IRC log of dawg on 2004-04-22

Timestamps are in UTC.

07:17:25 [RRSAgent]
RRSAgent has joined #dawg
07:18:12 [Yoshio]
Yoshio has joined #dawg
07:18:26 [DanC_AMS]
(break for network)
07:18:56 [AndyS]
AndyS has joined #dawg
07:19:46 [alberto]
alberto has joined #dawg
07:20:07 [Yoshio]
Hello there
07:20:22 [alberto]
Hi
07:20:30 [AndyS]
= DAWG F2F
07:20:31 [AndyS]
Dan: Hi!
07:20:31 [AndyS]
Present Alberto, Dirk, DanC, Yoshi, Hiroyuki Sato (NTT), Steve, DaveB, AndyS, Rob Shearer, Howard, Janne,
07:20:31 [AndyS]
Phone:
07:20:33 [AndyS]
Regrets: Bryan, Patrick,
07:20:35 [AndyS]
Scribe: AndyS
07:20:38 [AndyS]
== W3C Process /DanC
07:20:39 [AndyS]
Objective: Concensus on technology
07:20:41 [AndyS]
Finished when community agrees
07:20:43 [AndyS]
Working Drafts every 3 months
07:20:45 [AndyS]
Public Records
07:20:47 [AndyS]
Charter: WG can't unilaterally change it
07:20:49 [AndyS]
DaveB: Target: strawman decision??
07:20:51 [AndyS]
DanC: Not a big deal
07:20:57 [AndyS]
Important use cases, requirements
07:20:59 [AndyS]
Howard: origins of this WG?
07:21:01 [AndyS]
DanC: Community work (Alberto, Libby et al.) => WG
07:21:04 [AndyS]
Done by middle next year: (June 2005)
07:21:05 [AndyS]
== Use cases
07:21:17 [AndyS]
DanC: Will present at WWW2004 => will use Use Cases as illustration
07:21:17 [AndyS]
Similarly for the WG home page
07:21:17 [AndyS]
(network is up ....)
07:22:20 [SteveH]
SteveH has joined #dawg
07:23:11 [AndyS]
Jean-François arrives
07:24:20 [AndyS]
DanC asks for UC comments from around table
07:24:37 [AndyS]
Janne: Profium - SIR product
07:25:02 [AndyS]
Implemented UC 5.5 Multisource collection
07:25:29 [AndyS]
Serving info to different users
07:26:11 [AndyS]
Requires: queries on union of multisource (not model info)
07:26:51 [AndyS]
Have own QL (RDFQL)
07:27:20 [AndyS]
Query result as a graph, not variable binding, for further processing
07:27:33 [AndyS]
Do XSLT on query result
07:27:50 [AndyS]
++ Danc has questions, as had Steve
07:28:04 [AndyS]
s/had/has/
07:28:11 [AndyS]
Howard: Learning mode
07:28:46 [AndyS]
Data access and storage servers / figuring out usage
07:28:59 [js]
js has joined #dawg
07:29:13 [SteveH]
SteveH has joined #dawg
07:29:14 [DaveB]
DaveB has joined #dawg
07:29:18 [DanC_AMS]
Andy: ...
07:29:42 [DanC_AMS]
... several of the use cases look relevance... both [oops; I'm lost already]
07:30:27 [DanC_AMS]
... the need for a query language comes from dealing with large bodies of data; programmer time is used inefficiently unless/until you have a query language
07:30:49 [DanC_AMS]
... we haven't done much with provenance yet...
07:31:05 [DanC_AMS]
... We've done "find email address..." sort of stuff...
