The semantic web

Today I'm going to talk about the "semantic web" and how our own strategy with ocPortal relates to it.

The semantic web has it's roots in academic circles, and it's all about allowing computer software to better understand web pages, websites, and the content of the web in general. Currently pages are pretty much only understandable by humans, but by writing them in a special way so computers can interpret them all kinds of new possibilities are opened up. A good example is search engines - if search engines can better interpret website contents much more sophisticated queries can be written. The possibilities are endless though, and I'll explain some later in this post.

The semantic web never really took off in it's original and most sophisticated form, and there are a few reasons for this:

• Academics wanted to do too much too quickly. It got very tied up with Artificial Intelligence research - and just like plans for us to all have Robot Servants - was rather too ambitious to realise.
• There was a "chicken and egg" situation. There was no point in making "semantic web" interpreters until there were "semantic web" pages for them to read, and vice versa.
• There was a lack of imagination for practical uses.

There has, however, been some renewed interest recently. Some people are thinking of it as the next Internet revolution - "Web 3.0". We want to help break the "chicken and egg" situation and bootstrap the process and ocPortal 4.2 will represent our first step. We anticipate making further strides in the future and helping our users help define the future of the web.

We have implemented a number of technologies in ocPortal 4.2 to help make the first stage of "Web 3.0" happen:

• Dublin Core. Dublin Core is a meta-data system for storing attributes relating to resources. View screens for all resources in ocPortal (downloads, videos, news entries, etc) will provide meta data such as edit dates, descriptions, and creators.
• Dublin Core: ocPortal extensions. We have created our own extension of Dublin Core, exporting details such as which user has the most points, how many members have accounts on the website, and what the rating of content is.
• HTML5 'rel attribute' semantics. We have gone to quite some effort to mark-up hyperlinks so that their purpose is clearly identified. A perfect example is breadcrumbs are now identified as 'up'.
• ocPortal extensions for the 'rel' attribute. We've added some extra 'rel' values: print, site_map, privacy, contact, dorating, docomment, dovote, pageexpand, pagecontract, add, delete, report, track, untrack, report, reply, history, move, validate, unvalidate.
• We've added a new 'base_url' meta tag, which would allow software to identify what pages relate to what 'parent site' (until now there was no good way to do this - only humans could really tell, by analysing page styles or linking patterns). We hope others will follow.
• Improved support for OpenSearch, allowing external search engines to better tie themselves into our own one.
• Microformats, which allow making web page content that both humans and computers can understand. We'd already implemented hCalendar (calendar events), hCard (OCF member profiles), XFN and rel-nofollow. Now we've also added hReview (review catalogue), rel-enclosure (attachments), rel-license, and rel-tag.
• XHTML semantics: (ocPortal has included excellent 'XHTML semantics' for years, but I thought I would just mention that XHTML semantics are yet another part of the overall mix that enables new possibilities).

So what practical things will this allow to happen? The immediate answer is absolutely nothing. But once there are enough pages out there the innovation can really start.
Here are some possibilities for you to imagine:

• Google could do page rankings of sites based on the ratings users give them. Imagine doing an advanced Google search of your website and being able to order search results by ranking.
• Software could be written that reads your browsers bookmarks and scans all of them to find which sites have content written by you, and then giving you links to it all. Great for finding all your old forum topics, for example.
• Your web browser could be programmed to go up a breadcrumb path by pressing one of your extra mouse buttons.
• Google could run an advanced query to find all the largest forums on the Internet that have the word 'ghosts' in their title (let's pretend that you want to find the largest 'ghosts' forum but on a normal search you can't find it because you just get results that aren't forums, or mention the word 'ghost' casually).
• Your web browser could be scripted to find all topics that were created by 'Jim' and track them.

If you have your own ideas on what could be done, please post them!

We're really excited about the possibilities for the web and this is just the beginning of our plans. It pushes ocPortal to the forefront of real-world semantic-web enabled software, but we want to go a lot further and really drive innovation!

Chris Graham said

• We've added a new 'base_url' meta tag, which would allow software to identify what pages relate to what 'parent site' (until now there was no good way to do this - only humans could really tell, by analysing page styles or linking patterns). We hope others will follow.

The scuttle has it that IE8 will be capable of doing this with its 'tab/links' methodology, grouping them into similar coloured tabs depending on the association.

For the rest of your discussion - my mind boggles!

Great stuff indeed gentlemen.

Let's not forget that it's a visual world out there.  Don't let this great product with all it's power be sidelined because of it's appearance.

I realize that it's a conscious decision to keep it sparse so that one can more readily customize it, but there should at least be a sample theme in addition to the default that really stokes the fire!

Perhaps even some eye candy type mods available for quick implementation by the none technical that can be applied to any theme.  Example fancier more modern buttons, image rotators/sliders ( I believe this is already address in galleries upgrades in 4.2), gradients in buttons/background/header, etc.

my 2 cents, for what they are worth.   I know right, I'm not asking for a whole lot  

Elgg_Noob said

I realize that it's a conscious decision to keep it sparse so that one can more readily customize it, but there should at least be a sample theme in addition to the default that really stokes the fire!

I think the Theme Wizard would cover this. I've seen other CMS products have different themes but they really aren't that different. Just color changes really. And with this Theme Wizard, you have freedom of what colors you want. With the other CMS' you actually have to go into their CSS and other pages to change all that if you like the theme but want different colors. Kudos to the ocPortal Theme Wizard.

Elgg_Noob said

Perhaps even some eye candy type mods available for quick implementation by the none technical that can be applied to any theme. Example fancier more modern buttons, image rotators/sliders ( I believe this is already address in galleries upgrades in 4.2), gradients in buttons/background/header, etc.

I think for the beginner, the addon page works well. Perhaps when more coders come forward who play around with ocPortal, we'll see a lot more in there.