Friday, November 24, 2006

So much compelling technology and all freely available under LGPL .

On November 16-17, 2006 the final review of the SEKT project took place in BT´s installations in Ipswich, UK. SEKT is three years European IST project (running from Jan 2004 to Dec 2006) aimed at improving Knowledge Management using Semantic Technologies. The project was organised in several technology work packages (with the aim to create innovative and cutting edge technology), and three application (case study) work packages (to provide real life requirements and to provide a testbed). The three applications are related to

  • A large Digital Library
  • Decision support for (Spanish) Judges on duty
  • Knowledge management for an IT consultancy company

The technology developed can be divided in different groups

  • Text mining, ontology learning, clustering components to structure "hidden" information
  • Natural language processing components to extract semantic information from unstructured content
  • Ontology engineering software (editing, mapping and versioning)
  • Components for automatic annotation
  • An integration platform compliant with industry standards

Specific effort has been dedicated to scalability and methodological aspects of semantic solutions.

The SEKT project has produced an impressive list of software components that -if configured rightly- can solve many existing knowledge management problems. The good thing is that most software is freely available under the LGPL license.

So, any company (and of course university) with technological IT skills, can simply go to www.sekt-project.com and download the components of interest, and plug and play prototypes.

In addition, the project has produced a training website, with movie material to get quickly up to speed with Semantic Technologies. This includes movies taken from SEKT people giving tutorials and screen cams from the software and applications. There is also a website especially aimed at decision makers and IT managers.

The review went very well. The EC, the reviewers and the project partners were very satisfied. To be continued ...?

Wednesday, November 22, 2006

FP7, a happy surprise

As you’ve read before in this blog, I am attending the IST Conference in Helsinki. During the session on Intelligent Content and Semantics, Stefano Bertolo gave a practical presentation on FP7 proposals. The happy surprise concerns the so-called STR-D projects (Specifically Targeted Research Projects – Demonstration) geared towards field experimentation (“use cases”).

ISTConf2006-Helsinki-SemanticContent

Projects of this kind are (quoting from Bertolo, full presentation here):

  • centred around existing, promising but untried technologies
  • designed to go one step forward towards
    • packaging, configuring … and testing
  • assess suitability & viability
    • functionality
    • performance
    • usability (hide technical complexity!
  • within a well defined domain / user context
  • rigorous evaluation plans & metrics
  • active user involvement & feedback
  • adequate documentation of results (positive/negative)

Throughout my postings in this blog so-far you may have read some frustration about the difficulty in closing the gap between R&D results and the market. To me it seems that these STR-D projects could play a promising role in filling this gap. It allows, for example, to take results of existing R&D projects (e.g. the SEKT project), and develop them a step further into direction of the market. This form of project seems to acknowledge that it is hard and expensive (and involving high technology risk) to bring R&D results to the market, and therefore I applaud it.

Sunday, November 19, 2006

Innovation Funnels and Impact Sprays

On Nov. 14, 2006 the second annual review of the OntoGrid project took place in Brussels. The goal of the OntoGrid project is to design an architecture for the Semantic Grid; the long hoped for combination of Grid and Semantic Web. The idea is to add semantic metadata to Grid resources (so-called “semantic bindings”) such that Grid resources can be reasoned about, thereby paving the way for automatic discovery, selection and combination of Grid resources. In this sense the Semantic Grid and Semantic Web Services have several things in common (see my posting on the DIP project)

Many of the projects I am involved in are about developing and applying research results and technology to real life settings. That is, there is an important focus on “exploitation plans” of the research results, with the aim of commercial uptake in the market. In the OntoGrid project that is not the case. OntoGrid is a STRP (a Specific Targeted Research Project) whose focus is on research rather than on potential commercial exploitation of the project results.

Apart from mentioning that the review went very well; both the reviewers and the Commission were very happy with the project (you can read all about the project at the website www.ontogrid.net), we found an interesting distinction between research projects focused on commercial exploitation of the results, and research projects focusing on impact on the research communities. This difference is expressed in the Innovation Funnel versus the “Impact Spray” (coined by Carole Goble).

