Topic Maps 2008 Day 1 & Day 2
Better late than never…
Topic Maps 2008 - Day 1
Alex Wright of New York Times held the opening keynote of Topic Maps 2008. Wright talked about “Hierarchies, Networks, and the web that Wasn’t“.
His talk gave a quite interesting historic overview of how humans have tried to organize information in “topic maps-like” structures for, well, quite a while. He started with Europe and Thomas Aquinas and talked about the ideas of great minds such as Otlet (classification of knowledge, typed associations), Vannevar Bush (memex, links, trails), Douglas Engelbart (NLS Demo), Ted Nelson, leading up to Tim Berners-Lee and the WWW.
Wright implied that the web that is, is more primitive than the web that could have been, with typed associations (links), classification of knowledge, etc. It was an engaging and most interesting talk.
After Alex Wright, Lars Helgeland (psychologist) of Bouvet gave a talk about how “Topic Maps driven web sites Are Different“, allowing users to enter the same content from different angles and paths. Helgeland mentioned that knowledge is not hierarchical, and that the exercise of classifying things in hierarchies is an academic invention. Children do not classify by hierarchies, but learn to do so in school. I believe that this does also apply to people from non-western cultures, or with a different school of education than ours (western). After all, it was Aristotle and the Greeks who started doing this…
Helgeland points out that while we don’t think in hierarchies, most CMSs are limited to hierarchical categorization! Topic Maps driven web sites allow us to break free from this kind of thinking, and enable different perspectives & paths.
Lars did also mention that because topic maps are so flexible, they had been able to do a complete redesign of Forbrukerrådet’s web site without even touching the content.
After the opening talks, there were two parallel tracks on “Information Integration” and “Education and E-Learning”, followed by “Product Documentation” and “E-Government”, and “Knowledge Management” and “Digital Libraries”. I attended the Info. Integration, eGovernment and KM tracks, of which I found the ones from the latter two to be most interesting.
Semantic Interoperability with Topic Maps
The first talk during the eGovernment track, “Semantic Interoperability: The State of the Nation (Norway)“, was held by some people from Creuna/DIFI and presented challenges related to meta-metadata exchange (data about metadata, obviously).
Most applications are semantically rich “closed worlds”, and there is not much interchange taking place — the message being that technology itself does not lead to semantic interoperability. People need to agree, user needs vary, and the technology must be able to capture the semantics we have agreed upon.
The speakers presented LOS as an ontology for all public services in Norway. LOS defines 400 subjects based on user vocabulary, and includes synonyms, 2 levels of categories, multiple languages, and more, of which public sectors can agree upon. It is, for instance, implemented by both Norway.no and the municipality of Bergen. Not all sites use topic maps, although the result is topic mappish.
At this state, their main challenge is to extend the adoption of common ontologies, extend/merge the LOS ontology and to exchange ontologies. It might also turn out that the LOS subjects are too high level (as many would claim is Dublin Core’s problem too).
Altinn.no (one-stop reporting site for Norwegians), Regjeringen.no (the government) and Stortinget.no (”senate”) was also discussed. Stortinget.no’s upcoming version re-use concepts from Regjeringen.no and is an example of semantic interoperability between governmental portals. Today, both sites publish articles and documents related to the same / given events (on both sites). Tomorrow, Stortinget.no will publish information related to Stortinget, while Regjeringen.no will publish information related to Regjeringen only. By using semantic technologies (I believe this was EPiServer with TMCore), information from one portal can be used by the other portal, but still being maintained by the originating portal, and vice versa.
Altinn.no uses fragments of different ontologies, for example LOS and SERES, and is an ongoing project that uses semantic interoperability to deliver contextualized information — for example forms for use by a specific kind of company.
My note taking wasn’t too good during the second presentation on semantic interoperability. At this point, there was probably too little caffeine and sugar in my blood stream. The speakers (Marco Aarts and Peter Brown) did, however, give an interesting speak on “Semantic Interoperability: The State of the Union (Europe)“, in which they claimed that eGovernment of today is too old fashioned. There are lots of overlapping portals which present the same information. Topic Maps could help to standardize the way in which information is shared. Governmental bodies are in a position to define, use and enforce PSIs.
Knowledge Management with Topic Maps
After the last coffee break of Day 1, Gabriel Hopmans and Peter-Paul Kruijsen of Morpheus gave a very interesting talk about “Establishing Conclusive Proof in Forensic Data Analysis“. Morpheus and SBV Forensics have developed a topic maps driven system that analyze unstructured information sources, such as emails, and generate hypotheses about criminals — and others, I suspect.
