Sounds very cool, Jeroen.

This is lovely and works a treat.

Is there any way to block certain words being tagged? For example if I were running a Michael Jackson site, I wouldn't want 'Michael' or 'Jackson' tagged each time as all content will be related to this subject.

Heh - forget that last comment - found the file

autotag/vendors/php_kea/languages/en if anyone else is wondering too...

Great plugin - really good!

@Jeroen - Genius. Thanks! ...Where do I find information on Ukrainian and Russian (basically Cyrillic) stemmer algorithms, for your plugin?

A step in the right direction. This is what I was doing for a working search function, and what has to be in the background in the core.

Especially groups and user full names should be extracted into tags too. This will make users searchable by first and last names, make groups searchable by name. I also consider using a filter to convert unicode characters to lower latin equivalents to search letters like 'ü' using 'u', then filter the search form in the same way during a search. And I consider using a hidden seperate tags field, so people can enter any tags without getting confused with auto tags, or pad user tags like "utag:.....", so they can be shown / searched in different steps. Padding auto tags would require more space, as tagging is not a habbit of a casual user.

Ah, and the filter should also filter out 1 and maybe 2 letter words in extracted tags, and in search terms.

Interesting reading:

http://snowball.tartarus.org/texts/introduction.html

http://snowball.tartarus.org/algorithms/russian/stemmer.html

------

http://cpan.uwinnipeg.ca/dist/Lingua-DetectCyrillic

@UkrProgrammer the stemmers we used are based on the snowball algorithms. If you find a good algorithm for a new language that is written in PHP (because those are hard to find :), we can include it in the autotag. You can also try to include it yourself. The keaphp class autodetects available language stemmers and stopword files.

Very cool! Thanks, Jeroen! Genius!

Hi Jeroen,

Thank you, working good so far, only had an issue with cloned tags. I solved it by changin the plugin ordering, moved it on top of the plugins list, and then it worked nicely.

Regards,
Uddhava dasa

Really great plugin Jeroen as always

Great plugin! I have yet to have auto tag work on external pages. Have I mis-read something?

http://elggtest.ensci.us/keywords/

has example that uses stopwords data and also special keywords data to extract KWIC indexes

If anyone wants the code.. just ask and I'll zip the php code for you to download

wow, really nice plugin :) My master thesis was partly about this, this is a very interresting subject :)

zip it up please...Dhrup

I'd like to point out that Jeroen's plugin could form the basis of a multi-lingual tagging system. Faviki has done this by

1) autotranslating content to English via Google,

2) automatically generating a list of possible tags

3) symantically linking tags to a concept database like wikipedia

4) then returning the result translated back into the original language.

The benefit is that content in any language can be tagged and searched by readers of any other langauge.

@chiinook

hmm doesn't sound very feasible to me.. I think finding results in different languages would find an extremely rare usage, especially in a social network.

A complex plugin that works with no conflicts. Amazing.

@jeroen and ColdTrick once again...  FANTASTIC!

Thx!

Hi, Jeroen! Again, great plug-in!

I was wondering if it's possible to autotag user-created forms (ie, using Kevin's plug-in). Any thoughts? Anyone try it?

does this work with 1.6 ?

@jeroen

can this work for groups also ?

i dont really understand how this plugin works. Is it supose to show tags somewhere on my site ?

seems to work just fine on 1.6

How do I stop it from tagging words like "the" and "nbsp"?  These are the only two auto-tags I got in my testing...