Exploration of Open Data through Procedural Content Generation
A B Barros Gabriella · 2016
Procedural content generation for games has been studied extensively in the past (Shaker et al. 2015). In many ways, PCG often attempts to mimic a human’s content creation process, and many creative methods have been proposed. However, little work has been done towards integrating open data into PCG. Games that attempt to do so have appeared recently, called data games. (Friberger et al. 2013). They refer to the use of data to automatically generate game content. Such content should represent the data in a way that both is engaging and understandable. This way, the user could interact with and learn from information that may be otherwise difficult or exhausting to engage with. Additionally, users identify easier with what they know: they may find more interesting to explore their hometown as an apocalyptic scenario, or to interact with abstractions of their idols in arcade games. Therefore, this project aims at further exploring the space of data games. The main challenges of data games are data acquisition, transformation and balance. Data acquisition refers to how and where to obtain data. Transformation involves how to parse data into game content, and it differs depending on the data type, the content type and/or the game genre. Finally, balancing involves the task of delivering an enjoyable game experience, and ensuring that the original information is understandable, to some extent, while minimizing misinterpretation. To achieve this project’s goal, our approach consists of exploring how to generate different types of content, using different types of data. This involves the development of several projects of varying length. A general framework for data game generation has been proposed, as shown in Figure 1. It consists of two main parts: a crawler and a parser. The crawler is responsible for gathering and pre-processing data. The parser takes this data and transforms it into whatever game content is required. To generate different content, based on the same original data, it is only necessary to create a different parser. For example, the same information from Wikipedia can be used to generate items, dialogues or game levels. Our first project attempts a somewhat direct transformation: creating map levels based on real maps. Generated levels could be played in FreeCiv 1 , a game based on the Civilization series. OpenStreetMaps 2 provided geographical information that was parsed into the level