The main subject of my final graduation project was Social Networks and it’s a subject I really enjoy studying. I will say a few things about my Project, how the research was conducted, a small explanation about social networks and a download link to my thesis (pdf file, in Portuguese).
This research was accepted as a full paper, written in conjunction with my mentor, Pollyana Notargiacomo Mustaro, for the SBGames Symposium in 2006. The paper is in portuguese and you can download it from the official website. If you need an english version, please contact me.
A mathematical approach of this research was accepted as a full paper for the IADIS Multi Conference on Computer Science and Information Systems (MCCSIS) e-Society 2007. The paper entitled Social Network Analysis of Virtual Communities in Online Games also written in conjunction with my mentor was awarded an Outstanding Paper Award! This paper is only available in english, if you need it, please, contact me.
Now I will talk about my thesis. The project is written entirely in Portuguese, because I’m Brazilian, but I will explain a little bit of each chapter in English, and if you wish to obtain explanation about anything just contact me. It was my final project in college, and I chose Social Networks applied to online game players as the subject. For that research I had to use complex computational resources that were also part of my project. If you want to see the abstract in English, click here to view it on a pop up, or in Portuguese.
My mentor was my former "Computer in the Society" and "Methodology" teacher, Dr. Pollyana Notargiacomo Mustaro who studies social networks analysis. She is also interested in long distance education and other subjects.
In the first Chapter of my work I elucidate the history of electronic games over the years. Starting from Huizinga’s (Homo Ludens) concept that play is primitive and the first games of the 50s and 60s. Next is the birth of videogames, and then the new and modern MMORPGs of these days. That is an important chapter that links this history with technology and society history as well. In the end of this chapter I made a game history timeline related to my work and its most important facts. (see timeline - see in Portuguese)
In the first part of the next chapter I explain the social network theory, its main features and characteristics and how it can be studied and represented using graph theory. For that I give some examples of social networks using graphs and matrices, afterwards presenting some calculations that can be done to identify some properties of the network. I also give some details of graph theory that are important to the study of social networks.
Social Network Analysis is the study of the network of relationships between people, agents, corporations or groups of people/agents/etc or anything that can have relationships (called actors). This study analyzes how these relationships are bounded and related and it can verify many characteristics of these actors and their relationships. Those relationships are called ties, and they can be strong or weak, reciprocal or not.
|
A social network is represented mathematically as a graph, so you can calculate its properties from its matrix. An example of a simple social network is a group of people where I can ask each one who they like (note that if A likes B, it doesn't mean that B likes A). A graph for that social network is illustrated by the figure on the right.
Then, you can calculate properties from you network by appying graph teory formulae such as number of edges, number of elements, in and out degrees and its averages, variances and density. If an actor (person, element, etc) has more outgoing relationships than the average, you can say that it is a communicative actor. If an actor receives more edges than the average, it's a popular actor. A table with basic elements to calculate follows below:
|
There is a complex theory about social networks and if you want to have a better idea what that’s about, check wikipedia, or even a more complete material from Robert A. Hanneman from the University of California, Riverside. If you are interested in this subject and want more material or any book suggestions, feel free to send me an e-mail and ask.
On the second part of the chapter I talk about virtual communities and online games, and how those games build huge communities, especially MMORPGs. With this, there is a discussion about the identity assumed by the player when he/she controls his/her avatar, the characters they are playing, the social impact on culture and some concepts involved with playing.
Then I describe the game I chose for my research, Ragnarök online, because it has a huge Brazilian community. Ragnarök online is a Korean game represented in Brazil by Level Up! Games. It was one of the first MMORPGs to build an official server here and there are about 200 thousand players today. In the end of this chapter, I talk about all the communication tools available in online games, specially Ragnarök.
The fourth chapter is about the methodology I used in my research. First I explain the questionnaire used to collect data and then the technical specifications, the Servlets used for the questionnaire system and the plug-in used for exporting data. For my research, first I made an online survey system in Java to collect data. After gathering data (the period was really short, I wish I could've collected more data...), I developed a plug-in in Java to export the answers from the MySQL database to the Pajek .net format. After all that was done I started the analysis.
On the next chapter I show the results with graphics and the analysis. I also show the graphs of the social networks that have been found on the online research with the analysis as well. The research resulted in two main social networks, one about the friendship in-game and another about highest level players known. I will talk about each one separately next.
