<DanC> logger, chump C:
<logger> C:See [http://chatlogs.planetrdf.com/swig/2006-06-08#T15-30-15|discussion]
<EliasT> DanC, is SPARQL scheduled meeting coming soon?
<DanC> EliasT, you mean SPARQL/SQL/D2R? not clear. If you took the ball, maybe it could happen sooner than if I continue to hold the ball
<EliasT> I just glanced at the email, not swapped in.
<EliasT> DanC, will there be a teleconf?
<DanC> EliasT, well, let's try it. tel:+1.617.761.6200 the conference code is 83261
<DanC> anybody else, wanna join the voice channel? EliasT ? TimBL?
<EliasT> Zakim is not on here.
<EliasT> who can ping richard cyganiak?
<shellac> DanC, No sign of chris or richard (D2R) today
<DanC> ACTION: AndyS to ask Chris B and folks at ESWC to report to sparql-dev on any BOF that should arise
<timbl> ***** Teleconference NOW on bridging RDB and RDF and SPARQL *****
<timbl> We are missing Rob Crowell
<timbl> who is on vacation
<DanC> C: tel:+1.617.761.6200 the conference code is 83261
<bblfish> I can't connect to chatlog
<bblfish> oh well, now I can
<EliasT> shellac, can you join the call?
<shellac> not at those rates :-(
<EliasT> DanC, could he be invited? ;-)
<ericP> i hear that thre's a fr number for zakim
<AndyS> shellac, skype to call in?
<shellac> I believe skype causes problems
<EliasT> DanC, can Zakim call shellac?
<DanC> <Zakim> I am sorry, DanC; I do not know a number for shellac
<DanC> Zakim is authorized to call out to specific stuff
<DanC> skype seems to work about 40% of the time with W3C teleconferences. it's worth a try
TimBL: I discussed this with
BrianG w.r.t. HCLS...
... seems clear that HCLS is interested, but it's not clear that there are development resources there.
... many of these systems seem to have a mapping; should we try to converge/standardize that?
DanC: hmm... sparql-dev... use that as the forum to do this work? or to build an XG charter? ...
<DanC> ... technically, I'm interested in mappings that can be reduced to horn rules, the way InverseFunctionalProperty can be
TimBL: Dan, is there anything to point to re dbview's key stuff?
<EliasT> dbview.py http://www.w3.org/2000/10/swap/dbork/dbview.py
DanC: there's http://www.w3.org/2000/10/swap/db ; it should have the foreign key stuff, if it doesn't already
<timbl> Table of (domain id, personid) represents a many-1 mapping should be 1 arc each in RDF
<timbl> ^ Pattern.
<chimezie> I have to run :(, but I have an interest in writing a D2R mapping engine for rdflib
<chimezie> i'll have to catch up from the minutes
<DanC> mapping... as opposed to rules?
<chimezie> a mapping from TriplePatterns -> SQL via D2R
<DanC> turns out chimezie is affiliated with a W3C member org ...
<timbl> Cleveland clinic
<DanC> chimezie: I'm interested in D2R because it's expressed in RDF...
chimezie: D2R has already an
implementation to bootstrap from.
... interested in doing this for Python RDFLib.
<shellac> if you have any questions about squirrelrdf I can answer them here, but it's pretty simple
<AndyS> Mapping is http://jena.sourceforge.net/SquirrelRDF/#rdbmap
Andy: motivation for Squirrel is
that we wanted the mapping for cheap/free, unlike D2R where you
have to write the mapping
... we have an LDAP mapping in Squirrel too
<AndyS> ("we have" = shellac, not me)
<shellac> ok, I like those odds :-) just found a head set
<EliasT> E:|Live D2R server.
<DanC> discussion of D2R server and ajax interface doing SPARQL queries, and dbview that makes, for example, column names into browseable http URIs
<EliasT> I'm also wondering on the best practices to make primary keys into URIs.
<EliasT> ... maybe use bnodes and IFPs.
<shellac> I ended up using ...;key1=val1;key2=val2..
Andy: Squirrel makes URIs for database row things...
TimBL: yes, good; that's just like dbview. Have you tried tabulating the data?
<AndyS> Example from SquirrelRDF :: <http://example.com/people;id=1>
<shellac> but only because people wanted it. I tried IFPs, and there was interest in composite keys in the past
<EliasT> shellac, CFIPs?
