Data-Driven Entertainment/Education
Friberger et al. defines data games as “games where gameplay and/or game content is based on real-world data external to the game, and where gameplay supports the exploration of and learning from this data.” They also use a taxonomy to explore the categories of such games. They propose to use open and linked data to procedurally generate game content. One of their examples was Open Trumps based on the popular card game Trumps. The data source for the game content is a UN database of countries and demographics indicators. However, procedural content generation entails some transformation of data into the game content. As the authors caution, we must be careful of transformations that “lead to an unacceptable loss of veracity in relation to the original source.” In our work, the game content is the original data [1, 2, 3]. Moreover, our perspective in this endeavor to foster understanding is general instead of a game focused. Thus, it is from this general point of view that we classify data-driven applications.
The figure below illustrates the classification matrix: rows represent data velocity (static or dynamic) and columns the goals of the application. The first type of goals is informational: the application just provides specific information to the user usually through visualization; interaction with the data is minimal. The second type of goal focuses on interaction with the data for sensemaking; applications in this case are usually implemented as games.

| Informational/Visualization | Interactive/games | Static | 1 | 2 | Dynamic | 3 | 4 |
Table 1 Data-driven applications classification
1 & 3. (Static/Dynamic, Information/Visualization). A plethora of data-driven applications on the data.gov [4] fit in these categories. An example with a static dataset is Food-A-pedia, a tool from the USDA that helps users access nutritional information (calories, added sugar, alcohol, etc…) of over 1000 foods [5]. Another example is a set of application that allows the users to query for information that is not easily accessible or in a human comprehensible format. On the data.gov website, some instances are the ‘Search for Public Schools’ and its sibling applications. Or ‘Find a Health Center’ to help citizens find federally funded health center near them [6].
Data-driven applications with dynamic datasets are more common since they map well to our dynamic world. Moreover, the popularity of smartphones(loaded with sensors) allows us to collect and generate an ever-changing stream of data about ourselves and our environment. On the data.gov website, some examples are AIRNow which reports real time air quality information about a location, or Rodify where users can contribute or check real-time transit information [7, 8]. Other types of examples are interfaces to our personal health data collected through wearable devices like the Fitbit[9]. Finally, in academia, there have been efforts trying to understand the dynamic data produced by people on social networks. The Tweet Sentiment Visualization project is an implementation of such efforts. Based on a keyword supplied by the user, and a set of sentiments, the application supports different type of visualizations representing the relationships between the realtime tweets about the keyword and the sentiments [10]. [11, 12, 13] are other examples.
The major problem with these applications is that they are quick information delivery system. The perspective of the designer is that of a passive user, a consumer of the transformed data. Thus, they do not engage the users to explore the data, discover patterns and acquire new knowledge.
2. (Static, Interactive games). This category comprises mostly location-based games since most location information does not change or, games based on historical data. “Reliving the Revolution” is an educational game that combines augmented reality with location to provide virtual information in a physical environment. The game aims to teach historic inquiry, decision-making and critical thinking. The physical site of the game is the Battle of Lexington, in Lexington MA. Depending on their location of the site, the player interacts with different historical figures to gather evidence on the battle, and then collaborative evaluate who fired the first shot. A pilot study revealed that learning about the Battle through real historical figures’ accounts was an enjoyable experience for the participants [14]. Another project, “Ambient Wood” enabled children to practice their scientific inquiry skill to uncover the complex history or ecology of woodland. The participants received information on their mobile devices about their environment based on their position in the woodland. They were also equipped with different data collection sensors. The findings of two studies revealed that the participants were active in their scientific inquiries, including interpretation of information and collaborations with peers [15]. Although the applications in this category focus on learning, they do not map well to our dynamic world where data on which the system is based is ever changing. In other words, the users’ knowledge and skills are static and constant. However, it is possible to apply the ideas of these applications to dynamic datasets.
4. (Dynamic, Interactive games). Friberger et al describe Flight Leader, a typical application in this category. The datasource is streamed in real-time from the flightradar24.com site. Flight Leader, a real-time strategy game, enables the player to be an aircraft flight path controller. He must adjust the path of real flights or ghost flights and ensure there is no collision. The player loses if there is a collision or if he lands the plane at the wrong destination. Thus, he must understand the characteristics and constraints of flight traffic in order to play the game well [1].
Another set of applications that befit this category are fantasy sports games. The game lets the user act as the manager of a team of athletes and, compete against other teams managers. The users choose which athletes to start, substitute for injured athletes, or trade with other fantasy players. Each player’s score is based on statistics about the performance of their athletes, generated by a single athlete or a team during a real sport event. Thus, up-to-date knowledge of the sport in question is valuable for these games.