First I will tell you about the other results I got. About the age of the players, 49% were between 10 to 15 years old and 35% were between 16 to 20 years old. The players were predominantly male, 84% of them! One curious thing was that the few female players were older. They were mostly from the Brazilian states of São Paulo and then Rio de Janeiro. Most of them were students (25% of the first Grade - 11 to 14 years old in Brazil, 39% from High School, 21% from College and 2% from Graduate Studies). Their characters were mostly from high levels (32% from 81-99 and 47% from 61-80). The majority played more than 25 hours per week at 18h to 24h (35%) and 12h to 18h (30%). They also never go to real life events (46%), but instead play at home (54%) and play PVP mode sometimes (31%) or rarely (29%). This indicates that this is a virtual community, it doesn't need physical contact, but needs human contact. The last element highlights the importance of the community for the game, 58% of the players say the best thing in the game is playing with friends. For more interesting data and analysis, download my thesis or contact me.
Friendship Social Network
This social network was generated using data from the online survey. Players told who were their 10 best friends in the game and how they first met each one of them. There were four options:
1. If they met the person because they had the same level and played together for some time (indicated in the graph by purple).
2. If they met because they had chosen the same in-game options as that friend, such as guild, class, profession, characteristics, etc (indicated in the graph by blue).
3. If they went to the same place inside the virtual world of the game (indicated in the graph by green).
4. If they were already friends outside the game (indicated in the graph by orange).
To visualize this network I exported the data with my own plug-ins and then put it into the Pajek software. For this visualization I used the Fruchterman Reingold 3D algorithm.
In this network there are 341 vertices and 319 edges. The entrance and exit degree is 0,9354838709 (less than 1), which confirms that many actors don’t have any relationship or only receive relationships (some of those that were indicated and didn’t participate in the survey). Most ties were from type 2 (135 ties), showing that the options the player makes for the character are important for the relationships inside the game.
In this social network 3 main sub-networks were found relating to the main guilds found in the survey.They were studied separately where popular and keys actors were found (if those were to be removed from the network, there would be a disconnection in the graph). The most popular actor was the one with entrance degree 5.
With Pajek, it’s possible to export the social network in 3D (VRML), allowing you to see and navigate around it using any browser. You can see my friendship social network in your browser, just download or open the file (really small, 225kb). For that, you need a VRML viewer, like Cortona (recommended) or Microsoft VRML Viewer, its simple to install, free and just a second to download (and really cool!!). See a picture of the exported network in 3D below:
For more detailed studies and data, obtain the entire thesis.
Highest Level Players Social Network
This social network was generated with the online survey. I asked the participants who were the players that they knew were level 99 in game (highest level), and they could indicate up to 10 players. Those players that are level 99 have been in the game for more time, played more, with more people, and because of that, know more people and are more likely to have a larger social network.
To visualize this network I exported the data with my own plug-ins and then put it into the Pajek software. For this visualization I used the Fruchterman Reingold 3D algorithm.
In this network there are 118 vertices and 72 edges. The entrance and exit degree is 0,6101694915 (less than 1), which confirms that many actors don’t have any relationship or only receive relationships (those who were indicated and didn’t participate on the survey). The degree is even lower than the one in the friendship social network because many players didn’t even know any 99 level players.
In this social network a main sub-network was found, because two 99 level players answered the survey and they were indicated by other players as known to possess the highest level. They were studied separately where popular and keys actors were found (if those were to be removed from the network, there would be a disconnection in the graph). The most popular actor was the one with entrance degree 5.
With Pajek, it’s possible to export the social network in 3D (VRML) to see and navigate around it using any browser. You can see my friendship social network in your browser, just download or open the file (really small, 54kb). For that, you need a VRML viewer, like Cortona (recommended) or Microsoft VRML Viewer, its simple to install, free and just a second to download (and really cool!!). See exported 3d picture below:
For more detailed studies and data, obtain the entire thesis.
The last chapter is about conclusions and future works. The first thing to conclude here is that online games are also a revolution to the culture and society. With social network analysis, you can understand the behaviour of your community and then, plan commercial actions, marketing campaigns, etc. To simplify this study and the analysis, a computational system can be developed and implemented, not only to gather data, but also to compute the analysis. The friendship network is more connected - it has more sub-networks (cliques). On this network, the clan, and character choices are the most important for the player’s relationships. With social networks and virtual communities, people are now breaking geographical barriers and building relationship based on their personal interests and not on localization. More conclusions and future work can be found on my thesis.
For my Msc degree I am working to gather data from World of Warcraft and build a mathematical model based on its social network to prove that a MMORPG community is an emergent complex system. More about it later. :)