IRC log of dawg on 2004-04-23

Timestamps are in UTC.

06:41:09 [RRSAgent]
RRSAgent has joined #dawg
06:42:14 [Zakim]
Zakim has joined #dawg
06:49:21 [Yoshio]
Yoshio has joined #dawg
06:50:58 [DanC_AMS]
RRSAgent, pointer?
06:50:58 [RRSAgent]
06:56:29 [dirkx]
dirkx has joined #dawg
07:00:07 [Yoshio]
Goede morgen!
07:02:32 [alberto]
alberto has joined #dawg
07:02:54 [dirkx]
dirkx has joined #dawg
07:06:54 [HiroyukiS]
HiroyukiS has joined #DAWG
07:11:01 [Yoshio]
23:00 is reasonable, but does it mean 24:00 in winter time?
07:15:05 [DanC_jam]
DanC_jam has joined #dawg
07:15:06 [Yoshio]
7:30 am? hmm, I have to make my children prepared for the schools...
07:15:26 [thompsonbry]
thompsonbry has joined #dawg
07:18:33 [dirkx]
Tuesday 14.00 to 15.30 avoids DAWG and is not too baxd for everyone.
07:19:01 [alberto]
07:27:38 [JanneS]
JanneS has joined #dawg
07:40:10 [kendall]
kendall has joined #dawg
07:40:14 [DanC_jam]
yeah... there's some cool work at u. washington on scheduling with RDF.
07:40:27 [DaveB]
DaveB has joined #dawg
08:08:56 [DaveB]
08:11:10 [Yoshio]
Yoshio has joined #dawg
08:15:43 [kendall]
08:20:56 [Yoshio]
08:22:24 [Yoshio]
Yoshio's homework -->
08:22:49 [AndyS]
AndyS has joined #dawg
08:24:09 [kendall]
yoshio: oh, good, i'll look carefully at that
08:27:32 [AndyS]
2005 TP : February-March 2005, New England :
08:27:55 [ericP_]
ericP_ has joined #dawg
08:29:57 [JosD]
JosD has joined #dawg
08:32:05 [HiroyukiS]
HiroyukiS has joined #DAWG
08:33:26 [DaveB]
kendall discusses updated usecases doc\
08:33:30 [DaveB]
08:33:46 [DaveB]
took the 8 UCs with the most votes and tried to do the big rewrites, not everything yet
08:33:57 [DanC_jam]
(er... $Revision$???)
08:33:57 [DaveB]
struck all the benefits into the new section 4
08:34:03 [DaveB]
moved the TR canidates to sec 5
08:34:21 [DaveB]
CVs 1.25
08:34:39 [DanC_jam]
Revision 1.25 / (download) / (as text) - annotate - [select for diffs] , Fri Apr 23 08:13:40 2004 UTC (20 minutes, 49 seconds ago) by dbeckett2
08:34:40 [DaveB]
revision 1.25
08:34:40 [DaveB]
date: 2004/04/23 08:13:40; author: dbeckett2; state: Exp; lines: +656 -614
08:34:58 [DaveB]
a cleaner version by next Friday
08:35:01 [DanC_jam]
KC: draft by this coming Friday is likely
08:35:04 [DaveB]
30th April
08:35:21 [DaveB]
RS: next telcon Tues?
08:36:04 [DaveB]
Next telcon proposal 4th May 2004
08:36:19 [DaveB]
KC: intend to aim for draft for end May
08:36:34 [DaveB]
... like to have it ready around telcon of 18th
08:36:39 [DanC_jam]
(we're RESOLVED to meet 4 May. that's a refinement of the record of the decision we made earlier)
08:37:04 [DaveB]
AS: note that www2004 is mid may 17-20
08:37:28 [DaveB]
RS: would like to see drafts as soon as possible, even incomplete
08:37:33 [AndyS]
me : Who will be at WWW? I will (Wed-Sat)
08:37:43 [DaveB]
KC: my practice is to check in early, often
08:38:03 [kendall]
i won't
08:38:17 [DaveB]
ACTION: AndyS, EricP to review kendallC draft by 4 May telcon
08:39:49 [DaveB]
DanC - hypertext style - headings as targets of links, move the email pointers to seealso. maybe
08:40:07 [DaveB]
RESOLVED: kendall has done a great job
08:41:02 [kendall]
---Discussion of Candidate Technical Requirements---
08:41:26 [DanC_jam]
08:42:38 [DanC_jam]
q+ to suggest "tri-colored queries"
08:42:52 [SteveH]
SteveH has joined #dawg
08:42:53 [ericP_]
q+ to demo queuing
08:42:57 [DanC_jam]
08:43:12 [DanC_jam]
q+ danc2 to say "blueberry queries"
08:43:16 [DanC_jam]
ack danc_jam
08:43:16 [Zakim]
DanC_jam, you wanted to suggest "tri-colored queries"
08:43:22 [DanC_jam]
ack ericp
08:43:22 [Zakim]
ericP_, you wanted to demo queuing
08:43:24 [DanC_jam]
ack danc2
08:43:24 [Zakim]
danc2, you wanted to say "blueberry queries"
08:44:00 [ericP_]
q+ demo queue removal
08:44:02 [ericP_]
08:44:10 [ericP_]
q+ to demo queue removal
08:44:12 [ericP_]
08:44:25 [Phandaal]
Phandaal has joined #dawg
08:44:27 [DanC_jam]
poll: how many support "Queries supporting datatypes xsd:integer and xsd:float with operations: less than, equal, greater than"?
