OWL/CNL

Authors

Bijan Parsia, University of Manchester

Introduction

Editor's Note: Let's make the background reasonable but not overwhelming

ISWC 2008 CNL Group meeting

This is a summary of the meeting as written by Ann Cregan.

1. Since Rabbit and Sydney Syntax are conceptually quite close, we will move towards convergence (a core syntax). In practical terms this will be led by Rabbit (Catherine and Glen) since Anne and Tommie are not currently available to work on CNL development. However we would like to still be "in the loop" and may offer comments from time to time. I have not had a chance to talk to Rolf about to what extent he wants to continue his involvement, so Rolf, can you please respond on this? The idea would be to work together to agree the core syntax, in combination with user testing to ensure the features chosen is what actually works best in practice.

2. The approach will take the meta-level path of using words like "symmetric" and "transitive" rather than trying to do everything at the object level (which often leads to cumbersome use of variables.) We accept that the users will have to learn some new words. However, we think that if they are really going to step up to the task of ontology building, they need to learn these concepts anyway. We will support users to understand the words by providing examples and explanations within the user interface.

3. Since ACE is much more expressive that Rabbit/SOS, the approach there will be to try to have the core CNL as a subset of ACE, so that people may extend easily into ACE where appropriate. Catherine and Glen will need to work with Kaarel to work out a suitable mapping.

4. The core part of the CNL will be spec'd and put forward as a W3C member submission ASAP.

5. Bijan has taken an action item to set up a new wiki page for co-ordinating this work

6. We will reach out to identify some new users for testing the core syntax. We have identified Michel Dumontier as a good person to approach, also Jean-Sebastien Brummer who was at the meeting volunteered for some user testing. It would be a good idea to identify testers in several different domains so we get a good spread.

Syntax

Preliminaries

Editor's Note: Stolen from The OWL 2 syntax doc. We need to check with the W3C that this is ok.

Documents containing OWL 2 ontologies written in OWL/CNL syntax are sequences of Unicode characters [UNICODE] and SHOULD use the UTF-8 encoding [RFC3629].

The OWL/CNL syntax of OWL 2 is defined using a standard BNF notation, which is summarized in Table 1.

The grammar does not explicitly show white space. White space is allowed between any two terminals or non-terminals except inside prefix, reference, full-IRI, nonNegativeInteger, quotedString, languageTag, and nodeID. White space is required between two terminals or non-terminals if its removal could cause ambiguity. Generally this means requiring white space except before and after punctuation (e.g., commas, parentheses, braces, and brackets).

White space is a sequence of blanks (U+20), tabs (U+9), line feeds (U+A), carriage returns (U+D), and comments. Comments are maximal sequences of Unicode characters starting with a '#' and not containing a line feed or a carriage return. Note that comments are only recognized where white space is allowed, and thus not inside the above non-terminals.

Mapping to OWL

The current Rabbit syntax is shown here and is available for comment by the group. In particular differences between Rabbit and ACE that may mean ACE cannot be seen as a superset of Rabbit should be highlighted.

Note the syntax is not yet shown in the BNF style indicated above.

A summary of Rabbit including the semantics can be found here.

Glen Hart

References