As I wrote in IEEE Intelligent System (AI’s Future: Innovating in Business and Society, May/June issue, 2006, pp 72-73), The innovation funnel models how ideas become products. Having ideas is easy, turning them into concrete proposals is a bit more difficult, transforming that into a working prototype is much harder, and commercializing the software is a completely different story. Few ideas make it to commercialization and end up in a new product, service, or even company. A consequence of the innovation funnel is that as you move from idea to results, you need increasingly more investment. Having an idea is cheap; commercializing a software product or service might involve millions of dollars. The Innovation Funnel applied to OntoGrid is illustrated in the figure below. There are quite a few ideas and proposals for technology and/or components, but only a few will make it in the end to the mainstream market (may take easily 5 years, I estimate).

InnovationFunnel-Ontogrid


The impact spray is exactly the opposite. One starts with one main idea (Semantic Grid in this case), which leads to several proposals, which each may lead to several prototypes, etc. As the project creates more impact, and the research community takes up the individual results and starts using it, it quickly spreads out like a spray. Maybe s omething like the selfish memes of Richard Dawkins (Richard Dawkins, ``The Selfish Gene'', Oxford University Press, 1976). The figure below shows to Impact Spray applied to OntoGrid current state (© Carole Goble, as far as I know).

impact-spray

Wednesday, November 15, 2006

The 2006 edition of the IST Conference, Nov. 22, Helsinki

Next week, I will give two presentations at the IST conference about applications using Semantic Technology in enterprises and public organisations. One presentation is a general one discussing the market (estimated size, drivers, and inhabitants) for Semantic Technology, several applications (both for corporate and public Semantic Web), and some barriers this type of technology is likely to encounter. The presentation can be downloaded here.

Date, time and place: Intelligent Content and Semantics, 22 November 2006, 14:00-15:30, The Helsinki Fair Centre, room Lappeenranta

The other presentation discusses a semantic application for Spanish judges, and in particular recent judges. In order to become a judge in Spain, one has to pass a public exam held as a competition. Only the top performers are invited to become judge. Once accepted, the recent judge gets all responsibilities, but has still little practical knowledge. The system we are building provides a solution for this problem in the form of an intelligent FAQ system. We built a high-quality corpus of about 1000 frequently asked question-answer pairs, collected and maintained, based on interviews with more than 400 judges. The judges can query the system in natural language (e.g. "I have given an injunction of protection and the woman is asking me for a withdrawal of the measure. Should I withdraw it?"), and get as result a list of related question answer pairs. In addition, it provides a list of documents with related sentences to the answer.

The novelty of the system is they way in which relevant question-answer pairs are retrieved, namely based on domain semantics, represented in ontologies. We can say that –to some extent- the system “understands” the user query, and based on this understanding, provides relevant results. The presentation can be here.
The session is called: Building Semantic Knowledge Applications and the program is:

11.00 – 11.05 Introduction and overview
Dr. John Davies
11.05 – 11.15 A Semantic Application for Spanish Judges
Dr. V. Richard Benjamins
11.15 – 11.25 Large-scale semantic web applications
Professor Enrico Motta
11.25 – 11.35 A case study in semantic web services
DIP project representative
11.35 – 11.55 Providing Intelligent Content by Using Web Semantics and Web Mining
Dr. Pinar Senkul, Dr. Marko Grobelnik, Professor Josiane Mothe
11.45 – 11.55 The Role of Language - Extending the Semantic Web towards Web Pragmatics
Professor Kurt Englmeier
11.55 – 12.10 Cost estimation for ontology engineering
Prof. Rudi Studer
12.10 – 12.30 Panel discussion
Facilitated by Dr John Davies


Enjoy!

Wednesday, November 08, 2006

Semantic Web and Web2.0

Since some time, I am asking myself the question “what does Web2.0 mean for a company like iSOCO”, which is currently focusing on bringing Semantic Web technology to the market. Both technologies are expected to have high impact on businesses and society. But what is their relation? Are they complementary or competitors for achieving the same dream? When one says “Semantic Web”, one says “ontologies”; formal web representations of knowledge of a particular domain. When saying “Web2.0”, one says “communities”, “services” and much more (see http://www.oreillynet.com/pub/a/oreilly/tim/news/2005/09/30/what-is-web-20.html).

In my view, they are complementary, both helping to bring the web to its full potential for society and businesses. Rather than expressing this in text, I tried to add relevant tags (for Semantic Web and Web2.0) to all our Semantic Web applications we have been built over the past few years. The presentation can be downloaded here. Enjoy! It also includes a small introduction to iSOCO, but you can safely skip that.