08:44:38 [DanC_jam]
pollR1: how many support "Queries supporting datatypes xsd:integer and xsd:float with operations: less than, equal, greater than"?
08:44:42 [DanC_jam]
R1 +1
08:44:53 [SteveH]
R1 +1
08:44:53 [ericP_]
R1 +1
08:44:56 [JosD]
R1 +1
08:45:01 [Phandaal]
R1 +1
08:45:03 [DaveB]
R1 +1
08:45:04 [alberto]
R1 +1
08:45:04 [Yoshio]
R1 +1
08:45:04 [Phandaal]
Phandaal has left #dawg
08:45:16 [AndyS]
xsd:string needed?
08:45:27 [AndyS]
and something about dates?
08:45:28 [alberto]
xsd:date also nice to have
08:45:51 [Phandaal]
Phandaal has joined #dawg
08:46:10 [kendall]
+1 xsd:string and xsd:date
08:46:19 [Phandaal]
Phandaal has left #dawg
08:46:37 [janne]
janne has joined #dawg
08:46:40 [kendall]
comparing 2 variables harder than comparing variable and a constant
08:46:54 [JFBaget]
JFBaget has joined #dawg
08:46:56 [ericP_]
(quoth RobS)
08:47:05 [kendall]
andys: will make a diff in performance, but not really implementational complexity
08:47:30 [kendall]
users don't understand why "x < y" is more expensive than "x < 18"
08:47:35 [JosD]
q+ implementation experience with math:lessThan
08:47:45 [kendall]
(much, much more expensive)
08:48:13 [kendall]
Action: SteveH to draft and maintain a list of test sketch cases, including x < y, x < 18
08:49:10 [kendall]
ACTION SteveH: to draft and maintain a list of test sketch cases, including x < y, x < 18
08:49:32 [DanC_jam]
Howard: beware of comparisons across types. coersion, etc.
08:49:57 [kendall]
SteveH: agrees with Andy re: comparisons of var to constant
08:50:04 [JosD]
q+ to say that implementation experience with math:lessThan
08:50:09 [DanC_jam]
ack josd
08:50:09 [Zakim]
JosD, you wanted to say that implementation experience with math:lessThan
08:50:13 [kendall]
EricP: agnostic as to adding 2 var comparison to language
08:50:50 [kendall]
JosdR: implementation experience w/ these comparisons, hasn't run into performance difficulties.
08:51:19 [kendall]
RobS: in some forms of inferencing, 2-var comparisons much harder to reason about; OWL has problem expressing this kind of thing.
08:51:31 [ericP_]
q+ to ask RobS if this applies to query
08:51:36 [kendall]
RobS: rule langs make it easier to express these
08:52:06 [DaveB]
for myself, I've not looked at performance issues with this area, nothing to report yet.
08:52:24 [janne]
janne has joined #dawg
08:52:27 [kendall]
anyone get that last bit from Jos?
08:52:38 [janne]
q+ other datatypes than just xsd:integer and xsd:float?
08:52:56 [DanC_jam]
q+ janne
08:53:00 [DanC_jam]
ack ericp
08:53:00 [Zakim]
ericP_, you wanted to ask RobS if this applies to query
08:54:14 [SteveH]
08:54:32 [DanC_jam]
ack janne
08:54:56 [SteveH]
08:55:21 [kendall]
Janne suggests adding dates and date-times to list of datatypes
08:55:58 [alberto]
+1 for dates
08:56:12 [kendall]
Janne has implemented date comparisons in his QL.
08:56:34 [kendall]
AndyS gets many requests for such things.
08:56:48 [kendall]
AlbertoR gets lots of reqs too.
08:57:32 [kendall]
AlbertoR uses data types in RSS 1.0 feeds
08:57:43 [DaveB]
08:57:53 [ericP_]
q+ to mention SQL use of dates
08:58:06 [DanC_jam]
ack daveb
08:58:20 [JFBaget]
An other example would be the travel organization (train B must be after train A).
08:58:27 [kendall]
Andy doesn't implement it because RDQL predates datatypes in RDF; doesn't have a convenient datatype maths library to bundle in.
08:58:37 [DanC_jam]
q+ steve
08:58:49 [DanC_jam]
ack steve
08:58:49 [kendall]
SteveH hasn't had any requests; his users use ISO date strings for dates.
08:58:51 [DanC_jam]
ack ericp
08:58:51 [Zakim]
ericP_, you wanted to mention SQL use of dates
08:59:08 [kendall]
EricP people tend to use date types in SQL
08:59:09 [alberto]
agrees with Andy about the XSD story - good DAWG work will cosider basic scalar data types
08:59:17 [JosD]
kendall: it was about a typical implementation of resolution i.e. unification of terms and that takes care of the constants, vars etc
08:59:36 [kendall]
ah, thx
08:59:36 [dirkx]
JFBaget: date/time ok - but strong concepts like 'at this time' or leaves _after_' or 'time datatype' is more complex than that - as it often also needs a location; i.e. something can be at the same moment in time; yet have a different time stamp due to time zone's, date lines and the point of the observer.
09:00:11 [AndyS]
q+ to discuss plain literals
09:00:17 [DanC_jam]
poll: who supports DATATYPE-DT?
