Alexander Zook

Toward supporting stories with procedurally generated game worlds

Computer role playing games engage players through interleaved story and open-ended game play. We present an approach to procedurally generating, rendering, and making playable novel games based on a priori unknown story structures. These stories may be authored by humans or by computational story generation systems. Our approach couples player, designer, and algorithm to generate a novel game using preferences for game play style, general design aesthetics, and a novel story structure. Our approach is implemented in Game Forge, a system that uses search-based optimization to find and render a novel game world configuration that supports a sequence of plot points plus play style preferences. Additionally, Game Forge supports execution of the game through reactive control of game world logic and non-player character behavior.

Automated scenario generation: toward tailored and optimized military training in virtual environments

Scenario-based training exemplifies the learning-by-doing approach to human performance improvement. In this paper, we enumerate the advantages of incorporating automated scenario generation technologies into the traditional scenario development pipeline. An automated scenario generator is a system that creates training scenarios from scratch, augmenting human authoring to rapidly develop new scenarios, providing a richer diversity of tailored training opportunities, and delivering training scenarios on demand. We introduce a combinatorial optimization approach to scenario generation to deliver the requisite diversity and quality of scenarios while tailoring the scenarios to a particular learners needs and abilities. We propose a set of evaluation metrics appropriate to scenario generation technologies and present preliminary evidence for the suitability of our approach compared to other scenario generation approaches.

Skill-based Mission Generation: A Data-driven Temporal Player Modeling Approach

Games often interweave a story and series of skill-based events into a complete sequence---a mission. An automated mission generator for skill-based games is one way to synthesize designer requirements with player differences to create missions tailored to each player. We argue for the need for predictive, data-driven player models that meet the requirements of: (1) predictive power, (2) accounting for temporal changes in player abilities, (3) accuracy in the face of little or missing player data, (4) efficiency with large sets of data, and (5) sufficiency for algorithmic generation. We present a tensor factorization approach to modeling and predicting player performance on skill-based tasks that meets the above requirements and a combinatorial optimization approach to mission generation to interweave an authors preferred story structures and an authors preferred player performance over a mission---a kind of difficulty curve---with modeled player performance.

Creativity support for novice digital filmmaking

Machinima is a new form of creative digital filmmaking that leverages the real time graphics rendering of computer game engines. Because of the low barrier to entry, machinima has become a popular creative medium for hobbyists and novices while still retaining borrowed conventions from professional filmmaking. Can novice machinima creators benefit from creativity support tools? A preliminary study shows novices generally have difficulty adhering to cinematographic conventions. We identify and document four cinematic conventions novices typically violate. We report on a Wizard-of-Oz study showing a rule-based intelligent system that can reduce the frequency of errors that novices make by providing information about rule violations without prescribing solutions. We discuss the role of error reduction in creativity support tools.

AI for game production

A number of changes are occurring in the field of computer game development: persistent online games, digital distribution platforms and portals, social and mobile games, and the emergence of new business models have pushed game development to put heavier emphasis on the live operation of games. Artificial intelligence has long been an important part of game development practices. The forces of change in the industry present an opportunity for Game AI to have new and profound impact on game production practices. Specifically, Game AI agents should act as “producers” responsible for managing a long-running set of live games, their player communities, and real-world context. We characterize a confluence of four major forces at play in the games industry today, together producing a wealth of data that opens unique research opportunities and challenges for Game AI in game production. We enumerate 12 new research areas spawned by these forces and steps toward how they can be addressed by data-driven Game AI Producers.

Temporal Game Challenge Tailoring

Digital games often center on a series of challenges designed to vary in difficulty over the course of the game. Designers, however, lack ways to ensure challenges are suitably tailored to the abilities of each game player, often resulting in player boredom or frustration. Challenge tailoring refers to the general problem of matching designer-intended challenges to player abilities. We present an approach to predict temporal player performance and select appropriate content to solve the challenge tailoring problem. Our temporal collaborative filtering approach-tensor factorization-captures similarities among players and the challenges they face to predict player performance on unseen, future challenges. Tensor factorization accounts for varying player abilities over time and is a generic approach capable of modeling many kinds of players. We use constraint solving to optimize content selection to match player skills to a designer-specified level of performance and present a model-performance curves-for designers to specify desired, temporally changing player behavior. We evaluate our approach in a role-playing game through two empirical studies of humans and one study using simulated agents. Our studies show tensor factorization scales in multiple game-relevant data dimensions, can be used for modestly effective game adaptation, and can predict divergent player learning trends.

Examining Game World Topology Personalization

We present an exploratory analysis of the effects of game world topologies on self-reported player experience in Computer Role Playing Games (CRPGs). We find that (a) players are more engaged in game worlds that better match their self-reported preferences; and (b) player preferences for game topology can be predicted based on their in-game behavior. We further describe how in-game behavioral features that correlate to preferences can be used to control procedural content generation algorithms.

A framework for exploring and evaluating mechanics in human computation games

Human computation games (HCGs) are a crowdsourcing approach to solving computationally-intractable tasks using games. We outline a formal representation of the mechanics in HCGs, providing a structural breakdown to visualize, compare, and explore the space of HCG mechanics. We present a methodology based on small-scale design experiments using fixed tasks while varying game elements to observe effects on both the player experience and the human computation task completion. Ultimately, we wish enable easier exploration and development of HCGs, letting these games provide meaningful experiences to players while solving difficult problems.

Program synthesis as a generative method

Generative methods (also known as procedural content generation) have been used to generate a variety of static artifacts such as game levels. One key property of a generative method for a particular domain is how effectively the approach allows a designer to express the properties and constraints they care about. Generative methods have been applied much less frequently to dynamic artifacts such as boss behaviors, in part because the complex representation required to describe boss morphology and behavior is not amenable to existing generative techniques. It is challenging to describe a generative space of varied yet valid behaviors. Expanding on previous work that introduced a programming language for representing boss behaviors, we illustrate how such a language can be used by a designer to describe desirable design properties and constraints for bosses. That is, we define a generative space of bosses as a space of well-formed programs. We present a constructive algorithm that extends generative grammars to efficiently generate well-formed programs, and we show a complete example of generating Mega-Man-like bosses with complex attack patterns. We conclude that designing a generative space of dynamic behaviors can be fruitfully framed as a programming-language design problem.

Workshops Held at the Ninth Annual AAAI Conference on Artificial Intelligence and Interactive Digital Entertainment (AIIDE): A Report

The Ninth Annual AAAI Conference on Artificial Intelligence and Interactive Digital Entertainment (AIIDE) was held October 14–18, 2013, at Northeastern University in Boston, Massachusetts. Workshops were held on the two days prior to the start of the main conference, giving attendees a chance to hold in-depth discussions on topics that complement the themes of the main conference program. This year the workshops included the First Workshop on AI and Game Aesthetics (1 day), The Second Workshop on AI in the Game Design Process (1 day), The Second International Workshop on Musical Metacreation (2 day), The Sixth Workshop on Intelligent Narrative Technologies (2 day).