Curated content for content, computing, and digital experience professionsals

Tag: semantic web (Page 1 of 2)

W3C Publishes New Working Drafts for OWL 2 – Last Call

The W3C OWL Working Group has published new Working Drafts for OWL 2, a language for building Semantic Web ontologies. An ontology is a set of terms that a particular community finds useful for organizing data (e.g., for data about a book, useful terms include “title” and “author”). OWL 2 (a compatible extension of “OWL 1″ ) consists of 13 documents (7 technical, 4 instructional, and 2 group Notes). For descriptions and links to all the documents, see the ” OWL 2 Documentation Roadmap.” This is a “Last Call” for the technical materials and is an opportunity for the community to confirm that these documents satisfy requirements for an ontology language. This is a second Last Call for six of the documents, but because the changes since the first Last Call are limited in scope, the review period lasts only 21 days. For an introduction to OWL 2, see the four instructional documents: an “overview,” “primer,” “list of new features,” and “quick reference.” http://www.w3.org/2007/OWL, http://www.w3.org/TR/2009/WD-owl2-new-features-20090421/

Ontologies and Semantic Search

Recent studies describe the negative effect of media including video, television and on-line content on attention spans and even comprehension. One such study suggests that the piling on of content accrued from multiple sources throughout our work and leisure hours has saturated us to the point of making us information filterers more than information “comprehenders”. Hold that thought while I present a second one.

Last week’s blog entry reflected on intellectual property (IP) and knowledge assets and the value of taxonomies as aids to organizing and finding these valued resources. The idea of making search engines better or more precise in finding relevant content is edging into our enterprises through semantic technologies. These are search tools that are better at finding concepts, synonymous terms, and similar or related topics when we execute a search. You’ll find an in depth discussion of some of these in the forthcoming publication, Beyond Search by Steve Arnold. However, semantic search requires more sophisticated concept maps than taxonomy. It requires ontology, rich representations of a web of concepts complete with all types of term relationships.

My first comment about a trend toward just browsing and filtering content for relevance to our work, and the second one about the idea of assembling semantically relevant content for better search precision are two sides of a business problem that hundreds of entrepreneurs are grappling with, semantic technologies.

Two weeks ago, I helped to moderate a meeting on the subject, entitled Semantic Web – Ripe for Commercialization? While the assumed audience was to be a broad business group of VCs, financiers, legal and business management professionals, it turned out to have a lot of technology types. They had some pretty heavy questions and comments about how search engines handle inference and its methods for extracting meaning from content. Semantic search engines need to understand both the query and the target content to retrieve contextually relevant content.

Keynote speakers and some of the panelists introduced the concept of ontologies as being an essential backbone to semantic search. From that came a lot of discussion about how and where these ontologies originate, how and who vets them for authoritativeness, and how their development in under-funded subject areas will occur. There were no clear answers.

Here I want to give a quick definition for ontology. It is a concept map of terminology which, when richly populated, reflects all the possible semantic relationships that might be inferred from different ways that terms are assembled in human language. A subject specific ontology is more easily understood in a graphical representation. Ontologies also help to inform semantic search engines by contributing to an automated deconstruction of a query (making sense out of what the searcher wants to know) and automated deconstruction of the content to be indexed and searched. Good semantic search, therefore, depends on excellent ontologies.

To see a very simple example of an ontology related to “roadway”, check out this image. Keep in mind that before you aspire to implementing a semantic search engine in your enterprise, you want to be sure that there is a trusted ontology somewhere in the mix of tools to help the search engine retrieve results relevant to your unique audience.

W3C Opens Data on the Web with SPARQL

W3C (The World Wide Web Consortium) announced the publication of SPARQL, the key standard for opening up data on the Semantic Web. With SPARQL query technology, pronounced “sparkle,” people can focus on what they want to know rather than on the database technology or data format used behind the scenes to store the data. Because SPARQL queries express high-level goals, it is easier to extend them to unanticipated data sources, or even to port them to new applications. Many successful query languages exist, including standards such as SQL and XQuery. These were primarily designed for queries limited to a single product, format, type of information, or local data store. Traditionally, it has been necessary to formulate the same high-level query differently depending on application or the specific arrangement chosen for the relational database. And when querying multiple data sources it has been necessary to write logic to merge the results. These limitations have imposed higher developer costs and created barriers to incorporating new data sources. The goal of the Semantic Web is to enable people to share, merge, and reuse data globally. SPARQL is designed for use at the scale of the Web, and thus enables queries over distributed data sources, independent of format. Because SPARQL has no tie to a specific database format, it can be used to take advantage of “Web 2.0” data and mash it up with other Semantic Web resources. Furthermore, because disparate data sources may not have the same ‘shape’ or share the same properties, SPARQL is designed to query non-uniform data. The SPARQL specification defines a query language and a protocol and works with the other core Semantic Web technologies from W3C: Resource Description Framework (RDF) for representing data; RDF Schema; Web Ontology Language (OWL) for building vocabularies; and Gleaning Resource Descriptions from Dialects of Languages (GRDDL), for automatically extracting Semantic Web data from documents. SPARQL also makes use of other W3C standards found in Web services implementations, such as Web Services Description Language (WSDL). http://www.w3.org/

Web 2.0, 3.0 and so on

The recent Web 2.0 conference predictably accelerated some prognostication on Web 3.0. I don’t think these labels are very interesting in themselves, but I do admit that the conversations about what they might be, if they had a meaningful existence, expose some interesting ideas. Unfortunately, they (both the labels and the conversations) also tend to generate a lot of over-excitement and unrealistic expectations, both in terms of financial investment and doomed IT strategies. Dan Farber does his usual great job of collecting some of the thoughts on the recent discussion in “Web 2.0 isn’t dead, but Web 3.0 is bubbling up“.

One of the articles Dan links to is a New York Times article by John Markoff, where John basically equates Web 3.0 with the Semantic Web. Maybe that’s his way of saying very subtly that there will never be a Web 3.0? No, he is more optimistic. Dan also links to Nick Carr’s post welcoming Web 3.0, but even Carr is gentler that he should be.

But here’s the basic problem with the Semantic Web – it involves semantics. Semantics are not static, language is not static, science is not static. Even more, rules are not static either, but at least in some cases, syntax, and logical systems have longer shelf lives.

Now, you can force a set of semantics to be static and enforce their use – you can invent little worlds and knowledge domains where you control everything, but there will always be competition. That’s how humans work, and that is how science works as far as we can tell. Humans will break both rules and meanings. And although the Semantic Web is about computers as much (or more) than about humans, the more human-like we make computers, the more they will break rules and change meanings and invent their own little worlds.

This is not to say that the goal of a Semantic Web hasn’t and won’t generate some good ideas and useful applications and technologies – RDF itself is pretty neat. Vision is a good thing, but vision and near-term reality require different behavior and belief systems.

W3C Releases RDF/Topic Maps Interoperability Working Draft

The World Wide Web Consortium (W3C) Semantic Web Best Practices and Deployment Working Group has released the First Public Working Draft of A Survey of RDF/Topic Maps Interoperability Proposals. The document is a starting point for establishing standard guidelines for combined usage of the W3C RDF/OWL family and the ISO family of Topic Maps standards. The group expects to publish Survey and Guidelines Working Group Notes based on this draft. www.w3.org/TR/2005/WD-rdftm-survey-20050329/, www.w3.org/2001/sw/

« Older posts