds great promise. Narrative technologies
deepen our understanding of the human mind, and
enhance computational systems to more naturally
communicate with human users. Further, these technologies augment the human capacity for expression
to enable new forms of narrative experiences.
The Workshop on Intelligent Narrative Technologies aimed to advance research in artificial intelligence for the computational understanding and expression of narrative. Previous meetings of this
workshop have brought together a multidisciplinary
group of researchers including computer scientists,
psychologists, narrative theorists, media theorists,
artists, and members of the interactive entertainment
industry. From this broad expertise, the workshop focuses on computational systems that represent, reason about, adapt, and perform interactive and noninteractive narrative experiences. This includes
fundamental research in relevant fields such as natural language processing, believable agents, commonsense reasoning, and human narrative experiences.
Creating narratives through authoring tools and
narrative generation was a major theme at the 2013
workshop. Authoring tools augment human capabilities for crafting narrative experiences. Tools for validating interactive narratives (Kim Dung Dang and
Ronan Champagnat) and automatically converting
narrative events into virtual world actions (
Alexander Shoulson, Mubbasir Kapadia, and Norman I.
Badler) demonstrated the range of methods for supporting authors.
Narrative generation investigates how AI systems
can automatically create all or part of a narrative. The
workshop participants learned methods for generating aspects of narratives including character-centered
narrative plots (Bilal Kartal, John Koenig, and
Stephen J. Guy), language expression for narratives
(Marilyn Walker, Jennifer Sawyer, Carolynn Jimenez,
Grace Lin, Elena Rishes, and Noah Wardrip-Fruin),
and cognitive models of narratives (Justin Permar
and Brian Magerko).
Text mining and natural language processing
emerged as prominent new themes at this workshop.
Online blogs and social media provide a massive
repository of narratives and a rich source of knowledge for understanding and creating narratives. Several presenters discussed methods for mining personal narratives from the web (Andrew Gordon,
Luwen Huangfu, Kenji Sagae, Wenji Mao, and Wen
Chen) and deriving formal features of narratives such
as narrative clauses (Elahe Rahimtoroghi, Reid Swanson, Marilyn A. Walker, and Thomas Corcoran) or
character roles (Josep Valls-Vargas, Santiago On-tanon, and Jichen Zhu).
One of the workshop’s key strengths is bringing to-