07:31:22 [DanC_AMS]
... [something about "finding a part" that I didn't get the gist of]
07:32:48 [AndyS]
Janne: Has similar example to 5.5.1 in media domain
07:32:58 [AndyS]
DanC; That makes it a different UC
07:34:27 [AndyS]
DaveB/dajobe: Colleagues with query people- Libby et al
07:34:39 [AndyS]
Find stuff about X - web of information
07:35:21 [AndyS]
Example: in ILRT - support services for University (its ILRT)
07:35:29 [AndyS]
Eg Distance learning
07:35:41 [HiroyukiS]
HiroyukiS has joined #DAWG
07:35:49 [AndyS]
Wander round vocabulary to explore info space
07:36:17 [AndyS]
Will be building a thesaurus search
07:36:32 [AndyS]
Howard arrives
07:36:32 [AndyS]
Jos arrives
07:37:11 [AndyS]
DaveB will write up this educational metadata UC
07:39:18 [AndyS]
(DanC does intros for new arrivals)
07:41:02 [AndyS]
DanC sumarises for Kendall as editor
07:41:45 [DaveB]
DaveB has joined #dawg
07:42:02 [AndyS]
EricP arrives
07:43:30 [AndyS]
Jos: Medical use case (Dirk) - UC 5.9.1
07:43:43 [AndyS]
Real world problem
07:43:51 [AndyS]
UC is ordering X-ray
07:46:12 [AndyS]
Requirements: sort order, optional triples, complex graph patterns (more than 1 edge)
07:47:30 [AndyS]
RobS: Seen uses in military, medical records, manufacturing, supply chain
07:47:43 [AndyS]
5.10.1 : find employees by group - OWL addition
07:48:11 [AndyS]
Need to note upper layers and provide same interface as base RDF layer
07:48:32 [dirkx]
dirkx has joined #dawg
07:49:19 [AndyS]
SteveH: Built 3Store: RDFS scalable store
07:49:34 [AndyS]
Users have too much info, varying quality
07:49:49 [AndyS]
Fluid datasets (per second change)
07:50:15 [AndyS]
Interests in provenance (small "p")
07:51:14 [AndyS]
Customers - aerospace, medical infomatics, jet engines
07:51:27 [JosD]
JosD has joined #dawg
07:52:00 [kendall]
kendall has joined #dawg
07:52:17 [Phandaal]
Phandaal has joined #dawg
07:52:23 [AndyS]
Hiroyuki: interest context in documents
07:54:07 [kendall]
dirkx: where is the WC?
07:54:33 [AndyS]
Yoshio: Home appliances : users in the home, people who sell the products, ourselves - manufacturer of the products
07:55:02 [AndyS]
UC: TV sets, mobile phone, car appliances (navigation)
07:55:17 [AndyS]
eg 5.1.3
07:55:27 [kendall]
daveb: thx!
07:55:38 [AndyS]
5.6.1 - motorcycle parts
07:56:14 [AndyS]
5.2.2 - reviews of products
07:56:52 [AndyS]
Offers UC from Japanese POV
07:57:54 [AndyS]
Jean-François: Likes 5.10.1
07:58:15 [AndyS]
Mixing languages
07:58:38 [AndyS]
on the semantic web
07:58:46 [DaveB]
(not human langs?)
07:59:08 [kendall]
dirkx: thx
07:59:13 [kendall]
heh
08:00:28 [DaveB]
there's no logging bot here
08:00:39 [DaveB]
oops, I see it
08:00:41 [AndyS]
DanC: Fav UC to sell work: media, news aggregation and product metadata
08:01:09 [AndyS]
Fav UC to personally motivate: email queries
08:01:33 [AndyS]
More general than IMAP ql
08:01:54 [AndyS]
Also last call issue tracking from scraped HTML docs
08:02:06 [AndyS]
Geospacial queries as well.
08:02:34 [AndyS]
Kendall: Like ones in doc (!! he's the editor)
08:02:59 [AndyS]
esp. ones that can live on the public web
08:03:45 [AndyS]
Users: intelligence community
08:04:11 [AndyS]
Kendall notes: there is are more UCs to go in
08:05:14 [AndyS]
Alberto: asemantics/RDFStore - Customers: Euro Space Agency, Ministry of Water (NL)
08:05:15 [kendall]
two from steveh (bandwidth efficiency & chunking -- IIRC) and one from path (owl)
08:05:38 [AndyS]
asemantics: Hybrid datasets in XML
08:06:02 [AndyS]
thin layer of RDF to provide common view of heterogeneous datasets
08:06:26 [AndyS]
Aggreggation, also extend QL for eg. geospacial queries
08:06:53 [AndyS]
UCs: tell me about X
08:07:04 [AndyS]
Need provenance from aggregations
08:07:49 [AndyS]
Scalability is an issue: e.g. 1000 results
08:08:08 [AndyS]
RDF query is ambition in practical terms
08:08:31 [AndyS]
Dirkx: Integrate with GIS, and its extensions.
