With the same participants

CAD-GAME: Computer-aided game design

Project: Research project (funded)Research

Project participant(s)

Department / unit(s)


The aim of this project is to develop new tools for the testing, analysis and design of video games, based on computational analysis of data from live gameplay. We will be applying a range of data mining techniques to the captured gameplay data, and we will use the resulting insights to build tools which fine-tune the various parameters that control the way a game plays. As a design tool, getting richer information from game testing will reduce costs and increase the overall quality of our games. As a tool to run within a game, we will investigate methods for adapting how the game plays in response to various indicators of the player's behaviour. This will give us the ability to create much more subtle game experiences which are tailored for individual players. As a marketing tool, this potentially lets us segment our customer population in much more meaningful ways, opening up significant opportunities in marketing and cross-selling.The project objectives remained unchanged throughout the project and were:-

To build tools to capture, analyse and visualise gameplay data.
To conduct user testing studies of gameplay for different scenarios and to develop methodologies for approximating a user's enjoyment and immersion in a video game at fairly fine-grained intervals during the game.
To apply existing machine learning and data mining techniques to the problem of finding causal relationships between the gameplay data and the overall measures of enjoyment and immersion.
To develop novel data analysis techniques, tailored to this particular application area, that offer an increased understanding of the underlying dynamics of gameplay.
To use the resulting insights to build tools for testing and fine-tuning game designs.
To build in-game tools which modify game parameters in response to the player's behaviour as they advance in the game.
To develop a game demo based around the concept of adaptive game control, and to further deliver a full version of this game to customers.

The objectives reflect the combination of basic research into the underlying causes of player behaviour, and the application of the results in the development of tools and methods to make richer, more engaging video games.

Layman's description

As a form of media - increasingly, mass media - games have much in common with film, particularly in terms of business models. One reason for the high levels of risk in film is that only limited market testing can be done prior to product release, and the product is essentially fixed from then on. Games have traditionally followed this model, but it is important to remember that games are not just media - they are also pieces of software. One major lesson from the software industry is that capturing users' interactions with a product can be used to refine market offerings in several ways: by adapting it after release (e.g. adaptive interfaces), by using that data to iteratively refine products in future releases (e.g. error reporting in Microsoft and Apple), and to understand the various user groups in more depth. In essence, we are suggesting that this approach can be directly applied to games. The difficulty, however, is that we cannot use simple measures of task completion, because there is no task: we are simply trying to entertain. The main research challenge is to understand better just what the word 'entertain' really means in games, and how this might be reflected in patterns of player behaviour. To do this, we undertook studies in the the enjoyment and immersion players experience when playing video games in order to derive a concrete methodology for closely estimating the amount of enjoyment and immersion a player has, at fairly fine-grained intervals of a game. We also captured a multitude of different forms of data from players actually playing Rebellion's games, and we used advanced techniques from Artificial Intelligence to search for correlations with the player's overall enjoyment. These techniques could, in principle, be used during design to speed up the process of tweaking the various game parameters - where the enemies are, how effective weapons are, etc. More interesting, however, is the idea that data analysis tools that we developed can also run within the game itself, adapting the game in response to the player's behaviour. In this way, this work points the way to a new age of user-adaptive video games, within a "one game, many gameplays" paradigm, i.e., where each player has a unique game experience which is tailored to their personality, level of experience, playing style and mood.

Key findings

The project produced a new piece of software that can capture fine-detailed actions by players in games and using user interviews relate these to the experiences that players were having and how they understood what they were doing. This is the first step towards thinking about how to adapt games to provide different experiences through matching patterns of behaviour to particular experiences.
We also found several links between features of the game and experiences that people had for instance the skill and balance not only need to be balanced but that players need to perceive the balance to become engaged in the game. Also that interaction style influences how engaged players are: even though some interactions styles are more novel and just as effective, they do not engage the players in the same way.
Effective start/end date1/03/0928/02/12

Award relations

CAD-GAME: Computer-aided game design

Cairns, P. A.

EPSRC: £27,251.00


Award date: 17/10/08

Award: UK Research Councils


  • Tools for game data analytics

    Impact: Technological

Research outputs

Discover related content

Find related publications, people, projects, datasets and more using interactive charts.

View graph of relations