Meta Tags Hack in the works

Posted: Mon Feb 02, 2009 7:32 am

The truth about this mod is you will never be able to please everyone and there will always be someone wanting more from it.

However, if I was heading this project up I would play it like this.

Use a combination of both techniques. Allow users to enter specific keywords where they want them like articles. Then use dynamic titles or something like to gather the page contents and build a keyword list based on occurrence. (This step sounds complicated but it really is not)

Now that we have two arrays, compare them and strip out duplicates. So now we have the best of both worlds. It does sound somewhat complicated, but seriously the hardest part is modding the modules to allow chosen keyword insertion.

registered · Site Admin Joined: Jun 04, 2004 Posts: 6437

floppydrivez, if we give flexibility to admins, everyone could be pleased. First, default the tags from the article content and give authors / admins the ability to override it. It doesn't have to impact the module, as long as the meta tags (overrides) are maintained after the content is saved.

In that scenario, an external function could maintain the tags without modifying the content, but it needs to know the content ID. Modifying the module gives you the ability to adds the tags when adding the content. That's preferred, of course, but most could use an addon that allows the tags to be specified after saving the content. This gives you the functionality more quickly without having to modify the modules.

I'm almost finished updating my site to RavenNuke (finally!) after which I plan to focus on changes like these.

Posted: Mon Feb 02, 2009 10:16 am

Guardian2003, totally agree with you on this one.

Posted: Mon Feb 02, 2009 10:43 am

Agreed. The logic should be:
1. Check for overrides
2. If no overrides, generate META tags based on rules for content / page, module, site.

The trick is defining the rules - e.g. if there aren't enough keywords generated for a specific content page, should the module-level keywords (if they exist) be added to those generated for the content? If module-level keywords don't exist, should they be generated from categories or other groupings (e.g. forums, topics, etc.) within the module? What tags should be used on a category-level page? How should they be generated?

Posted: Mon Feb 02, 2009 11:50 am

I agree with the logic you posted kguske and also that the key is in the 'rules'.
Most of the JS library extensions and Classes I have looked at allow you to;
determine the maximum of single words
maximum number of key phrases i.e. where two or more consecutive words are repeated in the page content.
The one area these all fail in is that they focus almost exclusively on meta keywords and completely ignore other important meta data like 'title' and 'description'.

Unfortunately *nuke doesn't make it easy. If it was consistent with it's use of variables throughout it's core modules e.g. $category, $category_title, $content, $content_title it would certainly make converting the data globally into meta data much, much easier.

Posted: Mon Feb 02, 2009 12:32 pm

Aren't those JS extensions for displaying tag clouds?

Quote:

Unfortunately *nuke doesn't make it easy. If it was consistent with it's use of variables throughout it's core modules e.g. $category, $category_title, $content, $content_title it would certainly make converting the data globally into meta data much, much easier.

That's what the content classes are for - to make it consistent... Smile

Posted: Mon Feb 02, 2009 12:57 pm

Yes the JS ones are basically for making tag clouds and most of them use CSS for the placement of the tag data. Most of the PHP classes do something similar but there are a couple that are more useful in terms of storing the dynamically generated tags.

Posted: Mon Feb 02, 2009 1:18 pm

I think we are looking at the same scripts... Smile

Involved Joined: Sep 15, 2008 Posts: 352

I disagree it's the webmaster(s) that's in control of SEO not a user so user inputted Meta info is not what I would do. But just my opinion.

The method I explained has been live since on our site since 2003 and we have pretty good search engine placement. We reply on SEO alot since we are a full blown online retail store and good SEO means sales for us.

I would like to keep it simple: use the title for keywords separated by comma's and the hometext for the description limiting it 255 characters. Also have the meta info on the backend so the webmaster can add to improve etc....

I have my Meta hack complete for the news module I just haven't released it yet :>)

Posted: Mon Feb 02, 2009 1:39 pm

Alien73...I'm a little confused.... If user inputted meta isn't useful (even as a suggestion for the admin), what would it be important to use on creation of the content?

Using title for keywords - may not have sufficient keywords - why not use function as Guardian described that determines the most frequently used words (single and two word phrases) for the entire content? Of course, admins would still be able to override.

Using hometext for description - agreed - for news. What about other content types?

