In two recent articles Alex Iskold explores current issues and suggests a different, more top-down, approach to getting the semantic web up to speed. He stresses the importance of  building on top of the web as-is, instead of requiring developers to build a new web. He certainly has a point and this pragmatic approach by focussing on structured data in the form of API's, feeds, and enhancing pages with microformats or embedded RDF may very well be the way to go for now.

On the other hand I do wonder what tools need to be available for developers to quickly start using semantic web technology in their projects. For one storage is an issue, as regular relational databases aren't the best fit for efficiently storing and retrieving RDF triples. Gradually a new set of developer tools is evolving, e.g. YARS for storage and Virtuoso as a development platform. It's impressive technology, but still not as easily usable as say PHP combined with a MySQL database. The Ruby on Rails folks are lucky to have ActiveRDF, but still, it's not something you pick off the shelf and start using as you'll still have to figure out how to store your triples.

Does anyone have hands on experience with really easy to use semantic web tools to quickly build a web application? I would love to hear your suggestions.

As we have been dropping little bits about the next release of Elgg I can say we will also be taking a really good look at translations, making it work even better and efficient. Some changes will need to happen, e.g. every module, or plugin will have its own translation file and strings will need to be more consistently constructed (there currently is much overlap). This will mean extra work initially, but the result will be smaller and more coherent translation files  which should be much easier to maintain. Expect a full announcement of things to come next week.

Things are really ramping up, you'll probably have read the announcement about Curverider securing investment. Apart from this helping me become official Curverider employee #3, I think more importantly it will give Elgg the extra boost to continue as a true open source, standards committed platform, but also in covering new grounds and setting new standards. But probably most importantly it will give us extra means to better serve and support the community with tools that matter. I'm definitely looking forward to the times ahead.

Although the basic things we do on social networks don't vary much across sites (building a friends list, a profile) the semantics and approaches do vary across context which makes it hard to settle on a common export format. E.g. the buildup of a profile on a site for professionals will be different from one that focuses on video exchange. A lot of activity is happening in this arena, both on mechanisms of exchanging the data and the format itself. Data formats like Microformats, RDFa and eRDF seem to dominate discussions and allow for transport as well - it's just pointing the import tool at an URL and extracting the information since it is embedded in the content. Others are heavily lobbying for using OpenId as the cure-all, and in particular have high hopes that the attribute exchange  (AX) extension built on top of the authentication layer will solve the problem of having a unified way of exchanging information.

Personally I am biased towards RDFa as it provides all the benefits of RDF and offers a rich way of expressing meaning. It is also extensible, which I think cannot be said about Microformats and OpenId AX: the namespace is flat and I can see class and type collisions ahead. What also attracts me is the transport mechanism, information expressed in RDFa can be embedded in (X)HTML but could just as well be available as regular RDF. Maybe this touches upon the biggest problem I have with OpenId AX which is the exchange of attributes strapped on top of a single protocol which was designed in the first place as a way to prove you own or control a particular URL. Apart from that there isn't even a specific property set defined and the name/value pairs will, as I said earlier, very likely limit it to flat structures only.

I think it also important for people involved in social software to remember that users are involved in much more than their social networks, and it's important to keep an eye on larger developments like the semantic web. Portable social networks is something to work towards, but in the interest of the user the basics shouldn't get limited to a single set of applications.

mnh posted a photo:

mnh posted a photo:

mnh posted a photo:

Front side view

mnh posted a photo: