Can tags be pulled from titles?

A developer can write his/her own query and push tht trough the tagadelic API.

Same exact question. At first I thought it was my mistake and that I needed to retag my nodes, but then I understood It's "by design".
As I don't possess the power of the writing snippets with the Drupal Api do you think this exists already as some kind of another module additional to tagadelic or can we have a workaround maybe by making a new vocabulary with tags that come from the titles ? (I'm just saying I have no clue:( )
Otherwise, can we have such an option in next release?

You might have a look at the Keywords module. Seems to provide at least a similar functionality:

Module can save node's keywords as taxonomy terms to selected vocabulary. Later you can use this vocabulary to tell Meta Tags module use this terms for meta-tag "keywords" or use as usual tags.

Frank

Setting this back to "normal" as it might be beyond the scope of this module ;-)

Thanks for that reply. It kinda work although I am still wrestling with it. I am describing what I did in case someone else has the same question.

1. you install and activate Tagadelic (it adds couple of Blocks in your Blocks page you need to activate'em yourself but wait for that)

2. you install and activate Keywords . You make a NEW Vocabulary that you name.. "Tags" or something similar.

3. you go and setup Keywords so that with each CRON run it scans your nodes for keywords used often and makes em a tag and drops them to your new Vocabulary you just made (Tags).

4. you run cron 2-3 times with the setting to scan like 10 or 20 nodes at each cron run. Not more, it's slow.

5. you go to the blocks page, publish the "Tags to Tags" block or whatever its called.

So just to make it FOOL-PROOF 123, what you're doing is: USING TAGALICIOUS TO PRESENT A CLOUD MADE UP FROM A SPECIAL VOCABULARY YOU MADE WITH THE HELP OF KEYWORDS WHICH ADDED ALL THE TERMS IN IT.

( My problem is that I have shitty words since I have a greek website and I need to re-fix my stopwords )

OH BTW! IF YOU HAPPEN TO SCAN 20-40 NODES AND YOU'RE NOT SATISFIED WITH THE QUALITY OF THE TAGS, MAYBE YOU SHOULD GO FIND THE STOP WORDS FILE (IN THE MODULE FOLDER) , EDIT IT, ADD THE WORDS YOU WANT , DISABLE->UNINSTALL->DELETE THE KEYWORDS MODULE AND THEN REINSTALL IT SO YOU HAVE NEW FRESH KEYWORDS.

Here are some stop word lists for different languages: http://www.ranks.nl/resources/stopwords.html

unfortunately there is no list for Greek stop-words there . :(
Thanks anyway

I am afraid the "Keywords" workaround is kindda buggy. Reason is that building a Vocabulary of Tags ends up with an enormous vocabulary (I have many nodes in my portal since it's an ads site).
I think it would be 100 times more practical of Tagalicious BY DESIGN took tags out from titles. Huge save of time, database size, speed, etc.

ps: I am trying to delete this 'Tags' vocabulary right now and my query is still going after 3 minutes ...

Tagadelic is supposed to pull its tags from vocabularies. So IMO there are two possible solutions for your problem:

1. Restrict the "Keywords" module so it only scans the title of nodes but not the body.
2. Create your own module which automatically creates vocabulary items from all node titles.

Either way, this is not (and should not be) the direct responsibility of the Tagadelic module itself.

hth
Frank

I don't think this is a discussion on responsibilities, Frank, but mostly a question of whether we can improve community modules with features everyone seems to be asking about. We need to understand the needs before reaching into conclusions.

After some days of testing, I confirm that the 'Keywords' workaround is not the best one.
'Keywords' also overwrites custom tags you try to add, and replaces them with the one it thinks its best. (with the available settings, the one which is more frequently used) (usually a stop word or a word that doesn't represent the context of the node).

By trying to describe our problem and what we need I'd say:
An algorithm that scans ALL nodes (it's titles - or bodies) , then saves the words that are more frequently used not inside ONE node (as keyword is designed to do) but across all of them, and then this can produce a useful Vocabulary that Tagadelic can cloud.

Maybe this is mostly a thing for "Keywords" to do than Tagadelic, I admit. But it seems Keywords is kinda abandoned as a project,

What about taking over the Keywords module as a (co-) maintainer and enhance it in the way you described? That would make a very useful module indeed!

Frank

I would do it but I'll need couple of years of PHP&SQL studying in order to do that I'm afraid :/

Will have a closer look at "Keywords" and see what I can do.

EDIT
AFAICS the Keywords module already provides most of the desired functionality. What's missing is some configurability, e.g. restricting node types (#568730: Choose content types) or adding stop word lists for other languages (#648180: Make stop-word lists configurable).

EDIT 2:
I posted a quick hack over at #648730: Cleaned up module code + updated .pot file which lets you make the module only scan titles.

Frank

Thank you for this contribution Frank. I hope Keywords maintainers add these fixes to the head asap.
In addition to the title-only scan patch, my idea was to save terms in the Vocabulary only if they were encountered over a number a times (we define how many) in the total of the node titles.
This would help avoid saving all words from all titles in a Vocabulary in order to present'em afterwards with Tagadelic

Hi George,

In the Keywords module you can set

Minimum number of times keywords used:

How many times keyword should be used in text to be saved as term in specified vocabulary?
Recomended: 2 times.

But that might be on a per node basis. Counting over all the nodes is difficult, as the Keywords module only scans a certain number of nodes with every cron run. So it won't capture all nodes in one go.

Frank

PS:
The Keywords issue queue might be a more proper place to continue this discussion ;-)

Yeap you might be right about the issue queue :)
Actually I'm tempted to get into this and make a small module myself, about time!

peace

closing this. Not a Tagadelic issue.