08:08:44 [AndyS]
ESN (space): "partially enclosed"
08:08:53 [AndyS]
s/ESN/ESA/
08:09:26 [AndyS]
RDQL is like SQL makes it an easier sell to customers
08:09:43 [AndyS]
but does not solve all problems - needs extensions
08:09:51 [AndyS]
based on application domain
08:10:20 [AndyS]
Current use cases should exhibit these extension features
08:10:43 [AndyS]
EricP: Arbitrary graph query: esp 5.2.2. - News stories
08:10:54 [AndyS]
c.f. annotation server inc provenace
08:11:51 [AndyS]
Routinely use disjunction for evolution of DC properties
08:12:11 [AndyS]
Also "things that are about this"
08:12:26 [AndyS]
-- Once round room done --
08:12:41 [AndyS]
Restart at 10:30
08:24:59 [Yoshio]
Here's the URI for my additional use cases http://www.w3.org/2004/04/21-fuku/usecases
08:25:38 [Yoshio]
They may not be corresponding to what I said (^_^;)
08:26:11 [Yoshio]
I made them to cover some technologies I think needed and not covered by exsiting ones
08:27:27 [DanC_jam]
DanC_jam has joined #dawg
08:30:57 [JosD]
JosD has joined #dawg
08:31:47 [kendall]
10:30-12:30 DAWG Session
08:32:58 [ericP]
DanC: tell us what you want us to know about the use cases document
08:33:13 [ericP]
Kendal: i tried to describe the benifits but got tired
08:33:32 [JosD]
JosD has joined #dawg
08:33:35 [ericP]
... some use cases may not have had a unique benifit.
08:33:49 [ericP]
... if you see use caess that need benifits, ping me
08:34:16 [ericP]
RobS: who are we trying to show the benifits too?
08:36:06 [ericP]
RobS: compare each to writing a bunch of custom code
08:36:30 [ericP]
Kendal: just occured to me to move the use cases to a separate part of the doc
08:36:31 [HiroyukiS]
HiroyukiS has joined #DAWG
08:36:40 [ericP]
... everyones's use cases are different
08:36:48 [ericP]
... do we want a standard form?
08:37:08 [ericP]
... then benifits section would go away
08:37:30 [ericP]
RobS: is the use case all problem? part problem and part use case?
08:37:47 [ericP]
Kendal: there ar a few more that need to go in. interested in structural suggestions.
08:38:03 [ericP]
s/Kendal\b/Kendall/g
08:38:53 [ericP]
Jos: currently a there is a link between the use cases and the requirements
08:39:17 [ericP]
is *NOT*
08:39:23 [alberto]
alberto has joined #dawg
08:39:40 [ericP]
Kendall: we discussed this in a teleconf. would like to establish this link.
08:40:09 [kendall]
So I'll work on making those linkages explicit soon, per DanC's whisper to me just now. :>
08:40:26 [ericP]
RobS: are we trying demonstrate to markets or are we trying to establish technical requirements?
08:40:42 [ericP]
DanC: i think we have 3x to many right now.
08:40:47 [Yoshio]
Here's the URI for my additional use cases http://www.w3.org/2004/04/21-fuku/usecases
08:40:54 [ericP]
RobS: we are missing a few right now:
08:40:58 [Yoshio]
They may not be corresponding to what I said (^_^;)
08:41:05 [kendall]
Yoshio: I'll look at yr use cases once I get back home. :>
08:41:05 [ericP]
... returning tuples ala a JOIN in SQL
08:41:07 [Yoshio]
I made them to cover some technologies I think needed and not covered by exsiting ones
08:41:36 [ericP]
DanC: you mean returing more than one result?
08:42:07 [ericP]
RobS: yes. most of the use cases can be met without returning tuples.
08:42:27 [ericP]
DanC: we don't have to do requirements tracing all the way back to your use cases.
08:42:54 [ericP]
... if we get to a disagreement in the design, we may have to come back to use cases.
08:43:47 [ericP]
Kendall: we don't need a use case motivating every requirement but we need to tie the use caess we have to requirements.
08:43:58 [kendall]
AndyS: there's no harm in repeating benefits or tech requirements.
