WiNGS Lab @ UW

Link to paper: http://www.ece.ucdavis.edu/~chuah/paper/2008/imc08_facebook.pdf

1. In the facebook apps, this was defined as any action taken that involved the use of the app.
2. To find this information in PK, the data in the sql tables would be sufficient. Attacks, making items, building houses/flags, etc.
3. The table 'genericactions' seems to make a record most of these things in a single table.

1. In the facebook paper, this was used for gauging popularity, the studying of power-law over the apps, etc.
2. Subscribed users in PK: hard to determine for sure since not all the databases have been recording data since the game's release
3. We can find active users by first defining what an active user is, and then using the 'logins' and 'sessions' tables to identify those users.
4. It may be useful to determine the percentage of regularly active users within each platform (iPhone vs. iTouch vs. Android) or percentage within each speed bracket (3G wireless, etc.). However, we do not have specific network speeds for each unique user.
5. It is possible to make a connection to power-law like the paper. This will be something we may look into as research continues.

1. Very easy to get this, because almost all the data tables have a row specifically designated for regions.
2. Even Africa, Russia, and Latin America have people who play PK, so it should be interesting to analyze network information from these users.

1. The 'genericactions' table (and a couple others) record a 'return' command, which is probably the same as relocating. A latitude and longitude is recorded along with this.
2. If location is recorded after the relocation takes place, then we have the location of the user.
3. If location is recorded before, then we could use the next action of the user to get coordinates.
4. If all else fails, we could use the ip row in the 'sessions' table, but that might be a little unreliable.

1. Used in the facebook research paper to find out how much activity occurred on the apps during certain times of day and holidays.
2. For our purposes, it depends on how we define 'peak usage'. If there is a lot of communication between server and client when the character is not doing much, then we can define it as the number of users who are logged on at any given time. The 'sessions' table will give us this info.
3. If we define it as number of activities done, all activities are recorded with a time. Using that, it should be easy to devise an algorithm that will calculate peak usage.

1. The facebook paper reports data showing how friends using the app interact with each other.
2. In PK, it is possible to form a group with one or more members. There are six data tables representing info about groups.
3. If we wanted a more in-depth look, we would need activity tables to include a row saying whether the character was in a group or not during the time of the action.