09:01:20 [DanC_jam]
poll shows 9
09:01:41 [DanC_jam]
no: 3
09:01:42 [kendall]
poll: Bad idea to critical-path dates
09:01:50 [kendall]
poll shows 3
09:04:16 [dirkx]
So the NO does not means -against- it - it means 'worry about having it on the critical path'.
09:04:36 [ericP_]
q+ to ask if anyone thinks they've implemented this
09:04:48 [AndyS]
09:05:09 [kendall]
perhaps i miscribed the 2nd poll, but i'm confused now too.
09:06:21 [dirkx]
yes count: supports the issue. no count: worries about seeing the issue affect the critical path (as opposed to no meaning 'does not support the issue').
09:06:37 [kendall]
RobS: you can write a query that uses an arbitrary datatype, but we support only a very limited form of reasoning about it (whether the lexical forms are identical or not...?)
09:09:56 [kendall]
RobS wouldn't support the requirement to express queries as URLs
09:11:04 [kendall]
poll support requirement for queries as urls
09:11:07 [kendall]
poll shows 9
09:11:15 [kendall]
1 opposed
09:11:28 [ericP]
ericP has joined #dawg
09:14:32 [DaveB]
what are result templates?
09:15:15 [AndyS]
Constructing RDF from a variables and a request supplied form (c.f. SeRQL CONSTRUCT and rule (fwd))
09:15:21 [SteveH]
09:15:38 [kendall]
q+ what are result templates?
09:15:46 [kendall]
q+ to ask what are result templates?
09:15:52 [SteveH]
09:16:17 [JFBaget]
a) Queries using the media type of a document at a particular URL
09:17:02 [JFBaget]
b) Use that media type to give the result to the user (picture?) instead of the URL itself.
09:17:11 [kendall]
poll for support for "results in user-selectable result template" -- 10 against
09:17:56 [AndyS]
q+ to Howard
09:18:11 [kendall]
09:18:58 [AndyS]
query results to be user-definable
09:19:08 [AndyS]
(howard suggests)
09:20:12 [alberto]
about result templates/contructors SeRQL has implemented some for of those see
09:20:45 [kendall]
support for "results in user selectable Internet Media Types" -- 5
09:20:48 [kendall]
oppose 4
09:21:44 [ericP_]
09:21:59 [AndyS]
09:25:25 [AndyS]
Charter says "selecting instances of subgraphs from an RDF graph"
09:26:51 [AndyS]
09:28:58 [AndyS]
09:36:18 [AndyS]
q+ to say that "reports" could mean SUM() SORT() etc
09:39:22 [AndyS]
09:44:29 [kendall]
Query results can be formulated in such a way that graph traversal is not required.
10:01:16 [ericP_]
RDF query with reification:
10:01:59 [ericP_]
RDF query as a "direct graph":
10:02:04 [AndyS]
Will need reification for general case else problems with var/constants
10:02:09 [kendall]
ACTION Janne: write a use case to motivate the requirement of having queries written in RDF
10:02:55 [AndyS]
QEL uses reification at higher levels, does it not?
10:07:54 [janne]
Kendall, I said our use cases do not include ones that support the RDF encoded query. Now that i'm opposing its inclusion in requirements, I don't think I make the effort to motivate its inclusion again.
10:10:37 [kendall]
janne: okay, so i guess yr action item is discharged. :>
10:34:32 [DaveB]
updated CVS 1.4 with results of morning
10:58:33 [thompsonbry]
thompsonbry has joined #dawg
10:59:49 [kendall]
hi bryan
11:11:34 [thompsonbry]
hello kendall.
11:18:12 [thompsonbry2]
thompsonbry2 has joined #dawg
11:38:27 [thompsonbry2]
Can someone setup the projector for the PDF of my presentation? The URI is
11:41:34 [Zakim]
Zakim has left #dawg
11:42:53 [DanC_jam]
DaveB's laptop is hooked up to the projector, and I saw him flipping thru your slides earlier, thompsonbry2
11:43:14 [Zakim]
Zakim has joined #dawg
11:43:20 [DanC_AMS]
Zakim, this will be dawg
11:43:20 [Zakim]
SW_DAWG()8:00AM has been moved to #dawg by DanC_AMS
11:43:21 [Zakim]
ok, DanC_AMS; I see SW_DAWG()8:00AM scheduled to start in 17 minutes
11:43:34 [thompsonbry2]
Cool. Do you know if he has the updates from this morning? Or the one from yesterday?
11:43:41 [DanC_AMS]
no, I don't know
11:48:15 [alberto]
alberto has joined #dawg
11:53:58 [Zakim]
SW_DAWG()8:00AM has now started
11:54:05 [Zakim]
11:55:42 [thompsonbry2]
Dirk, can you dial into the bridge?
11:57:49 [Zakim]
11:57:49 [DanC_AMS]
alberto's trying it...
11:58:19 [DanC_AMS]
Zakim, ??P1 is Leiden
11:58:19 [Zakim]
+Leiden; got it
11:58:36 [DanC_AMS]
Zakim, Leiden holds DanC, Alberto
11:58:36 [Zakim]
+DanC, Alberto; got it
11:58:47 [alberto]
can you hear us?
11:58:59 [thompsonbry2]
I hear a lot of voices, nothing clearly.
11:59:19 [DanC_AMS]
I'm losing my voice; not sure how I'm gonna chair this afternoon.