08:44:15 [ericP]
AndyS: it has an inward and outward utility.
08:44:31 [AndyS]
AndyS has joined #dawg
08:44:59 [ericP]
Kendall: we want to go down to 5 or 6?
08:45:16 [ericP]
DanC: how many to folks want?
08:45:28 [ericP]
ericP: how many were in the XQuery spec?
08:45:39 [ericP]
DanC: not a precedent I want to follow.
08:46:20 [ericP]
Yoshio: how can we avoid missing important technical points if we restrict the number of use cases in advance?
08:46:31 [ericP]
DanC: we can always add them later
08:47:02 [ericP]
... We'll publish a document that will show people what we're doing.
08:47:43 [ericP]
... We only get one chance to make a first impression, but some typos will involve public reviewers
08:48:15 [kendall]
Document meant to help facilitate discussion over the design.
08:48:37 [ericP]
... When someone introduces an important use case, we may just agree to change the group's mind or we may need to go back to the use cases.
08:49:23 [ericP]
DanC: requirements are strict. if one is too hard, we must republish the requirements to remove it.
08:49:43 [ericP]
... Use cases can look like they salve the world's problems.
08:50:10 [kendall]
RobS: we could have layered/stratified Use Cases. 5 v. complex ones, 10 less detailed ones.
08:51:11 [ericP]
Danc: that appeals to me 'cause we can answer user X's use cases
08:51:12 [ericP]
JeanF: we can then list related problems.
08:51:31 [kendall]
RobS: clients often ask for use cases outside their area of expertise.
08:52:44 [ericP]
DanC: want to see the buzzwords in the TOC
08:53:50 [ericP]
[discussion of separate use cases and requirements documents]
08:54:22 [ericP]
[consensus seems to be one document?]
08:56:03 [ericP]
DanC: would like to start document with "Because there are no formal standards..." section
08:56:30 [ericP]
Kendall: will re-order
08:57:15 [jannes]
pls change <h2>-><h1> for Use cases section in html
08:57:37 [kendall]
AndyS - thanks
08:57:56 [ericP]
RobS: s/Use cases are used to determine/Use cases are used to illustrate/
08:59:19 [ericP]
DanC: remove DAWG from every place except the status section
08:59:39 [kendall]
jannes: gotcha
09:01:49 [JosD]
wrt definition of "use case" http://www.google.com/search?hl=en&ie=UTF-8&oe=UTF-8&edition=us&q=define%3A%22use+case%22&btnG=Search&meta=
09:02:04 [Yoshio]
Yoshio has joined #dawg
09:02:23 [ericP]
RobS: should we illustrate the starting RDF?
09:02:23 [ericP]
... would like links to external RDF.
09:02:23 [ericP]
EricP: we discussed sprinking graphs in the document.
09:03:07 [HiroyukiS]
HiroyukiS has joined #DAWG
09:03:23 [kendall]
ActionItem: RobS will generate some RDF (in some form) for 5.1.1.
09:03:54 [ericP]
ActionItem: EricP will generate a PNG
09:04:51 [DaveB]
DaveB has joined #dawg
09:05:48 [ericP]
RobS: re 5.1.1, what is John Smith a string? a URI? a popup selection?
09:06:34 [kendall]
dirkx: fwiw, i haven't been able to read mail over ssh because of packet loss (i assume)
09:07:48 [ericP]
Yoshio: is the user supposed to know the email in advance? if not, how can the user choose the correct one without extra information?
09:09:39 [ericP]
RobS: are we saying that the name is "John Smith"? or that the user knows about a person named "John Smith"?
09:10:25 [ericP]
... Steve Bastiseeny (SP?)
09:10:58 [AndyS]
AndyS has joined #dawg
09:11:23 [ericP]
RobS: I would like the use case to say which of the above we are addressing.
09:13:10 [ericP]
Yoshio: so the client just knows the email address? how can he pick the "best"?
09:13:25 [ericP]
EricP: let's address the case where he does know more.
09:16:10 [ericP]
[discuss: s/George wants to send an email message to John Smith./George wants to send an email message to a person named "John Smith".]
