Wikimedia Bugzilla is closed!

I ask to read Bug 2320 / Google sitemap support
(http://bugzilla.wikimedia.org/show_bug.cgi?id=2320) before reading this feature
request. There are remarkable similarities.

Google has released a project called "google Co-op" which currently consists of
two parts. I will focus on the second part in this feature request which is
called "subscribed links".

The main documentation of Subscribed Links can be found at
http://www.google.com/coop/docs/guide_subscribed_links.html

The very basic idea is to define a list of queries into a search engine and then
to give a predefined response.

Here is an example:

http://meta.wikimedia.org/wiki/Image:Demo_simple_bugzilla.png

is the direct result of 

<Results>
<ResultSpec id="Bugzillademo">
  <Query>Suda</Query>
  <Response>
    <Output name="title">Article Suda on Wikipedia</Output>
    <Output name="more_url">en.wikipedia.org/wiki/Suda</Output>
    <Output name="text1">The Suda is a massive 10th century Byzantine Greek
historical encyclopaedia</Output>
    <Output name="text2">of the ancient Mediterranean world. The derivation is
from the Latin suda</Output>
    <Output name="text3">meaning fortress or stronghold.</Output>

  </Response>
</ResultSpec>
</Results>

If you read the documentation of google Co-op and the Page on meta, you will
find a number of more interesting applications, especially when it comes to
metadata. De.wp has currently about 100.000 "persondata" entries.

The pattern query search allows it to have a search pattern like "When did
[name] die?" and actually to form a response in top of the results page, along
with a link to the relevant page.

However, this feature request aims to the very basic situation:

* For every lemma a single static response, formed from the first 160 characters
of plain text of a given article. The resulting XML file should be made
automatically every timeframe, uploaded on download.wikimedia.org, next to the
sitemaps file. 

Size limitations won't be an issue and these results are first shown to people
'subscribing' to a channel (http://www.google.com/coop/directory). Later, these
kind of results can be shown elsewhere, the same way, google already uses the
onebox.

A more verbose descriptions of the applications of google Co-op with wikipedia
data is shown on http://meta.wikimedia.org/wiki/Google_Co-op.