This week I attended a workshop organized by the IKS research project in Salzburg. To use the description from their website, IKS “will provide an open source technology platform for semantically enhanced content management systems.”
Semantic technology includes a lot of aspects: smart algorithms to let a computer guess the semantics of a text or an image, ontology design, querying and reasoning, user presentation and interaction, the semantic web, ... What is understood by a semantically enhanced CMS is still an ongoing exercise, a number of user stories can be found on the IKS wiki.
The IKS project is not something that happens behind closed doors. Interested CMS vendors can participate, and all code is released as open source, thus available to everyone.
The workshop brought us two different perspectives. The research-partners in the project had a more top-down, theoretical approach. They defined an “interactive knowledge stack” (IKS) and an architecture. For more information on the IKS stack and the general vision, check out their website.
The industry partners had a more hands-on approach and showed us actually working code.
Bertrand Delacretaz, an old Apache Cocoon buddy, presented (1, 2) FISE. FISE is a standalone server integrating semantic technology, such as semantic lifting and querying. FISE is to semantic technology as what SOLR is to Lucene: a stand-alone server with a REST interface, integrating lots of interesting functionality and making it easier to use. FISE is more than just a wrapper project though: it already offers added functionality by having the ability to combine the functionality of different “semantic enhancement engines”. The FISE code is publicly available so you can go play with it.
Another component being developed is an editor called Aloha that will make it easy to create semantically enhanced content. One of their goals is that something like “Insert person” should be made as simple as inserting an image or making text bold. For this purpose, a core editor framework, extensible by plugins, is being developed. I liked that they want to produce cleanly-structured HTML, hiding the differences in behavior of the browser-embedded editors: just try putting a word in bold or try to start a new paragraph by pressing enter and compare the HTML being produced by different browsers. In our Daisy CMS, we have an advanced server-side HTML cleanup component which brings the HTML in a canonical structure, but having this closer to the user is even better. A first public release should be in July, looking forward to it.
There were also a couple of impromptu presentations. Thomas Kurz presented a semantic wiki built on Kiwi. We were shown features like semantic tagging and RDF editing, but also the ability for authors to embed queries in pages, of which I can't help mentioning that Daisy had this since its first release in 2004.
Michael Dreusicke showed off his semantic CMS, where everything up to the word-level is linked data. Henry Story presented foaf+ssl, which tries to do something about the social network silos problem. Ozgur Killic presented an ontology bridge to map JCR or CMIS structures to ontologies. I learned about projects like Apache Clerezza, UIMA, OpenNLP and others.
Between the presentations, there was lots of time for discussions and networking. So to conclude, it was very interesting to be in this semantic atmosphere for two days, and I brought back a lot of interesting (outer-)thoughts.