09:16:37 [ericP]
[discuss: removing "best match"]
09:17:42 [ericP]
DanC: 5.1.1. is probably in the top 5
09:17:52 [ericP]
(per straw poll)
09:18:28 [alberto]
notices small typo in 5.1.1 - '.... tthe "To:" field ' ---> '....the "To:" field'
09:19:11 [kendall]
alberto: got it
09:19:38 [ericP]
[discuss 5.1.2.]
09:20:09 [ericP]
Andys: I like this one to demonstrate use for data *access*
09:21:39 [ericP]
Janne: suggest this get merged with another to provide concrete query
09:25:02 [ericP]
[waiting for another to merge 5.1.2. with]
09:25:16 [ericP]
[discuss 5.1.3.]
09:27:33 [ericP]
Alberto: wanted to deomonstrate data integration.
09:27:36 [Yoshio]
q?
09:27:49 [Yoshio]
* oh no zakim
09:28:10 [kendall]
i like this use case, and i like alberto's use case, but there needs to be more work done on it before we can do much with it.
09:28:12 [ericP]
RobS: that changes my view of this use case
09:28:19 [Phandaal]
Phandaal has joined #dawg
09:29:42 [ericP]
Janne: are we doing an RPC query or a notification request?
09:31:18 [ericP]
ACTION: Kendall and Alerto to work on 5.1.3.
09:31:55 [ericP]
ACTION: RobS to generate some RDF (in some form) for 5.1.1.
09:32:13 [ericP]
ACTION: EricP to generate a PNG for 5.1.1.
09:32:43 [kendall]
dirkx: any idea about setting MTU on mac 10.3?
09:33:24 [kendall]
thx
09:33:36 [ericP]
Yoshio: what about the inverse case? where she finds an article and wants to find related programs?
09:34:00 [DaveB]
DaveB has joined #dawg
09:34:49 [kendall]
seems to have helped me too
09:34:50 [kendall]
yay
09:35:06 [ericP]
ACTION: Yoshio to write up Kate finds an article and wants to find related programs
09:36:40 [ericP]
[discuss: 5.2.1.]
09:36:48 [ericP]
DanC: very general title
09:37:13 [ericP]
RobS: I don't want to imply Natural Language Processing. probably work pointers to the bots.
09:38:01 [ericP]
Kendall: more concrete title: Sharing Resources with a Friend
09:39:44 [ericP]
DaveB: people use these bots to say "this image depicts this person"
09:39:55 [ericP]
DanC: that would be good to add
09:40:13 [ericP]
RobS: family plugs in camera and it goes on the web
09:40:33 [ericP]
... stuff you say in IRC shows up in RDF and you can perform a robust query
09:40:46 [ericP]
... This photo was taken at this location.
09:41:09 [ericP]
JeanF: puts too much imortance on the generation of the RDF
09:44:16 [ericP]
(sorry, that was Janne)
09:46:39 [ericP]
RobS: "The person I was telling you about yesterday..."
09:46:52 [thompsonbry]
thompsonbry has joined #dawg
09:47:46 [ericP]
Kendall: trim the mentions of irc, link to chump bot.
09:48:14 [ericP]
... "Sharing tulip photos wiht a friend"
09:49:20 [ericP]
Alberto: could be a journalist who sends photos, articles and irc descriptions
09:50:06 [ericP]
s/descriptions/annotations/
09:50:16 [ericP]
ACTION: Kendall to recast this in terms of some tool for annotating photos.
09:50:40 [DaveB]
ref to chump bot, image annotation bot, photostuff, etc. but a little less irc
09:51:04 [alberto]
or some kind of online translation service for articles using IRC + RDF + FOAF/codepiction ideas - just give some thoughts..
09:53:23 [ericP]
[discuss: 5.2.2.]
09:54:19 [ericP]
Kendall: I have users who wish to share in the protocol benifits without relinquishing their query language
09:56:18 [ericP]
SteveH: does require current QLs to change to adopt the new protocol
09:57:27 [ericP]
ACTION: Kendall to add negotiation of query lang to candidate requirements
09:58:09 [ericP]
Yoshio: QL negotiations is related to extensibility and layers of impelementation.
09:58:35 [ericP]
... for some home appliances, it would be difficult to impelement a whole QL.