11:59:32 [alberto]
11:59:38 [thompsonbry2]
Is that why I can't hear you? ;)
11:59:48 [DanC_AMS]
12:00:30 [thompsonbry2]
The presentation is:
12:00:41 [thompsonbry2]
This is an update from the first posted version.
12:01:22 [AndyS_scribe]
AndyS to scribe
12:01:32 [DanC_AMS]
Zakim, who's on the phone?
12:01:32 [Zakim]
On the phone I see BryanT, Leiden
12:01:33 [Zakim]
Leiden has DanC, Alberto
12:01:43 [JosD]
JosD has joined #dawg
12:01:51 [DanC_AMS]
(double-checking version of slides)
12:02:58 [AndyS_scribe]
BryanT presents
12:03:29 [AndyS_scribe]
Xpointer and HTTP Range for extensible addressible system for RDF
12:03:32 [DanC_AMS]
slide 1 of 15: "XPointer and HTTP Range"
12:03:38 [DaveB]
byran: can you say "slide 2" etc. when you want to turn over
12:03:41 [AndyS_scribe]
Today : #frag isn't passed
12:04:11 [AndyS_scribe]
XPointer today is extensible, not scalable
12:04:36 [AndyS_scribe]
Slide 3
12:04:44 [DanC_AMS]
slide 3 of 15: "Normal HTTP Request & Response"
12:04:52 [DanC_AMS]
(we were out of sync on slide 2)
12:05:14 [AndyS_scribe]
GET RDF/XML via accept:
12:05:27 [DanC_AMS]
(andy, you don't need to copy stuff that's on the slides)
12:05:35 [dirkx]
Byte-ranges: is any of such foreseen to have semantic meaning in this context
12:06:04 [AndyS_scribe]
Dirk: use of bytae ranges?
12:06:12 [AndyS_scribe]
Bryan: will cover later
12:06:20 [AndyS_scribe]
Slide 4 of 15
12:06:23 [DanC_AMS]
slide 4 of 15 "W3C XPointer Framework 1.0"
12:08:00 [AndyS_scribe]
Slide 5 : HTTP "Range" header
12:09:08 [AndyS_scribe]
Slide 6 : RDF Data access w/ HTTP Range
12:10:07 [AndyS_scribe]
Slide 7 : Pros and Cons
12:11:15 [DanC_AMS]
how do clients choose which query language to use?
12:11:40 [AndyS_scribe]
Slide 8 : Pros and Cons
12:11:57 [DaveB]
are we on slide 8?
12:12:13 [kendall]
that was a bit weird :>
12:12:36 [AndyS_scribe]
Slide 9 : Browser Scenario
12:13:12 [dirkx]
dirkx; outstanding questions about ease to shoehorn this into existing lib's.
12:13:40 [dirkx]
dirkx: second question: /path/ element also possible if you accept ? query argument.
12:13:54 [Yoshio]
Yoshio has joined #dawg
12:14:31 [AndyS_scribe]
Next talk - questions queued
12:14:42 [thompsonbry2]
URI for the next presentation?
12:14:51 [AndyS_scribe]
Howard: Crash course on XQuery
12:15:02 [dirkx]
dirkx: outstanding question with respect to the cachability of ranges in the light of the current byte-range situation.
12:15:06 [kendall]
doesn't seem to be
12:16:06 [thompsonbry2]
dirkx: I suggest in the backing material the use of the Vary header for caching. Also, unlike byte ranges I would not suggest breaking transparency to combine "adjecent" sub-ranges.
12:16:40 [thompsonbry2]
Caching is going to be very difficult in any case when the triple store state is evolving.
12:17:53 [DaveB]
"there is always the promise technology will save us in another few seconds"
12:18:02 [AndyS_scribe]
Waiting - Dirkx Q to Bryan :
12:18:29 [AndyS_scribe]
How to get the range into header.
12:18:51 [AndyS_scribe]
Issue: caching
12:19:55 [AndyS_scribe]
With query string (foo?query) can be treated differently from foo/query (e.g. Google)
12:20:37 [SteveH]
(( ))
12:20:42 [kendall]
i'd like to confirm the claim that GET /foo?bar often gets uncached.
12:21:28 [kendall]
12:21:29 [dirkx] versus
12:21:50 [kendall]
the path encodings of the vars is cleaner, but -shrug-
12:22:13 [dirkx]
versus"> | API -> versus and RDF-Range: rdf(123) --> pass to libWWW
12:22:20 [AndyS_scribe]
Query string has been used for non-idempotent operations
12:22:28 [thompsonbry2]
We note one exception to this rule: since some applications have traditionally used GETs and HEADs with query
12:22:28 [thompsonbry2]
URLs (those containing a “?” in the rel_path part) to perform operations with significant side effects, caches
12:22:28 [thompsonbry2]
MUST NOT treat responses to such URIs as fresh unless the server provides an explicit expiration time. This
12:22:28 [thompsonbry2]
specifically means that responses from HTTP/1.0 servers for such URIs SHOULD NOT be taken from a cache. See
12:22:28 [thompsonbry2]
section 9.1.1 for related information.
12:22:39 [thompsonbry2]
(Section 13.9 of HTTP/1.1)
12:22:40 [AndyS_scribe]
Bryan quotes RFC
12:23:02 [DanC_AMS]
12:23:03 [DanC_AMS]
We note one exception to this rule: since some applications have traditionally used GETs and HEADs with query URLs (those containing a "?" in the rel_path part) to perform operations with significant side effects, caches MUST NOT treat responses to such URIs as fresh unless the server provides an explicit expiration time.
