Hi,all:

In the past two weeks, the major work is to mine user interest via observing
user behavior in the gpodder. I try to record top score keywords of one
episode which user chooses to download in the behavior model. Then use a
mining algorithm to pick up top-n keywords from this model to populate the
user profile.

I also tried to re-design the recommendation score algorithm. I want to use
new social tag feature in opencalais to score each episode.

I think until now, the basic work of recommendation has been done. In the
next phase, I would focus on the context-aware module which will not only be
in charge of user-awareness but also be aware of device and environment
context. I will try to design a framework to implement basic functions and
then other developers can easily use this framework to add context-aware
funtions into their  applications.

In the next two weeks, I will do some paper works. I will document the new
score algorithm and the framework. At the same time , I will fix some bugs
in current code then release a new version of gpodder , then get some
feedbacks from the community.

If you are interested in my project, then you can find much more information
on http://garage.maemo.org/projects/newssprite. You can also check out the
code on the svn repository.

Any comments and suggestions are welcome.Thanks.



-- 
--Feng GAO
--Maemo-GSoC09
_______________________________________________
maemo-developers mailing list
[email protected]
https://lists.maemo.org/mailman/listinfo/maemo-developers

Reply via email to