1500 different information systems are integrated using topic maps. Information sources are merged using PSIs for known entities, for example social security numbers. The system maintains historic data, such as name changes, by using topic maps’ scope mechanism. The hypotheses are built from unstructured data sources, based on domain keywords, such as “shotgun”, as well as other keywords - such as a person’s name.
Luckily, the Dutch police has decided that all of the generated hypotheses must be manually confirmed / dismissed before it is stored in the system. This is done in order to prevent false information, about things such as innocent people, to enter the system. I wonder if the same approach is used by the U.S. Department of Homeland Security, or whoever, should I happen to mention them, Bush and terrorism in the same sentence. Guess I’ll find out on my next trip to the U.S.
The next talk given during the KM track was a guy from Steria (Peter Crook), who talked about “Adding Value to a Quality Management System for Securities” using topic maps. The system was built on Sharepoint and TMCore (as I often do, I didn’t quite get why they chose to use Sharepoint for this). Crook was a good speaker though, and it was easy to understand that he felt that the new system was much better than the old one, which was based on Lotus software.
I didn’t manage to extract much new information from the last talk on the KM track, but learned that I will need to work on my ability to stay alert during long conference hours.
All in all a good day.
Topic Maps 2008 - Day 2
I failed to take good notes during the second day of the conference. This was partly due to laziness, and partly due to the fact that I was a bit tired from staying out late after the social event of the evening of Day 1.
Nevertheless, I got up and to the conference in time to catch David Weinberger’s keynote speech “Everything is Miscellaneous“. Weinberger gave a very interesting speech that went into topics such as how we all perceive the world differently. We can simply not agree on how to classify things. It might be obvious, but often technologists loose track of this fact (not all read history or philosophy, but I would recommend that more engineers do).
Weinberger partly disagrees with Alex Wright, and believes that the “limitations” of the web, such as one-directional links, really are a blessing. Weinberger claimed that the web Wright wanted would require centralized control, which in turn implies a less free web.
After Weinberg’s speech, the Royal Academy of Music talked about “The RAMline - Rewriting Musical History” using a topic maps based approach. They started by giving an overview of the problem at hand, before diving into detail. In my opinion, this was one of the best use cases presented during the entire conference. Mr. Weinberg has blogged about it, so I wont go into more detail here.
After the first break of day 2, in which I indulged myself in breakfast — cakes, I attended the Enterprise Architecture talk. The talks presented at this track were OK, but not too relevant for me. The most interesting was Stian Danenbarger’s talk about semantics, published subjects and 7 levels of interoperability. I shamelessly steal from Svein Ølnes’ blog:
- No interoperability
- Technical
- Syntactic
- Semantic
- Pragmatic
- Dynamic
- Conceptual
The transition from semantic to pragmatic interoperability is important: while semantic interoperability does only allow for one meaning, pragmatic interoperability is about multiple (different) meanings about the same subject under different contexts (supported by Topic Maps’ scope).
At the closing plenary Volker Stümpften was “Explaining the Complexity of Life with Topic Maps“. It was about integrating 100s of databases with biological data, allowing for new ways of organization, navigation and visualization. I wrote about this after TMRA 2007.
Steve Pepper gave an inspiring and visionary closing keynote “Everything is a Subject” about subject centric computing. He argued that too little of Topic Maps potential is being used (i.e. scope, PSIs and merging), and that applications are too document centric. What the users are interested in, are the subjects. He wants applications, file systems and operating systems that are subject-centric instead of document-centric; being able to associate e-mails, web sites, documents, etc., etc. with unique subjects. This must, or at least should, of course be built using Topic Maps 
Again, Joho The Blog has written more.
I missed the closing panel discussion, as I were on my way to the airport at that time. It seems to have been interesting, though, and Svein Ølnes has written What’s Wrong with Topic Maps? as a follow up to the discussion.
Conclusion
All in all a good and successful conference with lots of interesting and inspiring talks. I did especially enjoy the keynotes, the talks on semantic interoperability and knowledge management, as well as the tutorial day. Perhaps a tad long (3 full days), though, as there were lots of new information to process (me not being dual core).
Other conference bloggers:
- David Weinberger (multiple postings in the “conference coverage” category)
- Svein Ølnes - Day 1 & Day 2 (Norwegian)
- Kjernedryss (Norwegian)
Lars Marius Garshol was busy taking pictures.