10:02:41 [thompsonbry]
thompsonbry has joined #dawg
10:10:20 [JosD]
JosD has joined #dawg
10:17:26 [ericP]
ericP has joined #dawg
10:33:45 [kendall]
kendall has joined #dawg
10:57:43 [dirkx]
dirkx has joined #dawg
11:01:36 [DaveB]
DaveB has joined #dawg
11:05:29 [HiroyukiS]
HiroyukiS has joined #DAWG
11:30:00 [DanC_jam]
DanC_jam has joined #dawg
11:30:24 [Zakim]
Zakim has joined #dawg
11:30:29 [DanC_AMS]
Zakim, this will be RDFD
11:30:29 [Zakim]
ok, DanC_AMS; I see SW_DAWG()8:00AM scheduled to start in 30 minutes
11:32:46 [dirkx]
dirkx has joined #dawg
11:34:11 [HiroyukiS]
HiroyukiS has joined #DAWG
11:42:57 [Yoshio]
Yoshio has joined #dawg
11:43:13 [DanC_AMS]
(aha... found Ian's TAG/webarch slides: http://www.w3.org/2004/Talks/0121-ij-xmlgov/all.htm)
11:44:26 [Yoshio]
I put some chocolates on the coffee table. They are Japanese one, sorry, but please feel free to take them.
11:46:09 [AndyS]
AndyS has joined #dawg
11:47:43 [JosD]
JosD has joined #dawg
11:59:57 [alberto]
alberto has joined #dawg
12:00:31 [AndyS]
AndyS has joined #dawg
12:01:15 [DaveB]
editing http://www.w3.org/2001/sw/DataAccess/ftfreq.txt
12:02:00 [JosD]
help
12:02:24 [JosD]
aliad JosD JosD_scribe
12:02:29 [JosD]
alias JosD JosD_scribe
12:02:36 [ericP]
... end of notes from lasst session:
12:02:43 [ericP]
[discussion of tradeoff between compatibility and flexibility when using profiles]
12:02:43 [ericP]
Janne: could merge this one with 5.1.2.
12:02:43 [ericP]
Kendal: noted
12:02:43 [ericP]
AndyS: I would like to deal with the forward looking issue of future forms of query languages that we can't imagine yet.
12:02:46 [ericP]
RobS: would like it to show the other QLs that motivate lang negotiation.
12:02:49 [ericP]
[5.1.2. is top 10 for 1/2, top 5 for 1]
12:02:51 [ericP]
[discuss "benifits"]
12:02:54 [ericP]
AndyS: benifits and list of use cases with tieback to the benifits.
12:03:44 [JosD_scribe]
========== discussion abour requirements
12:04:39 [JosD_scribe]
RobS: don't see anything about tuples
12:05:42 [JosD_scribe]
btw DaveB is editing the list online and projected on the screen (in the meeting room)
12:07:41 [JosD_scribe]
DanC: proposing "Results of a query can have more than one variable in the bindings"
12:10:59 [JosD_scribe]
Jean-Francios: one graph can be obtained by different bindings (some discussion wether that is the case)
12:12:10 [JosD_scribe]
(... a test case woukd be in place here)
12:13:48 [AndyS]
UC 5.7.1 finding input and output documents for test cases
12:15:45 [JosD_scribe]
DirkX: query results can be a binding and on other occasions a graph
12:20:02 [JosD_scribe]
Jean-Francois: keeping track of the *triple* binding
12:21:56 [JosD_scribe]
Janne: agrees that returning graphs is sufficient (although there could be a compctness issue)
12:23:39 [JosD_scribe]
(JosD had similar argument for graphs)
12:24:38 [Yoshio]
http://www.w3.org/2004/04/21-fuku/usecases.htmlq?
12:24:44 [Yoshio]
q?
12:24:48 [Yoshio]
q+
12:25:18 [JosD_scribe]
Steve: remarks about contexts (lost it a bit)
12:26:20 [JosD_scribe]
RobS: graphs are far beyound what users can understand, return to users in the simplest way
12:27:27 [kendall]
the answer is "both" :>
12:28:40 [JosD_scribe]
Andy: 5.7.1 but added "streaming" and "sorting" and then we need bindings
12:30:25 [DaveB]
RobS said: if you know the queyr, very often you can transform the query into the graph having the bindings, the opposite is not necessarily true
12:30:50 [JosD_scribe]
RobS: you can transform bindings to graphs whereas the opposite is not always the case
12:34:56 [JosD_scribe]
ACTION: JanneS to provide an example of graph / binding
12:36:01 [JosD_scribe]
EricP: for negation and disjunction the "graph to bindings" is impossible (if I got that right??)