12:23:08 [DanC_AMS]
]]] --
12:23:15 [DanC_AMS]
12:23:20 [thompsonbry2]
12:23:57 [dirkx]
No worries -- is an break-out possible.
12:24:08 [DanC_AMS]
ACTION DanC: inform the TAG of the conflict between the HTTP spec and TAG's recommendation on issue 7
12:24:37 [dirkx]
Note that reality is a bit worse; i.e. heuristics which detect 'cgi-bin' etc are rife.
12:24:49 [AndyS_scribe]
Howard now ready ...
12:24:50 [DanC_AMS]
Zakim, remind us in 9 minutes that Howard's time is about up
12:24:50 [Zakim]
ok, DanC_AMS
12:24:51 [kendall]
12:25:12 [AndyS_scribe]
= XQuery: a whirlwind tour =
12:25:52 [dirkx]
However with the same token have the reverse risk.
12:26:04 [AndyS_scribe]
There are 7 node types in XML : book to Janne
12:26:33 [DanC_AMS]
(XQuery is actually *not* functional, I learned in an exchange witht the editors via
12:27:23 [DanC_AMS]
Howard's presentation materials:
12:27:32 [dirkx]
12:28:41 [DanC_AMS]
Howard: everything is a sequence; 1 is actually a singleton sequence
12:28:53 [AndyS_scribe]
s/=> 42/=> 24/
12:29:09 [AndyS_scribe]
in example "2", 3+1
12:32:20 [AndyS_scribe]
Syntactical can construct strange things
12:33:09 [AndyS_scribe]
<foo><bar/></foo>/bar evals to <bar/>
12:33:27 [AndyS_scribe]
That was "ripley query 2"
12:33:50 [Zakim]
DanC_AMS, you asked to be reminded at this time that Howard's time is about up
12:36:56 [AndyS_scribe]
DB people want to statically analyse the query
12:37:12 [thompsonbry2]
* That's why I wanted to start with a critique of XQuery ;)
12:39:45 [DanC_AMS]
Zakim, remind us in 9 minutes that Rob's time is about up
12:39:45 [Zakim]
ok, DanC_AMS
12:40:23 [DanC_AMS]
Rob: XQuery is a pretty simple functional langauge. the complication comes from the XPath navigation, axes, and such
12:40:33 [AndyS_scribe]
= Rob Shaerer: Using XQuery for RDF: costs and benefits of one approach =
12:40:59 [AndyS_scribe]
Good to use FLWOR for SQL databases
12:41:10 [AndyS_scribe]
Tried using XPath but moved to ...
12:41:10 [DaveB]
FLWOR = For, Let, Where, Order by, Return.
12:41:40 [DanC_AMS]
RobS: we started using XPath, but that turned out to be hard/complicated, so we went a different way...
12:41:48 [AndyS_scribe]
:related(subj, obj, pred, model) [syntax not exact]
12:42:07 [AndyS_scribe]
q+ to talk about Robie's work
12:42:36 [AndyS_scribe]
:individuals, :properties from other API calls
12:43:26 [kendall]
doesn't appear to be
12:43:35 [AndyS_scribe]
Results are formatted by XQUery inc. as RDF/XML
12:43:42 [kendall]
he's cat'ing files to STDOUT in linux xterm. Old Skool!!
12:44:23 [AndyS_scribe]
return is sequence in XQUery that is concat
12:45:21 [DanC_AMS]
"satisfies" is part of XQuery syntax?
12:46:28 [DanC_AMS]
RobS shows query4, which demonstrates disjunction
12:46:45 [AndyS_scribe]
disjunction by "or" in XQuery with base access into RDF
12:47:15 [AndyS_scribe]
q+ to ask about efficient implementation issues
12:47:42 [DanC_AMS]
(indeed, shows "satisfies" syntax. current draft: 12 nov
12:48:39 [AndyS_scribe]
NI want to query OWL which is why model is passed to :related
12:48:45 [Zakim]
DanC_AMS, you asked to be reminded at this time that Rob's time is about up
12:49:28 [AndyS_scribe]
ni:models() => ni:rdfModels -- all interpretations of ontology
12:49:46 [DanC_AMS]
hmm... ni:models() ... brain cramp.
12:50:44 [AndyS_scribe]
"true" defn as true in every interpretation
12:51:29 [AndyS_scribe]
Points about "some Model" and "every Model"
12:52:11 [Yoshio]
12:53:06 [Yoshio]
* cause the whole world is not certain or we can't konw everything :)
12:53:32 [DanC_AMS]
quite, but how is that relevant to this work?
12:55:06 [AndyS_scribe]
Summary: these operators can be used to build a query system (RobS to email presented queries)
12:55:18 [thompsonbry2]
Is it true in all "models" that someone did something? If this is reasoning in multiple worlds, then you are basically asking if there is any world (that I can conceive of) in which something is not true. So, you are testing assumptions. If you have an assumption, then you have uncertainty.
12:55:38 [thompsonbry2]
In general, crisp logics are not the best for reasoning about uncertainty.
12:55:47 [Yoshio]
we sometimes want the answer without specifying the possible world, don't we?
12:56:17 [Yoshio]
marginalized possible world could have uncertainty.
