The Semantic Web And Policy Workshop (SWPW) will be held on 7 November 2005 in conjunction with the 4th International Semantic Web Conference in Galway, Ireland. The workshop will cover policy-based frameworks for the semantic web as well as the use of semantic web technologies in policy frameworks for other application domains such as multiagent systems, grid computing, networking, and storage systems. Submitted papers should describe original research results or articulate a position, describe an application or demonstrate a working language or system. Papers must be submitted electronically by 25 July 2005; decisions will be announced on 5 September with final camera ready copy due on 30 September.

The W3C’s Semantic Web Best Practices and Deployment Working Group aims to “provide hands-on support for developers of Semantic Web applications.” Their approach is to develop and publish notes explaining good ways to tackle common KR problems in RDF and OWL. For example, given RDF’s underlying binary relations, what are good ways to encode the n-ary relationships needed by many domains? If you are building ontologies or just trying to understand how RDF and OWL should be used, you need to take a look at these.

The working group has published three new working drafts:

• SKOS Core Guide. “SKOS Core provides a model for expressing the basic structure and content of concept schemes (thesauri, classification schemes, subject heading lists, taxonomies, terminologies, glossaries and other types of controlled vocabulary).”
• SKOS Core Vocabulary Specification. This document gives a reference-style overview of the SKOS Core Vocabulary as it stands at the time of publication. It also describes the policies for ownership, naming, persistence and change by which the SKOS Core Vocabulary is managed.
• QuickGuide to Publishing a Thesaurus on the Semantic Web. “This document describes in brief how to express the content and structure of a thesaurus, and metadata about a thesaurus, in RDF.”

These join some very useful previous working group notes and working drafts, including the following:

• Representing Classes As Property Values on the Semantic Web. “This document addresses the issue of using classes as property values in OWL and RDF Schema. It is often convenient to put a class (e.g., Animal) as a property value (e.g., topic or book subject) when building an ontology. While OWL Full and RDF Schema do not put any restriction on using classes as property values, in OWL DL and OWL Lite most properties cannot have classes as their values. We illustrate the direct approach for representing classes as property values in OWL-Full and RDF Schema. We present various alternative mechanisms for representing the required information in OWL DL and OWL Lite.”
• Defining N-ary Relations on the Semantic Web: Use With Individuals. ” In Semantic Web languages, such as RDF and OWL, a property is a binary relation; that is, it links two individuals or an individual and a value. How do we represent relations among more than two individuals? How do we represent properties of a relation, such as our certainty about it, severity or strength of a relation, relevance of a relation, and so on? The document presents ontology patterns for representing n-ary relations and discusses what users must consider when choosing these patterns.”
• Representing Specified Values in OWL: “value partitions” and “value sets”. “Modelling various descriptive “features” (also known variously as “qualities”, “attributes” or “modifiers”) is a frequent requirement when creating ontologies. For example: “size” may describe persons or other physical objects and be constrained to take the values “small”, “medium” or “large”; rank may describe military officers and restricted to a specific list of values depending on the military organisation. In OWL such descriptive features are modelled as properties whose range specifies the constraints on the values that the property can take on. This document describes two methods to represent such features and their specified values: 1) as partitions of classes; and 2) as enumerations of individuals.”

The standard view of the semantic web assumes that we will not have a single consensus ontology for a given domain, but many, each with its own base of users and applications. Thus it’s essential that we have good techniques and tools to translate information expressed in one collection of ontologies into another. One of the issues that we have not yet faced head on is that most of these mappings will probably be approximations. Here’s a good overview of the Bayesian approach to OWL ontology mapping being developed by Yun Peng and his students.

A Bayesian Methodology towards Automatic Ontology Mapping, Zhongli Ding, Yun Peng, Rong Pan, and Yang Yu, AAAI Workshop on Contexts and Ontologies, July 09, 2005.

This paper presents our ongoing effort on developing a principled methodology for automatic ontology mapping based on BayesOWL, a probabilistic framework we developed for modeling uncertainty in semantic web. The proposed method includes four components: 1) learning probabilities (priors about concepts, conditionals between subconcepts and superconcepts, and raw semantic similarities between concepts in two different ontologies) using Naive Bayes text classification technique, by explicitly associating a concept with a group of sample documents retrieved and selected automatically from World Wide Web (WWW); 2) representing in OWL the learned probability information concerning the entities and relations in given ontologies; 3) using the BayesOWL framework to automatically translate given ontologies into the Bayesian network (BN) structures and to construct the conditional probability tables (CPTs) of a BN from those learned priors or conditionals, with reasoning services within a single ontology supported by Bayesian inference; and 4) taking a set of learned initial raw similarities as input and finding new mappings between concepts from two different ontologies as an application of our for-malized BN mapping theory that is based on evidential reasoning across two BNs.

mSpace is an interaction model to help explore relationships in information. It is a research project developed by the School of Electronics & Computer Science (ECS) at the University of Southampton.

mSpace combinds the use of information semantics and a flexible UI interface. It allows users to explore new information that they only have limited knowledge about. One problem that the mSpace project tries to address is the following:

What if you want to find something from an domain where you have a general interest but not specific knowledge? How would you find classical music you might enjoy if you don’t know what Beethoven or Berlioz sounds like? What a Sonata is? The difference between Baroque or Romantic? What do you type into Google?

You can find mSpace technical reports and demo apps here.