Let's confirm the table design and develop the retrieval method that uses that table design, then we can release your news hack without causing conversion problems / conflicts with a tool that supports all content types. As I previously mentioned, I hope to turn up the heat on development of this quickly as it's been too long and this is something many want, based on feedback here.

Posted: Mon Feb 02, 2009 2:03 pm

My method does everything automatically then you have the choice to edit. When you add content using a keyword heavy title is always the best bet anyways. The automatic way would use the Request URI so meta tags are created in real time grabbing the id from the mysql table then showing the title and description. I believe this is faster and editing the modules wouldn't even be necessary. My current hack uses both methods.

I do like what Guardian's idea. "most frequently used words"

Posted: Mon Feb 02, 2009 3:42 pm

Thanks, Alien73. That was my plan, too, and what I started on last year. So let's:
- default / generate the tags based on rules (title = content tile, description = first 255 characters of content, keywords = top keywords and key phrases from content)
- store them in an override table when they are first generated (so we don't keep generating them every time they're displayed)
- provide a facility for an admin to generate the defaults and maintain the overrides (it might be easier to do this from a block when displaying the page, but we could also use an enhanced sitemap for this purpose)
- modify the modules to allow addition / maintenance with content

I'll update the code and send you a preview to review (hopefully, I can get a day off this week to focus on it - probably Thursday).

Posted: Mon Feb 02, 2009 3:51 pm

Darn I forgot one very essential ingredient because my head is full of cold.....
There *must* be some facility to accommodate 'stop words' - for example when using the dynamic approach of building the meta data, words like

Quote:

a about an are as at be by com de en for from how what where when........

should automatically be skipped.

Posted: Mon Feb 02, 2009 4:25 pm

Yep. Got that covered.

Posted: Mon Feb 02, 2009 4:58 pm

Sounds good...

Posted: Mon Feb 02, 2009 9:11 pm

kguske wrote:

Yep. Got that covered.

is it language aware?

Posted: Mon Feb 02, 2009 9:15 pm

Don't remember, but I think not. You should be able to configure the stop words in whatever languages your site supports.

Posted: Mon Feb 02, 2009 9:43 pm

Good for me if at least is editable...

Posted: Mon Feb 02, 2009 9:49 pm

It might not be too hard to default from a language file, but I'd think you should be able to override it as your site might use some words that aren't common that you wouldn't want to make keywords.

Posted: Thu Feb 05, 2009 11:53 pm

I didn't get to a point where I can release test code yet, but I still made major progress on this today. The overrides are working beautifully, dynamic generation of the keywords with stop word removal...it's starting to take shape. A few more issues to resolve on the output side, then I start working on the maintenance functions.

Exciting!

Posted: Fri Feb 06, 2009 8:03 am

Exciting indeed. Certainly brightened up my day Smile

Posted: Fri Feb 06, 2009 8:27 am

uhm... im eager what you got under the hook.

Posted: Fri Feb 06, 2009 9:01 am

Im totally ready to see it, have some questions though... Reading over this and past posts has been confusing. Will this one replace the dynamic titles as well? If so, is there a way it could be set to not use the title from the module thats in home as the title for home, if that makes sense? Also, with the other dynamic mods, any chance of collaborating and creating one single module? Thanks

Posted: Fri Feb 06, 2009 9:07 am

Sure... when it's all said and done it will be a powerhouse of meta and titles!

Posted: Fri Feb 06, 2009 9:19 am

Yes, it will replace dynamic titles (though, in some ways it has evolved from that).

Yes, it will use the site title, description and meta tags for the home module on the index page.

Yes, collaboration is always welcome. The part I'm working on supports dynamic HEAD tags (title, description and keywords) that can be generated from the content and overridden at multiple levels. However, overriding at the category, subcategory and content levels requires that the content already exist (i.e. there is no way to specify the overrides when the content is created). Enabling the overrides to be generated when the content is created requires modifications to the modules. There is no reason the modules can't use the same functions for generating the overrides when creating the content, and allowing the admin to override on submission / approval.

alien73 is working on this for the News module, but other modules (namely Content) would also need this. Since jestrella developed the Content Plus module that will eventually become the standard for RavenNuke, I suspect he will want to do that for that module. But there are other modules that would also benefit from being able to specify the HEAD title, description and keywords (tags) on content creation (and, for that matter, being able to edit them with the content): calendar, reviews, encyclopedia, ... The only one I would caution against spending too much time is the Forums module.