UKUUG: Automated Website Synthesis

Siu-Wai Leung’s talk Automated Website Synthesis was quite an eye-opener for me at UKUUG 2003:

Basically, Siu-Wai presented a proof-of-concept how to apply Artifical Intelligence (AI) and Semantic Web technologies to transform annotated content to a Website. Read it again! His talk was not only about how to present annotated content on a Website, no, it was rather about how to automatically create a Website from annotated content! In fact, what he talked about was ontology-based web content development and design.

He perfectly demonstrated the use of such an approach for aviation accident reporting. The tools he used are the HEIDI ontology (HEIDI is the abbreviation for “Harmonisation of European Incident Definition”). Furthermore, upon any aviation accident, a sophisticated pilot report form has to be filled in. Furthermore, data is automatically being recorded storing e.g. the altitude and speed of the airplane. Based on that data, Siu-Wai was able to create a graph showing the altitude of the plane, or a tree of events of the aviation accident.

The Website prototyping was done using a Simple Website Interface Model (SWIM), the Web Modelling Language (WebML). More on the technology behind can be found in Siu-Wai’s paper (PDF).

At the end of his talk, Siu-Wai summarized that the ontology-based mapping of information, Website, and perception is an AI problem (that’s his mapping hypothesis). He presented some ideas of how this could be achieved in the future. I very much liked his idea to use genetic programming to mutate the design of single Webpages or complete Websites to find the most suitable format to present the information to the user – an intelligent way of adaptive and personalized C/KMS.

That was a cool talk 😉 The aviation accident reporting demo seems like a perfect prototype to exemplify the possible advantages of AI and the Semantic Web. Also, the idea of genetic mutations of info visualization acknowledges the reality that every individual has a different way of filtering information and offers a higher flexibility in personalized content presentation then the theme-based CMS currently provide.

Thesis 2: Think Different!

Most Semantic Web or AI discussion is about finding adequate information with intelligent retrieval systems. Ontologies are used to identify content based on formalized knowledge, either by automatic statistical reasoning or by human annotation. Usually, knowledge management applications concentrate on one ontology for a certain area of expertise to provide retrieval systems for a specific group of users aka experts.

The Semantic Web will be very much about communication/interaction between different, even contradictory world views. Ontologies are the computational equivalent to value systems because they define terms/concepts and their relationship. For future content or knowledge management systems, this means, that they must be able to handle multiple ontologies. Especially, if we think of innovation as based on new insights, multiple ontologies applications become interesting to knowledge based economies. Just imagine applying different, maybe contradictory ontologies aka world views to a specific or global information pool.

One major prerequisite for such a sophisticated system of multiple and interactive realities and interpretations based on ontologies, is that ontologies become easy and inexpensive to create. Also, they need to be transfered between applications. In this respect, Weblogs could server as an initial start, because here individuals categorize their Weblog entries based on self-made, usually flat structure of categories. Although this does not meet the requirements of ontologies, it would be interesting to see Weblogs not only interact on the basis of Weblog entries (e.g. via trackbacks), but also related to their catalogs/subjects/categories.