cept
The MMGS sends events about user actions to
the Game Usage Miner (GUM), if the GUM
discovers patterns of interest (e.g. player lo-
goff probability) they get passed on to Recom-
mender Service (RS). When the RS receives a
pattern it will give a recommendation back to
the MMGS about how to alter the player ex-
perience. The purpose of this is to improve the
gaming experience for players. Methods used
in the RS can e.g. be collaborative filtering
or case-based reasoning. Another purpose of
the GUM is to provide realtime metrics of the
game, typically macro numbers such as mood
of the community (e.g. number of players likely
to log off in the near future).
ing in non-discrete game-state massive multi-
player games with information gathering from
web browsing. The purpose of the information
web usage mining, respectively.
(i.e. common or extended log format) created
by the web server.
that supports game usage mining in massive
multiplayer games.
of user's active period, i.e. before browsing to
another web site or stop surfing.
it is usually very simple to determine t
instead of idling, or a least be logged out au-
tomatically after a particular idle time of the
client (in order to avoid high bills).
caused by timeout for a period t
purely event driven. When the client (browser)
requests a web page or an object it causes a
log entry line (LogEntry
is usually limited to whole seconds, so if the
web server receives several requests in a second,
they are written to the web log in an arbitrary
order.
real-time actions in the game (e.g driving a car,