A Game Architecture for Emergent Story-Puzzles in a Persistent World

Resource type
Date created
2005-05-30
Authors/Contributors
Abstract
This paper presents the design of a semi-emergent architecture for graphic adventures. The proposed architecture is able to create new significant story-puzzles on the fly, combining preauthored material in a coherent and emergent way. As a test-bed of this architecture we use a simple detective game inspired on the famous boardgame Cluedo. INTRODUCTION There are a lot of adventure games in the market and some of them try to tell a re-playable story that change and become slightly different for each player. Blade Runner is a good example of this kind of adventures in which characters do not act always in the same way. However this approach has the disadvantage of the same structure between each generated story. As each game takes a long time to be solved (usually several game sessions), it is not expected a player to finish completely the game more than once. The aim of our architecture is not to create a stand-alone long-play story and then make it up with slight changes. In fact, our assignment is to generate automatically short stories that happen in a closed and persistent world, in which coherence between each game session is maintained. The player must solve chained mysteries that happen in the same environment, using the conclusions reached in previous mysteries to solve new ones. The clues are obtained from the virtually recreated environment and the social behaviour that characters shows (hatreds, quarrels, debts, favours and secrets). Every element in the context is maintained coherently between stories by the plot generator of the system, that in each game establishes the crime, the scene, the culprit and his motivations. The major part of the story unfolds during the interaction between the player and the non-player characters (NPCs), based on the actions and reactions of the autonomous agents that implement the NPCs, their behaviour, goals and plans. STORY-PUZZLE GENERATION Automatic construction of story plots has always been a longed-for utopian dream in the entertainment industry, especially in commercial games that are fuelled by a large number of story plots with only a medium threshold on their quality. Of course, nowadays computers are not able to generate automatically complex stories like the ones that human writers can write. But now it is possible to take advantage of the combinatory power of computers to build some simple story-puzzles. Basically there are two approaches in the automatic plot generation: centralized and distributed. The first one is based on an intelligent system that controls the development of the plot. The second one is based on a multi-agent system in which the plot emerges from the behaviour of autonomous characters. Developing the game concept, the game designer faces the interactive dilemma: if the story plot is created with a lot of detail the characters are more constrained and the range of generated stories will be narrow, on the other hand the autonomy of the characters takes control over the plot and it is not easy to assure that the result will be valuable. The implementation of the centralized approach is quite complex lacks a truly emergent behaviour. But the consistency is also a hard problem in the distributed generation, especially if the model grows during the design process and becomes more sophisticated. THE GAME ARCHITECTURE We propose a double-layer architecture. The so-called story layer has an abstract plot generator that sets the initial game state, the location of objects and characters and the social relations between them. All these avatars inhabit in the persistent world, known as agent layer. This layer controls the simulation progress, using the guidelines imposed by the story layer. This architecture is a mixed approach to the problem of generating story plots from a set of narrative components analysed in terms of Description Logic. An AI-process is defined to generate plots from the user actions, the preauthored plot specifying and a random initial setting for the story, using an ontology to measure the semantical distance between elements taking part in the story development. Formal logics and ontologies allow the system to maintain coherence and structure in the global puzzle. The agents use the common sense-think-act cycle. Their actions take place over a module that simplifies the interaction with the objects of the virtual environment. APPLICATION Our proposed game belongs to a classical genre: detective stories and unsolved mysteries. The player plays the role of an investigator that tries to find the culprit from the cast of characters. Our approach allows the game designer to have control over both layers. Firstly, abstract generation of each story plot can be controlled establishing the theme, mood and background of the game, the set of possible crimes, scenes, suspects and their motivations. In the second place, the autonomous characters also have a set of different personalities for the designer to combine. CONCLUSIONS Our design lets the system to adapt the content of each game session to the player, depending on the player skills solving the proposed story-puzzle, the system can deduce the way the user is thinking, creating a user model that will improve next generations of mysteries. To sum up, using this architecture adventure games can improve the replayability at the same time that the player experience is personalized and optimised REFERENCES "An Oz-Centric Review of Interactive Drama and Believable Agents". Mateas, M. 1999. "Blade Runner". Westwood Studios, 1997 "Causal Normalisation: A Methodology for Coherent Story Logic Design in Computer Role-Playing Games". Craig A. Lindley and Mirjam Eladhari "The Description Logic Handbook Theory, Implementation and Applications". Franz Baader, Diego Calvanese, Deborah McGuinness, Daniele Nardi, Peter Patel-Schneider 2003
Description
Contact: Federico Peinado, DSIP - Universidad Complutense de Madrid, fpeinado@fdi.ucm.es
Copyright statement
Copyright is held by the author(s).
Language