12:56:30 [thompsonbry2]
Sure. I am just saying that if you are trying to reason about uncertainty then using crisp logics is not the best course.
12:56:38 [AndyS_scribe]
Impl: canonicalise RDF and implement ni:related
12:57:27 [AndyS_scribe]
Jos: is canonical RDF solved?
12:57:46 [AndyS_scribe]
Rob: canonical means a std rep (not ordering etc)
12:58:00 [DanC_AMS]
ack andy
12:58:00 [Zakim]
AndyS_scribe, you wanted to talk about Robie's work and to ask about efficient implementation issues
12:58:48 [kendall]
thom: andy s asking about relation to jonathan robie's work 'syntatic web' from a few years ago
12:58:55 [AndyS_scribe]
q/andyS: Rel. to syntactic web (Robie)
12:59:06 [DanC_AMS]
cf The Syntactic Web
12:59:06 [DanC_AMS]
Syntax and Semantics on the Web
12:59:06 [DanC_AMS]
Jonathan Robie
12:59:09 [AndyS_scribe]
Rob: don't know Eric: its similiar
12:59:11 [DanC_AMS]
12:59:39 [AndyS_scribe]
Rob; users not interested in iterating over triples
13:00:11 [AndyS_scribe]
... they think of the assertions: not graph is a list of edges to be iterated over
13:00:17 [AndyS_scribe]
node centric
13:01:01 [AndyS_scribe]
DanC; similar but non trivial diffs
13:02:05 [AndyS_scribe]
Current impl: some optimisations : may combine with existing Xquery engine
13:03:24 [AndyS_scribe]
Rob: users do not think in terms of a graph (do not like current RDF APIs)
13:03:57 [Yoshio]
* then how?
13:04:20 [AndyS_scribe]
User thinks in nodes and relationships (OWLish view?)
13:04:31 [thompsonbry2]
Some users just want know when the milk in the fridge is running out or going bad.
13:05:17 [Yoshio]
What's the difference between graphs and nodes and relationships?
13:05:36 [thompsonbry2]
I.e., this is an issue of application encapsulation levels. RobS is thinking developers = users, right? Just like an SQL developer.
13:05:50 [kendall]
13:05:55 [Yoshio]
I see.
13:05:55 [AndyS_scribe]
Jos: can arcs be variables?
13:06:22 [AndyS_scribe]
Rob: had list all properties : not used and removed
13:06:47 [AndyS_scribe]
Rob: could be done but no user has asked for it.
13:09:28 [DanC_AMS]
(re XQUery being functiona, here's a pointer in to the exchange I had with the editors )
13:10:15 [thompsonbry2]
Who are these people who know X-Query?
13:10:35 [DanC_AMS]
well, there are 1500 last call comments. So somebody's reading the spec pretty carefully.
13:10:37 [kendall]
it's a good question; i guess many of them are NI clients.
13:10:59 [AndyS_scribe]
Rob: have one QL everyone knows and has invested in
13:12:17 [thompsonbry2]
Is this the same population? XQuery is not a simple thing. And doing inference over XQuery seems hard. I have done a lot of fancy things with SQL, but the scalability always breaks down and you need to use a different processing model eventually.
13:13:22 [DanC_AMS]
DanC: ni:models() has an implicit background document? RobS: yes, that's the norm in XQuery
13:13:24 [Yoshio]
Is tree structure suited/natural to express relationships?
13:14:34 [AndyS_scribe]
Rob: could have XML doc with each RDF model as subnodes but need to optimise in practice
13:14:40 [DanC_AMS]
thompsonbry2, why is XPointer relevant to the range thing? it's just a fragid syntax for RDF/XML, no? Where does XPointer come in?
13:16:50 [thompsonbry2]
Just the framework: specifies encoding for URI, processing model (left-to-right, first match is returned), extensible - people can define their own query languages. You loose some of that if you drop the XPointer framework and just map the RDF query onto the Range header, e.g., "Range: rdf=....".
13:17:17 [thompsonbry2]
Broader context, is I was thinking about sub-resource addressing as a general problem and RDF as a specific use case.
13:18:22 [thompsonbry2]
btw - I am NOT suggesting the use of the xpointer() scheme. I think a lot of people conflate the xpointer() addressing scheme (which is very complex and did not make it to Rec) with the XPointer Framework.
13:18:45 [AndyS]
AndyS has joined #dawg
13:18:47 [DanC_AMS]
"encoding for URI"? you mean how to URI-escape a string? you don't need XPointer for that, do you?
13:19:15 [DanC_AMS]
the left-to-right stuff premiered in XPointer. I can see that.
13:20:06 [thompsonbry2]
No. But it does specify how to do it (encoding) in a way that supports the namespace processing context and left-to-right processing order. The XPointer Framework is really dirt simple.
13:20:28 [AndyS_scribe]
disucssion on XQuery's role (which scribe lost)
13:20:49 [AndyS_scribe]
This appraoch works well with rest of XML tool chain
13:21:05 [DanC_AMS]
ok, so the use of the XPointer framework has nothing to do with XML elements/attributes and everything to do with fragment syntax stuff, thompsonbry2?
13:21:44 [thompsonbry2]
Yes. I would think that an RDF query language would be grounded in the RDF model, not the syntax. Should have made that point in the presentation.
13:22:00 [DanC_AMS]
maybe you did; maybe I just needed to hear it twice ;-)
13:22:27 [kendall]
Other XML-based media types are also encouraged to use this framework in defining their own fragment identifier languages.
