Collaborative filtering?

[Dan Bolser]

On 28 April 2011 02:59, Aldrin Montana <drinm...@gmail.com> wrote:
> Dan,
>
> Your suggestion sounds like it yields itself better to collaborative
> filtering.

Collaborative filtering?

> and a machine learning toolbox? sounds cool @___@
>
> -Aldrin
>
> On Wed, Apr 27, 2011 at 3:58 PM, Rubens Camargo <rbnsc...@gmail.com>
> wrote:
>>
>> Hello Kalyan,
>>   Shogun is a machine learning toolbox. Maybe you wanna see it:
>>   http://www.shogun-toolbox.org/
>>
>> Cheers,
>>   Rubens
>>
>> 2011/4/27 Dan Bolser <dan.b...@gmail.com>
>>>
>>> Hi kalyan,
>>>
>>> Thinking about MACHINE LEARNING, which is surely in demand, here is a
>>> random suggestion (I'm sure others can come up with better ones):
>>>
>>> Why not develop an extension [1] for MediaWiki [2] that reads the
>>> users 'watchlist' [3] and provides links to pages that the user may
>>> also want to watch? Perhaps this inference could be done using GRAPH
>>> THEORY?
>>>
>>> Actually, MediaWiki is PHP, which you don't list... However, you could
>>> do something similar via a 3rd party server set up to query the
>>> Wikipedia API [4-6]! That would be really cool :-D
>>>
>>> Thoughts?
>>>
>>> [1] http://www.mediawiki.org/wiki/Developing_extensions
>>> [2] http://www.mediawiki.org/wiki/MediaWiki
>>> [3] http://www.mediawiki.org/wiki/Manual:Watchlist
>>>
>>> [4] http://www.wikipedia.org
>>> [5] http://www.mediawiki.org/wiki/API
>>> [6] http://en.wikipedia.org/w/api.php
>>>
>>> On 27 April 2011 14:18, kalyan babu <kalyanba...@gmail.com> wrote:
>>> > HI.
>>> >     I am a 3rd year student and my skills are C,VB,ORACLE,GRAPH
>>> > THEORY,MACHINE LEARNING.I want  to work on a open source project ,can
>>> > anyone
>>> > suggest some
>>> > On Wed, Apr 27, 2011 at 4:24 PM, Dan Bolser <dan.b...@gmail.com>
>>> > wrote:
>>> >>
>>> >> Hmm.. I don't see that message there. Have you got a link?
>>> >>
>>> >>
>>> >> On 27 April 2011 11:43, kalyan babu <kalyanba...@gmail.com>
>>> >> wrote:
>>> >> > Yup.....
>>> >> >
>>> >> > On Wed, Apr 27, 2011 at 1:58 PM, Dan Bolser <dan.b...@gmail.com>
>>> >> > wrote:
>>> >> >>
>>> >> >> Did you ask this on GSoD?
>>> >> >>
>>> >> >> On Apr 27, 2:17 am, kalyan babu <kalyanbabuiit...@gmail.com> wrote:
>>> >> >> > HI.
>>> >> >> >     I am a 3rd year student and my skills are C,VB,ORACLE,GRAPH
>>> >> >> > THEORY,MACHINE LEARNING.I want  to work on a open source project
>>> >> >> > ,can
>>> >> >> > anyone
>>> >> >> > suggest some
>>> >> >> >
>>> >> >> >
>>> >> >> >
>>> >> >> >
>>> >> >> >
>>> >> >> >
>>> >> >> >
>>> >> >> > On Wed, Apr 27, 2011 at 6:24 AM, Dan Bolser
>>> >> >> > <dan.bol...@gmail.com>
>>> >> >> > wrote:
>>> >> >> > > In the interest of openness, I should point out that there are
>>> >> >> > > other
>>> >> >> > > options:
>>> >> >> >
>>> >> >> >
>>> >> >> > >
>>> >> >> > > >
>>> >> >> > > > > >http://blog.lydiapintscher.de/2011/04/25/not-accepted-into-gsoc-heres...
>>> >> >> >
>>> >> >> > > What else have people found to do instead of GSoC?
>>> >> >> >
>>> >> >> > > On Apr 26, 7:42 pm, Carol Smith <car...@google.com> wrote:
>>> >> >> > > > Hi everyone,
>>> >> >> >
>>> >> >> > > > On Tue, Apr 26, 2011 at 11:39 AM, William Holbrook <
>>> >> >> >
>>> >> >> > > > holbrook.wmich...@gmail.com> wrote:
>>> >> >> > > > > Awesome idea -- I'll check it out!
>>> >> >> >
>>> >> >> > > > We don't need everyone on this mailing list replying about
>>> >> >> > > > whether
>>> >> >> > > > they
>>> >> >> > > plan
>>> >> >> > > > to join the group or not. Please don't spam the mailing list
>>> >> >> > > > any
>>> >> >> > > > further
>>> >> >> > > > with these emails.
>>> >> >> >
>>> >> >> > > > Cheers,
>>> >> >> > > > Carol
>>> >> >> >
>>> >> >> > > --
>>> >> >> > > You received this message because you are subscribed to the
>>> >> >> > > Google
>>> >> >> > > Groups
>>> >> >> > > "Google Summer of Code Discuss" group.
>>> >> >> > > To post to this group, send email to
>>> >> >> > > google-summer-...@googlegroups.com.
>>> >> >> > > To unsubscribe from this group, send email to
>>> >> >> > > google-summer-of-cod...@googlegroups.com.
>>> >> >> > > For more options, visit this group at
>>> >> >> >
>>> >> >> > > >http://groups.google.com/group/google-summer-of-code-discuss?hl=en.
>>> >> >
>>> >
>>> >
>>
>
>

[Aldrin Montana]

Everybody's "watchlists" are represented as a vector of weights or values of some sort. Based on how one user's watchlist compares to another, you can make suggestions to one user using an item in the other user's watchlist or suggest an item based on the items in the former user's watchlist.

The basic idea is that you can construct a contextual understanding of what one user may or may not like based on what other users do or do not like that have similar watchlists and similar ratings of items that are in both users' watchlists.

For a good example think of Amazon's "Other users that looked at this item also looked at these items:".

Not sure if that ended up making sense, it's hard to be general and concise using just text.

-Aldrin

[Dan Bolser]

On 28 April 2011 14:52, Aldrin Montana <drinm...@gmail.com> wrote:
> Everybody's "watchlists" are represented as a vector of weights or values of
> some sort. Based on how one user's watchlist compares to another, you can
> make suggestions to one user using an item in the other user's watchlist or
> suggest an item based on the items in the former user's watchlist.
> The basic idea is that you can construct a contextual understanding of what
> one user may or may not like based on what other users do or do not like
> that have similar watchlists and similar ratings of items that are in both
> users' watchlists.
> For a good example think of Amazon's "Other users that looked at this item
> also looked at these items:".
> Not sure if that ended up making sense, it's hard to be general and concise
> using just text.

Yup, makes sense. Unfortunately I just found out [1] that Wikipedia
watchlists are private :-(

Can't think of a way round that unfortunately.

Cheers,
Dan.

[1] irc://irc.freenode.org/#mediawiki