12:37:22 [ericP]
you can only reconstruct a graph from a set of bindings and the original query if the orginal query is a simple conjunction or has variables separating the different possible paths for a solution
12:38:41 [ericP]
eg, a solution for ?a to the query (?a dc10:title "foo") || (?a dc:title "bar") will not lead to a graph
12:38:47 [JosD_scribe]
Jean-Francios is projecting some slides showing *triple* bindings
12:40:53 [JosD_scribe]
... suggesting that the computation of graph-to-bindings can be NP expensive
12:43:45 [JosD_scribe]
Andy: 5.1.1 motivates bindings (more than 5.7.1)
12:44:28 [AndyS]
Other way round - 5.7.1. is a better motivation
12:50:10 [JosD_scribe]
Albert: both are relevant at the end of the day
12:51:07 [Yoshio]
* I feel like pushing the SUBMIT button :)
12:51:39 [JosD_scribe]
(DirkX also made a similar remark before which I forgot to scribe...)
12:52:12 [JosD_scribe]
RobS: can we put the question as "why do we need graphs?"
12:56:57 [JosD_scribe]
DanC asks for push or break - it is push
12:57:42 [JosD_scribe]
straw poll shows that majority wants bindings, but still no consesus
12:58:21 [JosD_scribe]
start from the top of the list of requirements
12:59:35 [JosD_scribe]
queries with optional triples: yes 9 no 1
13:00:10 [DanC_jam]
DanC_jam has joined #dawg
13:01:24 [thompsonbry2]
thompsonbry2 has joined #dawg
13:03:13 [JosD_scribe]
queries with disjunction: yes 4 no 3 chat more 2 - 3
13:05:50 [JosD_scribe]
Steve: would allow disjunction on a node; on triples it is a nightmare
13:07:35 [JosD_scribe]
EricP: implementation experience and examples, do union of results
13:09:16 [JosD_scribe]
... mechanically merge the results but 5 goes to 25
13:11:19 [JosD_scribe]
DanC: use conjunction on the triples and owl disjunction
13:19:41 [JosD_scribe]
discussing "Query results using 1) graph entailment OR 2) treating the graph as a fixed object"
13:22:39 [JosD_scribe]
Kendall: motivating use case 5.5.1
13:25:10 [JosD_scribe]
proposal to move it to the very bottom
13:27:28 [JosD_scribe]
"queries expressing arbitrary RDF data types" yes 6 no 3
13:29:46 [JosD_scribe]
"datatypes xsd:integer, xsd:float with operations less than, equal, greater than" has sufficient support
13:31:13 [JosD_scribe]
============== time for a break======================
13:48:58 [DanC_AMS]
. http://www.w3.org/2004/Talks/0121-ij-xmlgov/all.htm
13:51:25 [ericP]
DanC speaks on web architecture
13:53:51 [alberto]
alberto has joined #dawg
14:04:57 [ericP]
AndyS: are there issue trends?
14:06:31 [ericP]
[DanC does a flyby of the document with the comments]
14:06:51 [ericP]
DanC: section 2 (identification) is the winner
14:07:06 [Yoshio]
Yoshio has joined #dawg
14:10:58 [ericP]
DanC: I don't have enough to data to discuss publications -- back to use cases
14:13:20 [ericP]
[discuss 5.3.1.]
14:13:42 [ericP]
top 10 for 5 folks
14:15:52 [ericP]
DanC: we may eliminate disjunction from the requirements, but it would still be useful to show folks how to do it.
14:15:58 [ericP]
top 5 for 1
14:17:26 [ericP]
ericP: is this the spokesmodel for disjunction?
14:17:32 [ericP]
DanC, Kendall: yes
14:17:50 [ericP]
[discuss 5.4.1.]
14:18:51 [ericP]
RobS: expressing the RDF that we're querying affects the query.
14:19:27 [ericP]
DanC: does the DB have great-circle arithmetic support?
14:19:50 [ericP]
DanC: one agency is happy if the QL has +-<>
14:20:07 [ericP]
... another agency wants great-circle, square root.
