Exploration of Open Data through Procedural Content Generation

This paper explores the emerging field of data games, in which open data is used to procedurally generate game content that is both engaging and informative. Its central argument is that integrating open data into procedural content generation can make otherwise dry or difficult information more accessible, since players tend to engage more readily with material they already know, such as their hometown or familiar public figures. The author identifies three core challenges for the field: data acquisition, transformation of data into game content, and balancing gameplay enjoyment against faithful, understandable representation of the source information.

The proposed approach is built on a general framework with two components: a crawler that gathers and pre-processes data from sources such as Wikipedia, DBpedia, OpenStreetMap, and Wikimedia Commons, and a parser that converts that data into specific game content. A key feature of this design is reusability, since the same crawled data can produce different outputs such as items, dialogues, or levels simply by swapping in a different parser. The framework is demonstrated through two projects. The first generates Civilization-style map levels for FreeCiv from OpenStreetMap geographic data, with resource placement handled by image processing and starting-position balance tuned through evolutionary algorithms. The second, a work in progress called Data Adventures, produces a detective game in which non-playable characters and locations derive from Wikipedia articles and the relationships between articles define the path between victim and suspects.

The paper is a position-oriented overview rather than an empirical study, and it does not report survey sizes, quantitative evaluation results, or player data. Its scope is primarily conceptual and project-based, covering the design and early development of several prototypes. Planned future work includes improving gameplay in the adventure game, designing a deckbuilding game with themed cards, and integrating data games with the Video Game Description Language. The paper closes by raising open questions about mapping textual or demographic data onto game attributes and mechanics, noting that less human-readable sources may demand more creative parsing strategies.