Talk:WMDE contract offers/Rewrite CatScan

Integration into MediaWiki

The integration of CatScan into MediaWiki is important. The potential users for CatScan aren't just editors (who are used to the toolserver), but mostly readers, who are likely to be confused if they visit a page on toolserver.org ("is this still Wikipedia? It doesn't look like Wikipedia!").

It'd be nice to be able to integrate CatScan directly into MediaWiki, as an extension. The extension could be just a front-end for an API running on the toolserver. This would mean no loss in performance, as all the actual work is still done by the toolserver.

If there's a front-end in MediaWiki, there are more ways to use CatScan. Instead of providing just a link to an external CatScan tool, there could be:

• A special page where you can select categories and filters (same as current tool page, but with more familiar to the reader than an external page)
  • If Manual:$wgEnableMWSuggest is enabled, find categories more easily.
• "Find similar pages" on each page, which just leads to the special page, but all of the categories in the article are pre-selected (you don't have to copy&paste all category titles).
• More useful features I haven't thought of yet :)

I believe that CatScan could be really useful for readers, if there is a proper integration into MediaWiki and the user interface. I know that this probably conflicts with the specifications of this contract offer, but imho it should be considered nevertheless. --Church of emacs 10:53, 5 April 2009 (UTC)

Integrating into MediaWiki would of course be very cool. In theory. In practice I do not see how it could become fesible, the queries are simply too heavy for live use. It would even be a püroblem if the tool would become too popular.

A complete rewrite based on Lucene could make at least some common types of quries much faster. I would very much like someone to experiment with this.

Full integration of this into mediawiki would more or less equal the adoption of SAemantic MediaWiki. Which is also a cool tool, but I do not expect it to ever scale to the size of Wikipedia. -- Duesentrieb ⇌ 13:28, 16 April 2009 (UTC)

Features

• At least the functionality (but not the implementation, which is interesting too) shown here, and all the functionality of the existing catscan should be available with the new catscan extension.

Extension:DynamicPageList

Extension:Simple Forms

http://semeb.com/dpldemo/index.php?title=DPL_Simple_Forms_Test

• It should be possible to integrate it into every wiki, not only into wikip... ones, there should be no implicit connection to the toolserver.
• I like catscan and I'm using it quite often, but sometimes I've the need to intersect more then two categories.
• Sometimes I although need Cat1 and cat2 but not cat3
• It should be possible to fill a page with the infornmation by putting the selection statement on top of the page and saving it. Every call to this page will give you the actual results automatically.

Thanks --Joergens.mi 05:33, 8 April 2009 (UTC)

I fulley agree, except for the last bit. That is simply not feasable. Deep category intersection is expensive and must be cached. -- Duesentrieb ⇌ 13:25, 16 April 2009 (UTC)

Rewrite CatScan, who started?

Did someone start to do this job? Нирваньчик 12:57, 26 October 2011 (UTC)