14:21:43 [ericP]
Alberto: this use case is useful to clarify extension syntax to solve queries
14:23:06 [ericP]
DanC: i expect this discussion in the design. Patrick has raised it.
14:24:43 [ericP]
... beware of interoperatbility sacrifices with too many optional features
14:24:58 [kendall]
i'm learning to hate that argument. mainly because it's true. :<
14:25:07 [kendall]
er, valid and its premises are likely true, i shld say
14:26:05 [ericP]
AndyS: i dropped functions for this reason
14:29:39 [Yoshio]
are we discussing on the premises?
14:32:52 [ericP]
top 10 for 9
14:32:57 [ericP]
top 5 for 7
14:33:26 [Zakim]
Zakim has left #dawg
14:34:21 [Yoshio]
* I was amazed by Zakim saying so
14:34:25 [ericP]
[discuss: 5.5.1. Finding unknown human persons]
14:34:44 [ericP]
CONSENSUS: recast as media thing.
14:35:14 [ericP]
top 10 for 10
14:35:25 [ericP]
top 5 for 5
14:37:29 [ericP]
Yoshio: content is good but the intelligence/military aspect may put people off
14:38:23 [ericP]
DanC: it is expected that recasting will address this
14:38:38 [ericP]
[discuss:: 5.6.1. Finding information about motorcycle parts]
14:39:04 [ericP]
RobS: wasn't motivated by the "tell me about" aspect of this
14:39:33 [ericP]
[DaveB: describes his "tell me about" mechanics]
14:40:01 [ericP]
RobS: interactive browsing interests me
14:41:01 [ericP]
[SteveH discusses his implementations of "tell me about"]
14:41:32 [JosD]
JosD has joined #dawg
14:43:35 [ericP]
ACTION: SteveH to get a write-up of the semantic browser use case
14:47:02 [ericP]
top 10 for 8
14:47:10 [alberto]
in relation to Semantic Web broweser UC see "How to make a Semantic Web browser" http://www.ai.mit.edu/people/dquan/www2004-browser.pdf - good/inspiring perhaps
14:47:28 [ericP]
top 5 for 0
14:48:32 [Yoshio]
http://www.w3.org/2004/04/21-fuku/usecases.html
14:48:54 [ericP]
[discussion of supply chain use cases]
14:50:41 [ericP]
ACTION: Kendall to look at Yoshio's use cases.
14:51:18 [ericP]
[discuss: 5.7.1. Finding input and output documents for test cases]
14:51:32 [ericP]
unique feature: triple orienated
14:51:44 [ericP]
top 10 for 9
14:51:58 [ericP]
top 5 for 4
14:52:47 [ericP]
[discuss: 5.7.2. Describing software configurations]
14:53:47 [ericP]
top 10 for 9
14:53:53 [ericP]
top 5 for 2
14:56:14 [ericP]
[discuss: 5.8.1. Avoiding traffic jams]
14:56:37 [ericP]
Kendall: the magic can go away by fixing the last sentence
14:57:30 [kendall]
well, really, i don't think there is magic in it, but -shrug-.
14:57:44 [kendall]
we do web service composition backed by an HTN planner
14:58:12 [ericP]
RobS: could move the three queries from the use case to the solution
14:58:41 [ericP]
... would document that the RDF stores are independent
14:58:48 [ericP]
... would not go beyond that.
14:59:34 [ericP]
Jos: maps are world-wide. you just go for the part relevent to you
14:59:55 [Yoshio]
* +1 to dirk
15:00:14 [ericP]
RobS: There exists public RDF stores. One stores traffic, one stores weather, ...
15:00:50 [kendall]
dirk: yum!!
15:01:08 [JanneS]
JanneS has joined #dawg
15:10:58 [eikeon]
eikeon has joined #dawg
15:19:30 [DaveB]
path case http://lists.w3.org/Archives/Public/public-rdf-dawg/2004AprJun/0130.html
15:19:32 [DaveB]
PatH
15:21:41 [ericP]
ADJOURN
15:21:47 [JanneS]
JanneS has left #dawg
15:23:05 [alberto]
alberto has left #dawg
17:05:35 [thompsonbry]
thompsonbry has joined #dawg