New Posting Features - Automatic Thread Tagger

Automatic Thread Tagger


Description
When a user submits a new thread this modification will automatically take keywords from the thread title and use these as tags. You can use Automatic Thread Tagger to propose the user AJAX tags for his new thread, or it assigns new tags after saving the new thread. It can add the translated thread prefix to the tags.
Additionally, you can tag existing threads via maintenance and also scheduled tasks.

This modification is a successor to the terminated Automatic Thread Tagger by MrEyes:
https://vborg.vbsupport.ru/showthread.php?t=179927

As an example, if a user submits a thread with a title of:
"Fish Food for Cats!"

The thread will be automatically tagged with:

- Fish
- Food
- Cats

If the user also submits an actual tag of "Fish" this will not be duplicated. Any rules you have setup for tagging will be respected.
If you choose to do so this product will also automatically tag threads created by incoming RSS feeds.

Demo
I cannot show you the process of creation, but here is a list of tags generated by Auto Thread Tagger:
http://www.insideearth.net/tags.php?langid=5
http://www.insidesupcom.de/tags.php?langid=1


Automatic Tagging of existing threads
You can tag existing threads via maintenance or scheduled task/cron. They will be created with a special flag so they can be easily identified and deleted. Manual assigned tags are not touched. Maintenance is also working if Automatic Tagging is disabled via settings. Great if you want to test some settings. Automatic Tagging will take the date of the thread creation and also the userid of the creator. This process can be automated by running a scheduled job once a night.

Please keep in mind that tags that were proposed via AJAX are not tagged as auto tagged and therefore cannot be identified as such (and therefore not deleted automatically). If you want to retain the auto tagged flag you should disable AJAX and enable the tagging after the thread has been saved. As an alternative way you can also disable this and let new threads be tagged in the night from the scheduled job.


Installation / Upgrade
1. Upload all files from "upload" to your server, take care of the directory structure
2. Import "product-auto_thread_tagger110.xml" as a product, overwrite if it's already installed
3. Check settings
4. Run maintenance / Auto Tag Threads to tag existing threads (needed if you want to use the cron)

After install, and by default the modification is disabled, this will allow you to play around with configuration before switching it on.


Troubleshooting
If you report a bug please post the thread title that created it, without this I cannot test it and improve the language parsers.

* If no threads are tagged you will have to check the following:
- Is the modification enabled? Is the action you are testing enabled? (vBulletin tagging, whole auto thread tagger system, AJAX, new threads)
- Are the words you are using badwords or filtered out?

* Cron/Scheduled Task is not tagging all threads.
- The cron is limited to 500 (you can change this via settings) threads per run to avoid heavy impact on server. Make sure you run maintenance auto tagger before this to tag old threads. You can check the scheduled tasks log to see if it is running correctly.
Important: If a thread title does not meet minimum requirements to be included in tags (f.e. one word thread titles, too short words), it will be forever in this queue.

* I'm using polish, arabic, turkish, etc.. language and the tagger is not working like it should.
- If not already replaced, replace the filter replacement '&'=>'and' with ' & '=>'and' (a space before and after &)



Todo
What comes next? You decide. Tell me what you are missing and I'll look if it can be integrated.


Why thread title and not thread text?
Parsing the thread text for tags is an extremely unlikely addition as this would require some fairly heavy processing to ensure quality of tags.


What are Stopwords?
Stopwords is the name given to words which are filtered out prior to processing of tags.
The user Hostboard on vBulletin.org posted some resources regardings this:
https://vborg.vbsupport.ru/showpost....&postcount=380



History
1.2.0, 9th August 2008
- Fixed error with missing threadid's
- Fixed error with AJAX and prefix
- Fixed error with not indexing tags via cron
- Added polish, spain, english stopwords
- Compatibel with vBulletin 3.8

[Phalynx]

Originally Posted by Kuntz

All of you are suggested to use this as it fixes two bugs in the CRON file. The first bug is in relation to RSS threads not being tagged at all. This was due to the fact the creator used incorrect syntax to get NULL fields, so no fields would get returned. The second bug prevented threads from being tagged if they were created in a 'wierd' method. What is that? I'm not sure, but I had about 20 threads that weren't tagged but were marked as being tagged, the datelines were spread apart pretty evenly, so I'm assuming the creators of the topics created the threads, they had tags, then those creators for some reason removed their tags manually afterwards. This leaves the thread marked Auto-Tagged, but there are no tags. So now the cron will re-tag these threads. This is all a guess, but who cares, all your tags are belong to me.

For the first part of the NULL I agree, thanks for pointing me to this. It seems it works on some MySQL versions and on some not.

Also there is no mark for already tagged threads. But there is a mark for threads tha cannot be tagged, f.e. because of title containing only bad words or just too short words. This is an attempt to not try to tag these threads every night.
The autoskip option is going to reset to 0 by editing the title.

I will incorporate the changes to the official release if someone can confirm that cron is now working.

[Kuntz]

Erm, according to my DB, all my threads are 'autoskip = 2' once they're tagged by the CRON Auto-Tagger.

There are two kinds of untagged threads in my DB:

-Threads with NULL as the tag list
-Threads with a blank tag list (this is different then NULL)

-Threads set to NULL have an autoskip value of 0
-Threads that are blank have an autoskip value of 2

For my DB and installation, the above CRON works perfectly fine, and on top of that, tags the RSS feed as it comes in. My two RSS feeds are set to check every 30 minutes, I then edited my CRON to run the RSS script at '00' and '30', and edited the autotag CRON to run at '00' and '30' as well. Then I 'reset' my two RSS feeds and when the clock hit the top of the hour, it checked my RSS feeds, added the topics, and tagged them, all in one go! Been working great all day so far.

[Phalynx]

Check line 68-76 in functions_autotagger.php

[tekguru]

So have the Cron, etc fixes gone ino this Mod yet? Working out whether to install now or later.

[Ploz]

I think it would be useful to add an option to show the text of the first post below the thread title in the "Threads Tagged with" listings...

[Kuntz]

Originally Posted by Phalynx

Check line 68-76 in functions_autotagger.php

Some of my threads are set to '2', and some are set to '0'. It seems pretty random. =\

Maybe don't do !taglist OR !isarray() and instead just do !isarray() ??

[mcyates]

I keep getting this error when uploading the product

XML Error: Invalid character at Line 42

[Mazinger]

Tag RSS Feeds?
If yes the auto tagger will also attempt to tag incoming RSS feeds.

In order for RSS tagging to work you will have to manually edit one of the default vbulletin files. Please refer to the product page on vBulletin.org for further information.

Any help with that?

[Kuntz]

Originally Posted by Mazinger

Any help with that?

outdated info

[Hostboard]

Line item #5 in 1.1.0 GOLD, 31th July 2008 release notes addressed this.