Getting people to agree on a single ontology has always a problem in the Semantic Web research. There are two schools of thinking. Some people believe that in the future all ontologies will be defined by some kind of standard bodies or special interest groups. Some others believe that there will be many different ontologies flowing around, and standard ontologies will emerge as the result of an “evolution” process — i.e., good ontologies will get used and bad ontologies will be forgotten.

I think the latter is more likely to happen than the former. The new tag service of the Technorati website is a good example. Here is a short description of the service from a Slashdot post:

Technorati (a search engine for blogs) has a new ‘tag’ service. If your blog tool of choice uses Categories, has a RSS/Atom feed, and pings technorati, then you’re done. If not, you can add tags via a new tag markup. The twist is that Technorati is working with Del.icio.us (a social/sharing bookmark manager website) and Flickr (a social/sharing photo web site) to read their tagged content! So Flickr pictures, Del.Ico.us bookmarks, and blog posts all on one page! Here’s an example result for the tag Toronto. There is some documentation as well. One current limitation is that there is no way to do tag intersection as with del.icio.us (i.e. http://del.icio.us/tag/toronto+food ) like http://www.technorati.com/tag/toronto+Food. Tagging (also know as Folksonomies) was the topic recently on Slashdot: Folksonomies In Del.icio.us and Flickr.”

This is a good overview paper with the perspective of someone in library and information science.

Folksonomies - Cooperative Classification and Communication Through Shared Metadata, Adam Mathes, UIUC,December 2004. “This paper examines user-generated metadata as implemented and applied in two web services designed to share and organize digital media to better understand grassroots classification. … Conclusion. A folksonomy represents simultaneously some of the best and worst in the organization of information. Its uncontrolled nature is fundamentally chaotic, suffers from problems of imprecision and ambiguity that well developed controlled vocabularies and name authorities effectively ameliorate. Conversely, systems employing free-form tagging that are encouraging users to organize information in their own ways are supremely responsive to user needs and vocabularies, and involve the users of information actively in the organizational system. Overall, transforming the creation of explicit metadata for resources from an isolated, professional activity into a shared, communicative activity by users is an important development that should be explored and considered for future systems development.”

Interest in ontologies has gone down and up over the past 20 years and its been very strong in the last five years. Designing a good ontology for a complex real world topic is hard and made especially so by the usual goal that it be relatively independent of any small set of driving tasks. There are so many ways you can go wrong — too simple, too complex, too philosophical, to pragmatic, non-extensible, too big, too small, too brittle, too loose. And how do you evaluate the one you come up with? Sometimes it seems that ontological engineering requires graduate level training in way too many advanced topics — knowledge representation, logic, databases, philosophy.

While the semantic web movement hasn’t changed any of these problems, it has opened up new avenues by making this a problem by and for the web — an open, distributed, heterogeneous environment in which people and software agents create, publish, search for, combine, exchange and use information.

One interesting phenomenon is a number of sites which are using what some call folksonomies — informal tagging systems developed bottom up by their users. Examples of sites that use folksonomies include flickr, furl, del.icio.us and Google’s gmail. As a way to build an ontology, you can’t get much simpler that this — the tags form a flat one-level taxonomy of classes. You can attach a set of tags to an object (URL, picture or email message) and find objects indexed by a set of tags. What you can’t do are things like (i) define relations between tags (e.g., declare that rdf is a subtag of semanticWeb or that NYC and newYork are equivalent); (ii) form combinations of tags other than intersection (e.g., find pictures tagged as domesticatedAnimals OR pets but NOT cats); and (iii) define and use properties (e.g., this picture depicts an animal whose owner is a person with lastname=”finin”).

This is not a great leap forward for classification theory and the basic approach is quite common (e.g., see the use of faceted classifications in library science or polyclave classification systems in Biology), but what is interesting is letting a community of people develop and share folksonomies in a natural way with the hope that consensus vocabularies will naturally emerge.

Flickr, furl and del.icio.us allow you to make public your tags and tagged objects and to search over those of others, introducing an interesting social dimension. In the natural course of things, users will tend to converge around a set of tags to denote a common shared concept. This is accelerated by the fact that, for del.icio.us and furl, users are tagging objects from a common universe of URLs. Simple statistical techniques can reveal tags that are related or similar in that they’ve been used by different people to classify a common object. If shared consensus sets of tags do emerge in these communities as they grow, it will be significant.

Can we extend the expressive power of these systems, say by using RDF and introducing some of the features of RDFS and (even) OWL, resulting in folksologies. It’s a good question. We can do it, of course, but will the result be as easy for people to learn and use? That’s an even better question.

IBM and Massachusetts General Hospital Announce Effort to Improve Information Sharing Among Cancer Researchers “Effective tools for information management, integrated tightly with underlying computing and data infrastructures, are key to life sciences researchers gaining new insights into complex problems,” said David Grossman, Distinguished Engineer, IBM Internet Technology Group. “In addition, the use of semantic web technologies to integrate many sources and formats of data with advanced modeling algorithms is particularly helpful for this type of large-scale collaborative project.”

“There is an urgent need to develop a common, unifying infrastructure that enables the integration and sharing of knowledge about cancer — both in terms of disparate data and distinct computational tools — with the goal of modeling cancer as a complex dynamic system,” said Dr. Deisboeck. “While advances in cancer research and new technologies have generated a wealth of new data and insight, all too often the lack of shared systems and standards makes integration of this crucial knowledge difficult or impossible.”

Cwm 1.0 has been released. Cwm is a general-purpose data processor for the semantic web, implemented as a forward chaining reasoner which can be used for querying, checking, transforming and filtering information. Its core language is RDF, extended to include rules, and it uses RDF/XML or RDF/N3 serializations.