From: [email protected]
Date: 02/24/04
Hi everyone, Attached (text doc), please find my notes from today's telecon. thanks, Said JC TELECON: Tuesday, February 24, 2004. --------------------------------------- Participants: Mike Harold Sandro Peter Said Pat Benjamin Regrets: Ian, Deb Announcements ------------- 1. Harold: A new use case with over 2000 facts for a regional business rules analysis application in New Brunswick (NRC Semantic Web Lab) ToDo: Harold will send us the link 2. Glen Newton of NRC sent the updated proposal to Vivien (W3) today. ------------------------------------------------------------------------------- Notes related to the discussion on SWRL Built-Ins. The document has been updated by Said and Mike (latest update in CVS). Section 8.7 ----------- [Sandro]: suggets to remove the distinction (functions/operators). head and tail (link to OWL or better to RDF) [Pat] Why not use RDF instead of having a complete new system? [Pat] Counter argument: OWL-to-RDF embedding uses lists in a particular way. It could be problematic if SWRL uses it differently All agree. [All] remove first and rest from the actual list Section 8.8 ----------- Move it to the issues list document. Same for Section 8.9 [Mike] Sent an email about constructors and accessors. No comments yet. Said agrees with Mike's proposal Section 8.5 ----------- [Pat] Questioning the list and some of the operators. Suggested to look into Mac OS and its use of calendaring conventions. Pat will send a link and Said will look into incorporating such conventions. [Mike] simplifying the built-ins list with overloading some of them: instead of having numeric equals, date equals, string equals, etc...let's have just equals, lessthan, greaterthan, etc. [Said] the overloading follows OO conventions and simplifies things. [Pat and Sandro] Need to be clear about what to expect as returned values. Example: "3 lessthan a". System needs to generate an error for mismatch. [Pat]: a logical way to think of the overloading is as a statement of identity and query. [] what does the implementation of the built-in deliver? [Pat,Mike] A type mismatch rather than just True or False. [All] the system will be required to check for illegal overloading. Section 8.1 ----------- [Sandro] suggest to Add type predicate [Said] yes agreed [Pat] We should consider looking at RDF data types. Why not take advantage? No need for a new datatype handling. [Mike]both SWRL datatype and RDF datatype refer to XML schema Part II [Pat] connection to RDF? [Sandro] use of URIs [Mike] but also the knowledge base (match against RDF data) [Pat] Data types: why not just use XQuery why 'swrlo' namespace? [Mike] XQuery does not provide a namespace. Also, issue with function/operator name conventions. [Pat] string diff. Should we keep it? [Benjamin] encourage people to use it. [Pat] may be an issue with implementors... [Sandro] it is underspecified. [Said] We can make it optional [All] suggest to move this to the extensions list (in issues list document along with 8.8 and 8.9) [All] Need to clarify the list and distinguish between primitive and derived builtins. [Benjamin]Should discuss this. Could be an issue for forward reasoning engines. [Pat] general comment. Issues viewing the documents with some fonts. [Pat] Math operators...need to clarify and be more specific with the types. dro, Pat] How about unary plus? [All] hesitating to keep or remove. Finally decided to keep it in. Doesn't hurt...no impl required. [Sandro] in the heading provide an explanation about the mapping. [All] agreed Action Items: ------------- Said and Mike will make the changes and update the document. Pat will help with section 8.1
This archive was generated by hypermail 2.1.4 : 02/24/04 EST