13:22:39 [Zakim]
13:22:40 [thompsonbry2]
Only indirectly - by the choice of the query language notation in the example.
13:22:54 [thompsonbry2]
Whoops! Leiden just dropped the telcon.
13:23:12 [ericP]
bryan, just a sec...
13:23:26 [kendall]
dirk working on it
13:23:27 [thompsonbry2]
Re: Kendall, yes.
13:23:31 [ericP]
or top network engineers are on the case
13:23:54 [AndyS_scribe]
Rob: it is called "Rex" - RDF extensions for XQuery
13:25:10 [Zakim]
13:25:16 [DanC_AMS]
Zakim ??P1 is Leiden
13:25:28 [ericP]
Zakim, ??P1 is Leiden
13:25:28 [Zakim]
+Leiden; got it
13:25:33 [Yoshio]
*rex tremendae majestatis
13:25:59 [thompsonbry2]
And cannons are the last argument of kings.
13:26:02 [DanC_AMS]
"tremendae majestatis" translate pls?
13:27:02 [AndyS_scribe]
Jos: is the query over the RDF closure?
13:27:22 [thompsonbry2]
~ great majesty ~ I think
13:27:26 [kendall]
it's from the requiem
13:27:31 [kendall]
er, requiem mass
13:27:32 [ericP]
13:28:59 [DanC_AMS]
(Break thru 3:35 pm AMS time)
13:29:03 [DanC_AMS]
er... 3:45
13:29:05 [Zakim]
13:29:10 [Zakim]
13:29:11 [Zakim]
SW_DAWG()8:00AM has ended
13:29:12 [Zakim]
Attendees were BryanT, DanC, Alberto, Leiden
13:29:22 [DanC_AMS]
Brian, do you want to join the next sessino?
13:31:59 [eikeon]
eikeon has joined #dawg
13:32:15 [kendall]
hi dan
13:34:53 [thompsonbry2]
Dan, I can. I can't really hear the discussion over the phone though. Do you think
13:35:31 [thompsonbry2]
that we will return to the XPointer+Range thing in the general discussion? If so, then I can rejoin in 20min or so.
13:41:46 [Yoshio]
At the end of the day, I'm not sure how we can treat the non-tree structures within the XML representations...
13:44:01 [thompsonbry2]
Yoshio, I agree. Not the right layer. XML should be serialization and interchange, not the core processing language model.
13:45:40 [Yoshio]
Yes, although XQuery has many nice features...
13:53:12 [thompsonbry2]
kendall, do you have a URI for your presentation?
13:53:39 [kendall]
nope, i'm not giving it; i spent last night working on the UC document instead. (well, last night and a large part of this morning)
13:53:46 [dirkx]
dirkx has joined #dawg
13:54:20 [janne_scribe]
13:54:50 [janne_scribe]
RDQL: an implementation experience (just one?)
13:55:31 [DanC_AMS]
Zakim, remind us in 9 minutes to move on or something
13:55:31 [Zakim]
ok, DanC_AMS
13:55:41 [DanC_AMS]
(ooh... ah... keynote transitions)
13:55:42 [janne_scribe]
slide 2: what is RDQL
13:56:30 [janne_scribe]
RDFQL has SQL like syntax for querying RDF
13:56:43 [janne_scribe]
13:56:59 [janne_scribe]
slide 3: simple enough
13:57:48 [janne_scribe]
SFWAU structure
13:57:54 [janne_scribe]
slide 4: rdql example
13:58:57 [janne_scribe]
FROM clause fetches the model from the given URL
13:59:01 [janne_scribe]
slide 4: an implementation
13:59:05 [janne_scribe]
slide 5 that is
13:59:38 [janne_scribe]
13:59:44 [janne_scribe]
slide 6: needed extensions
14:00:02 [thompsonbry2]
q+ To ask, what about closure for the result sets (e.g., as RDF/XML)?
14:00:22 [janne_scribe]
slide 7: some issues
14:01:14 [janne_scribe]
question marks give headache as they are placeholders in ODBC/JDBC context
14:02:05 [janne_scribe]
working on evaluation algorithms by improving indexes
14:02:55 [janne_scribe]
indexes are built using known query patterns
14:03:32 [janne_scribe]
current merging does not consider bNodes correctly
14:04:31 [Zakim]
DanC_AMS, you asked to be reminded at this time to move on or something
14:04:35 [janne_scribe]
slide 8: an abuse case
14:06:06 [janne_scribe]
DaveB: what are the percents?
14:06:50 [janne_scribe]
SQL/Oracle type operator for substring matching
14:07:23 [Yoshio]
q+ to ask the role of USING part, can it be used to pose the premises?
14:07:49 [janne_scribe]
dirk: jdbc/odbc stack works nicely with current syntax
14:08:17 [AndyS]
Yoshio - no: USING is for namespace/prefix declarations
14:08:45 [Yoshio]
14:08:50 [thompsonbry2]
but that would be a nice feature -- is somethign true if you assume x, y, z.
14:09:14 [janne_scribe]
EricP: question about multiple FROM sections... (didn't get the end of the question)
14:09:34 [AndyS]
Would need transactions probably
14:09:37 [janne_scribe]
slide 9: our DAWG wishlist
14:10:10 [HiroyukiS]
HiroyukiS has joined #DAWG
14:10:16 [janne_scribe]
dirk: want to leverage existing infrastructure with rdql
14:10:19 [DanC_AMS]
Zakim, ??P24 is Leiden
14:10:19 [Zakim]
+Leiden; got it
14:11:00 [ericP]
ericP's question: can you enumerate which triples you wish to match at which source?
