A growing issue in the development of realistic and entertain-ing interactive games is the need for mechanisms that support ongoing natural language conversation between human players and artificial non-player characters. Unfortunately, many methods for implementing natural language generation(NLG) induce a significant burden on the author, do not scale well, or require specialized linguistic knowledge. We formalize the notion of typed-templates, an extension of standard structures employed in template-based NLG. We further provide novel algorithms that, when applied to typed-templates, ameliorate the above issues by affording computational support for authoring and increased variation in utterance and scenario generation. We demonstrate the efficacy of typed-templates and the algorithms through a user study.

RTAA* is probably the best-performing real-time heuristic search algorithm at path-finding tasks in which the environ- ment is not known in advance or in which the environment is known and there is no time for pre-processing. As most real- time search algorithms do, RTAA∗ performs poorly in presence of heuristic depressions, which are bounded areas of the search space in which the heuristic is too low with respect to their border. Recently, it has been shown that LSS-LRTA∗, a well-known real-time search algorithm, can be improved when search is actively guided away of depressions. In this paper we investigate whether or not RTAA∗ can be improved in the same manner. We propose aRTAA∗ and daRTAA∗, two algorithms based on RTAA∗ that avoid heuristic depressions. Both algorithms outperform RTAA∗ on standard path-finding tasks, obtaining better-quality solutions when the same time deadline is imposed on the duration of the planning episode. We prove, in addition, that both algorithms have good theoretical properties

Meaningful choice has often been identified as a key component in a player's engagement with an interactive narrative, but branching stories require tremendous amounts of hand-authored content, in amounts that increase exponentially rather than linearly as more choice points are added. Previous approaches to reducing authorial burden for computer RPGs have relied on creating better tools to manage existing unwieldy structures of quests and dialogue trees. We hypothesize that reducing authorial burden and increasing agency are two sides of the same coin, requiring specific advancements in two related areas of design and technology research: (1) dynamic story management architecture that represents story events abstractly and allows story elements to be selected and re-ordered in response to player choices, and (2) dynamic dialogue generation to allow a single story event to be revealed differently by different characters and in the context of dynamic relationships between those characters and the player. This paper describes SpyFeet, a playable prototype of a storytellingsystem designed to test this hypothesis.

Due to the changing nature of videogames, the component-based architecture is the design of choice for managing game entities instead of the traditional static class hierarchies. A component-based architecture lets programmers edit entities as collections of components, which provide the entity with new functionalities. Such architecture promotes flexibility but makes the code more difficult to understand because entities are built at runtime by linking components. In this paper we present a semi-automatic process for moving from a class hierarchy to a component-based architecture. Through the application of Formal Concept Analysis we propose a novel technique for automatically identifying candidate distributions of responsibilities among components.

We study the problem of learning probabilistic models of high-level strategic behavior in the real-time strategy (RTS) game StarCraft. The models are automatically learned from sets of game logs and aim to capture the common strategic states and decision points that arise in those games. Unlike most work on behavior/strategy learning and prediction in RTS games, our data-centric approach is not biased by or limited to any set of preconceived strategic concepts. Further, since our behavior model is based on the well-developed and generic paradigm of hidden Markov models, it supports a variety of uses for the design of AI players and human assistants. For example, the learned models can be used to make probabilistic predictions of a player's future actions based on observations, to simulate possible future trajectories of a player, or to identify uncharacteristic or novel strategies in a game database. In addition, the learned qualitative structure of the model can be analyzed by humans in order to categorize common strategic elements. We demonstrate our approach by learning models from 331 expert-level games and provide both a qualitative and quantitative assessment of the learned model's utility.

In this paper, we discuss approaches to dialogue in interactive video games and interactive narrative research. We propose that situating interactive dialogue in the simplified expectations of video games is a profitable way to investigate computational dialogue simulation. Taking cues from existing physical simulations such as combat, we propose a hypothetical game environment and design goals for an embedded interactive dialogue system. We present a modular framework targeted at that environment, which is designed to enable incremental development and exploration of dialogue concepts. We describe this framework together with a work-in-progress system for simulating simple in-game negotiation dialogues.

Interactive narrative environments offer significant potential for creating engaging narrative experiences that are tailored to individual users. Increasingly, applications in education, training, and entertainment are leveraging narrative to create rich interactive experiences in virtual storyworlds. A key challenge posed by these environments is devising accurate models of director agents’ strategies that determine the most appropriate director action to perform for crafting customized story experiences. A promising approach is developing an empirically informed model of director agents’ decision-making strategies. In this paper, we propose a framework for learning models of director agent decision-making strategies by observing human-human interactions in an interactive narrative-centered learning environment. The results are encouraging and suggest that creating empirically driven models of director agent decision-making is a promising approach to interactive narrative.

In this paper, we are considering advanced pathplanning problems that feature finding paths for multiple units subject to rich path constraints. Examples of richer constraints are the following of other units or to stay out of sight of a specific unit. Little attention has so far been given to richer pathplanning problem where the objective is more than reaching a specific destination from a starting point such that the path length is minimized. Richer pathplanning problems occur in many complex real-world scenarios, ranging from computer games to military movement planning. In this paper, a novel way to formally specify such problems and a new local-search strategy to solve such problems are proposed and demonstrated by a prototype implementation. Among the design goals are real-time computability as well as extendibility for new constraints and search heuristics.

In this paper we focus on proactive behavior for non-player characters (NPCs) in the first-person shooter (FPS) genre of video games based on goal-oriented planning. Some recent approaches for applying real-time planning in commercial video games show that the existing hardware is starting to follow up on the computing resources needed for such techniques to work well. Nonetheless, it is not clear under which conditions real-time efficiency can be guaranteed. In this paper we give a precise specification of SimpleFPS, a STRIPS planning domain expressed in PDDL that captures some basic planning tasks that may be useful in a first person shooter video game. This is intended to work as a first step towards quantifying the performance of different planning techniques that may be used in real-time to guide the behavior of NPCs. We present a simple tool we developed for generating random planning problem instances in PDDL with user defined properties, and show some preliminary results based on SimpleFPS instances that vary in the size of the domain and two well-known planners from the planning community.

Probabilistic models were fit to logs of player actions in the card game Dominion in an attempt to find evidence of personality types that could be used to classify player behavior as well as generate probabilistic bot behavior. Expectation Maximization seeded with players' self-assessments for their motivations was run for two different model types — Naive Bayes and a trigram model — to uncover three clusters each. For both model structures, most players were classified as belonging to a single large cluster that combined the goals of splashy plays, clever combos, and effective play, cross-cutting the original categories — a cautionary tale for research that assumes players can be classified into one category or another. However, subjects qualitatively report that the different model structures play very differently, with the Naive Bayes model more creatively combining cards.