Fantasy sports are very popular. In 2013 in the United States, there were 33.5 millions users and the number is expected to rise in subsequent years [16]. This reputation has made fantasy sports an attractive model for some data-driven interactive applications. A well known example is FantasySCOTUS, which is a Supreme Court fantasy league for law students and high school students. It engages law student and encourage young citizen to learn about constitutional law and current Supreme Court decisions. The game play involves predicting how each member of the Supreme Court will vote on a case. Points are awarded based on the correct predictions. To do well in the game, the player should know the voting history of each member, and what politics are associated with the current case. FantasySCOTUS is enjoying great success including a population of more than 10000 users and features in the New York Time or CNN.com[17]. Although not with the same success, there is an emergence of data-driven games based on the fantasy sports model including Fantasy Film League, Fantasy Congress, and Wall Street Survivor [18, 19, 20, 21, 22]. The advantages of these applications are first that even if they are not deployed in an educational environment, they still foster scientific inquiry (data collection/exploration, analysis, interpretation, and prediction). This is confirmed in our recent study of fantasy sports attitudes and practices. Our findings showed that even when the players had the option to let the game engine recommends athletes to them, the majority of them preferred collecting evidence on athletes(past statistics, news reports, peers or experts’ opinion), analyzing the evidence(comparison, ranking) and making predictions (selecting the best players based on their analysis) [16]. The other advantage of these fantasy-sports style applications is they depend on historical data and current data streams. As a consequence, the player must continuously engage data updates and changes in order to play well. The desired effect is that this continuous engagement shall expand the player’s knowledge and render it more sophisticated. Thus, this category is well suited for our endeavor.

[1] Marie Gustafsson Friberger, Julian Togelius, Andrew Borg Cardona, Michele Ermacora, Anders Mousten, Martin Møller Jensen, Virgil-Alexandru Tanase and Ulrik Brøndsted (2013): Data Games. Proceedings of the Procedural Content Generation Workshop at FDG.
[2] Julian Togelius and Marie Gustafsson Friberger (2013): Bar Chart Ball, a Data Game. Proceedings of Foundations of Digital Games (FDG).
[3] Andrew Borg Cardona, Julian Togelius, Aske Walther Hansen, Marie Gustafsson Friberger (2013): Open Trumps Data Game.
[4] Data.gov. https://www.data.gov/
[5] Food-A-pedia. https://www.supertracker.usda.gov/foodapedia.aspx
[6] Data.gov Applications. https://www.data.gov/applications
[7] AirNow. http://m.epa.gov/apps/airnow.html
[8] Roadify. www.roadify.com/
[9] Fitbit. http://www.fitbit.com/apps
[10] Tweet Sentiment Visualization. http://www.csc.ncsu.edu/faculty/healey/tweet_viz/tweet_app/
[11] Justin Cranshaw, Raz Schwartz, Jason Hong, Norman Sadeh (2012). The Livehoods Project: Utilizing Social Media to Understand the Dynamics of a City. Sixth International AAAI Conference on Weblogs and Social Media
[12] Takeshi Sakaki, Makoto Okazaki, and Yutaka Matsuo. 2010. Earthquake shakes Twitter users: real-time event detection by social sensors. In Proceedings of the 19th international conference on World wide web
[13] Michael Mathioudakis and Nick Koudas. 2010. TwitterMonitor: trend detection over the twitter stream. In Proceedings of the 2010 ACM SIGMOD International Conference on Management of data (SIGMOD '10)
[14] Schrier, Karen. Revolutionizing History Education: Using Augmented Reality Games to Teach Histories. Thesis. 2005. http://cmsw.mit.edu/revolutionizing-history-education-using-augmented-reality-games-to-teach-histories/
[15] Yvonne Rogers and Sara Price. Extending and Augmenting Scientific Enquiry through Pervasive Learning Environments. Children, Youth and Environments , Vol. 14, No. 2, Collected Papers (2004) , pp. 67-83. http://www.jstor.org/stable/10.7721/chilyoutenvi.14.2.0067
[16] Dzodom, G. and Shipman, F. Data-Driven Web Entertainment: The Data Collection and
Analysis Practices of Fantasy Sports Players. 2013.
[17] FantasyScotus. http://www.fantasyscotus.net/about-fantasyscotus/
[18] Fantasy Film League. http://www.fantasyfilmleague.com/
[19] Summer Movie League. http://summermovieleague.com/Welcome.aspx
[20] Fantasy Congress. http://www.fantasycongress.net/112/index.php
[21] Wall Street Survivor. http://www.wallstreetsurvivor.com/
[22] Realius. http://predict.realius.com/