14:11:03 [ericP]
AndyS: yes. we were thinking of something like FROM...FOR...FROM...FOR.
14:11:51 [janne_scribe]
testing charter with the bottom line (insert, update, delete)
14:12:35 [janne_scribe]
RobS: summarization functions like count()?
14:12:42 [janne_scribe]
alberto: don't have it, would be nice
14:13:40 [janne_scribe]
steveh: do variables need be prefixed?
14:14:05 [janne_scribe]
AndyS: QNames are ok
14:14:13 [thompsonbry2]
14:14:50 [janne_scribe]
RobS: only simple OR -> can you do general disjunction?
14:14:59 [janne_scribe]
Alberto: current implementation due to efficiency
14:15:35 [janne_scribe]
haven't faced the need quite yet
14:15:55 [ericP]
ack thompsonbry
14:15:55 [Zakim]
thompsonbry, you wanted to ask, what about closure for the result sets (e.g., as RDF/XML)?
14:16:32 [janne_scribe]
thompsonb: no serialization to rdf/xml in the select clause
14:17:36 [AndyS]
c.f. SeRQL's CONSTRUCT (instead of SELECT) is an RDF template: substitute in variables to patterns
14:17:59 [DaveB]
see for SeRQL
14:18:00 [DanC_AMS]
Alberto: our output generally goes thru XSLT. [?]
14:18:17 [ericP]
AlbertoR: we return all the solutions in regular-ized RDF and that goes through an XSLT
14:18:19 [DanC_AMS]
... we use the result of the query as an XML source.
14:18:31 [DanC_AMS]
(did he say the solutions become RDF/XML?)
14:18:32 [ericP]
ack Yoshio
14:18:32 [Zakim]
Yoshio, you wanted to ask the role of USING part, can it be used to pose the premises?
14:18:33 [thompsonbry2]
got it.
14:18:37 [janne_scribe]
The result set is managed as a normalized rdf/xml via xslt
14:18:54 [Yoshio]
what if we specify a (URI for ) RDF graph to use as premises in the USING part?
14:19:58 [janne_scribe]
rdql = grep for rdf graphs - no premises for now
14:20:30 [DaveB]
DaveB will be showing
14:21:38 [janne_scribe]
jdr: can you have multiple FROMs
14:21:56 [janne_scribe]
alberto: yes you can, the semantics should be clarified
14:23:14 [janne_scribe]
AndyS: you could mimic premises by having premises at FROM clause 1 and model in FROM clause 2
14:23:37 [janne_scribe]
RobS: is this extensible to OWL/rules world?
14:24:07 [janne_scribe]
AndyS: focus on runtime evaluation where inference is supported via virtual triples
14:24:18 [thompsonbry2]
Then what about creating a data source to contain those premises? CREATE <source> FROM <query> (or INSERT ... INTO <source>)
14:25:06 [janne_scribe]
DaveB starting on impl. experience
14:25:31 [janne_scribe]
slied 1
14:27:08 [DanC_AMS]
(what's the difference between Redland and Raptor?)
14:27:16 [janne_scribe]
slide 2
14:27:41 [janne_scribe]
easy/medium/hard things described
14:28:52 [janne_scribe]
slide 3
14:30:26 [janne_scribe]
14:30:44 [janne_scribe]
a very quick demo
14:31:33 [thompsonbry2]
q+ to ask about evolution of api / QL to resource centric
14:31:46 [janne_scribe]
PatH not available for DQL talk
14:31:52 [ericP]
ack thompsonbry
14:31:52 [Zakim]
thompsonbry, you wanted to ask about evolution of api / QL to resource centric
14:32:02 [DanC_AMS]
(thanks EricP for chairing help)
14:32:21 [janne_scribe]
what does resource centric mean?
14:32:50 [janne_scribe]
triple store = uri (= resource centric)
14:33:01 [DanC_AMS]
agenda + wrap-up
14:34:29 [janne_scribe]
EricP? anything else before wrap-up?
14:36:13 [thompsonbry2]
14:36:16 [DanC_AMS]
ciao, thompsonbry2
14:36:18 [DaveB]
bye bryan
14:36:22 [alberto]
14:36:30 [Zakim]
14:37:18 [Yoshio]
Tot ziens!
14:38:18 [DanC_AMS]
(jann's action to motivate.... WITHDRAWN)
14:39:08 [janne_scribe]
I will keep the action about "not being able to use variables to span a correct graph" for a while (before I give up)
14:39:12 [Yoshio]
14:39:58 [alberto]
Yoshio: nice!
14:41:24 [alberto]
alberto has left #dawg
14:41:27 [janne]
janne has left #dawg
14:41:37 [Zakim]
14:41:38 [Zakim]
SW_DAWG()8:00AM has ended
14:41:39 [Zakim]
Attendees were BryanT, Leiden
15:11:06 [eikeon]
eikeon has joined #dawg
17:03:49 [Zakim]
Zakim has left #dawg
19:57:37 [eikeon]
eikeon has joined #dawg
21:37:22 [eikeon]
eikeon has joined #dawg
23:42:58 [eikeon]
eikeon has joined #dawg