W3C launches the Web Ontology Working Group

Background

An ontology, in computing, is an explicit description of the concepts in a domain and the relations between them: which classes exist, which properties their members have, and which classes sit inside which others. Written in a machine-readable language, such a description lets software do more than match keywords. If a vocabulary states that every professor is a person employed by a university, and a dataset records that someone is a professor, a program that understands the vocabulary can draw conclusions the data never spelled out. The W3C chartered the Web Ontology Working Group, known as WebOnt, to produce a standard language of this kind for the Web, and the Recommendation it produced was developed as part of the W3C Semantic Web Activity ^[2]. That activity is the W3C's programme for making data on the Web directly processable by machines, rather than only readable by people.

The group did not start from a blank page. Its charter cites the DAML+OIL W3C Note as an important influence and as the starting place for the group's deliberations ^[1]. In other words, an existing W3C Note describing an ontology language was taken as the point of departure, and the working group's job was to turn that material into a reviewed, standardised language for the Web.

What happened

The working group's own page records 1 November 2001 as the date the group started ^[1]. From there, its published schedule set out a sequence of milestones.

The first was a face-to-face meeting in January 2002 at Lucent, in New Jersey ^[1]. Face-to-face meetings matter in standards work because most of the rest of the process runs over calls and email; a week of sitting in the same room is often where disagreements about the shape of a language get settled.

Next came the first working drafts: the abstract syntax, reference and guide documents were scheduled for 29 July 2002 ^[1]. These three answer different questions. An abstract syntax defines the formal shape of the language, independent of how it is written down. A reference document catalogues the constructs and what each one means. A guide explains to a newcomer how to use the language and when to reach for which construct.

The schedule then lists 31 March 2003 for last call on the requirements, guide, semantics, reference and overview documents ^[1]. Last call is the point at which a specification is frozen for outside review: the working group asks the wider community to read the text and object, before the design hardens. By this stage the deliverables had grown to five documents, adding a statement of requirements, a formal semantics, and an overview for general readers to the three earlier drafts.

The endpoint came on 10 February 2004, when OWL became a W3C Recommendation ^[1]. A Recommendation is the final stage of the W3C process: the specification is published as a stable standard that others can implement against, rather than as a draft that may still change. Measured from the group's start, this took about two years and three months.

The group's life ended soon afterwards. A note dated 28 May 2004 says that as of Monday the 31st of May the working group would officially come to an end, having achieved all it was chartered to do ^[1]. Closing on completion of the charter, rather than drifting on, is how the W3C process is meant to work.

Why it matters

The value of a standard ontology language is that vocabularies can be shared. Without one, two organisations wanting to publish overlapping descriptions of genes, or of library holdings, or of products, each invent their own format, and software written for one cannot read the other. With one, a vocabulary published by one group can be read and reused by another, and tools built to process the language work across domains.

That is what makes the formal parts worth the effort. The abstract syntax and the semantics are not decoration: they pin down what a construct means, so that two independent implementations can agree on what follows from a set of statements. The guide and the overview, meanwhile, exist so that people who will never read the semantics can still use the result.

OWL was the name the working group's effort produced, and the Recommendation is the artefact that outlived the group. The working group itself ended in May 2004 because its chartered work was done ^[1]; what remains is the published specification, developed within the W3C Semantic Web Activity ^[2], which other people could then build on without needing the group to still exist.