<timbl> It maps a graph to a database
<shellac> CIFPS: http://blogs.sun.com/roller/page/bblfish/20050825
DanC: an interesting way to do mapping is with rules; note Ora's stuff rewrites queries using rules. yeah... wilbur
<DanC> H:|Wilbur Semantic Web Toolkit for CLOS
TimBL: yes, to add RDFS smarts to wilbur's query, he rewrites the query with all the subproperties
EliasT: I'd be interested in
using Squirrel to expose an SPARQL endpoing to Roller Blogging
... hopefully using an Atom-OWL ontology.
<ericP> DanC asked me to note that http://www.w3.org/2003/01/21-RDF-RDB-access/ was originally written to translate Algae2 queries to SQL but since the SPARQL parser uses the same abstract syntax tree, it also rewrites SPARQL to SQL
<DanC> turns out chimezie is affiliated with a W3C member org, so we have 3 member orgs on the phone, which is enough to make an XG
<timbl> Cleveland clinic
TimBL: proposed scope?
... mapping language?
<timbl> XG for a mapping ontology?
Andy: start with a survey?
... each of the existing systems has a certain amount of system-specifc stuff...
Elias: use semantic media wiki to collect survey material
<DanC> discussion of other relevant W3C members... noting Jim M's talk at XTech and such.
<timbl> Question: Is this just sparql, or is mapping to virtual RDF pages also interesting to people?
<DanC> topic names?
<DanC> RdfAndSql ?
<DanC> hmm... SNORQL?
<DanC> danja, can we use that name for this effort?
<DanC> hmm... http://esw.w3.org/topic/SNORQL is not taken
<AndyS> wiki-ize: "NotesToRDF"
<shellac> not SPASQL!
<AndyS> shellac did LDAP
<DanC> discussion of whether this is exactly SQL shaped or also LDAP, notes, IMAP... DanC expressed "let's keep it SQL shaped" but is become sympathetic to an expanded scope.
<DanC> Scribe: Dan Connolly
<shellac> I also have my eye on spotlight, google desktop, etc...
<DanC> i.e. I'll mail some notes to public-sparql-dev
<DanC> Andy: ok, for now, http://esw.w3.org/topic/RdfAndSql ... we can change it later.
<DanC> RESOLVED: to start with http://esw.w3.org/topic/RdfAndSql for collecting survey materials and to use email@example.com
<DanC> C:we're pretty content to use sparql-dev
<DanC> C:all are encouraged to collect relevant work in http://esw.w3.org/topic/RdfAndSql
<DanC> C:Andy is nominated to coordinate for a while.
<DanC> logger, chump C:
<logger> C:See [http://chatlogs.planetrdf.com/swig/2006-06-08#T16-33-50|discussion]
<ericP> i've started a survey list, including a parameter of interest to me: whether it requires some apriori schema knowledge
<DanC> elaborate that flag just a bit, please? I'm confused about the dbview flag in particular, ericp
<ericP> yeah, i'm thinking i got it wrong there
<ericP> whether you need to tell the tool about the database (pks, fks)
<LeeF> Is this topic only appropriate for RDF to native ("legacy") relational models, or is discussion of best practices for modeling the RDF model in an RDBMs (triple tables, materialized views, subject tables, etc.) appropriate as well (and therefore also talk of mapping SPARQL to SQL on top of these graph structure tables)?
<DanC> I'm much less interesetd in triple tables. but you might add stuff and see what sticks, LeeF
<LeeF> Well, to some extent they're two very different topics (even if both could reasonably be called RdfToSql), which is why I asked the question.
<shellac> in some cases there is a close connection, such as property tables
<LeeF> that's true, shellac_
<DanC> LeeF, I'm inclined to judge relevance by looking at running code. If there's running code that does what you're talking about, it's quite possibly worth noting it in the survey and seeing what happens
<DanC> if it's just an idea, I'm much less interested
<AndyS> Make a separate page and link to it?
<DanC> oops; did I just clobber Lee's edits?
<DanC> nope. good.
<AndyS> The using SQL for RDF storage is a big area in itself.
<DanC> I see http://esw.w3.org/topic/TripleStoreScalability is already there.
<DanC> and http://esw.w3.org/topic/LargeTripleStores
<DanC> and http://esw.w3.org/topic/LargeQuadStores (hmm... are these linked to each other?)
<DanC> and http://esw.w3.org/topic/StoringRDF
<DanC> ah... http://esw.w3.org/topic/AdapterArchitecture seems to be the general case of LDAP/IMAP adapter...
<LeeF> thanks, AndyS and DanC. A separate topic (and one of those extant ones might be good) is probably the best way to go.
<DanC> C:what number are we up to?
<DanC> C8: all are encouraged to collect relevant work in [http://esw.w3.org/topic/RdfAndSql|the RdfAndSql esw wiki topic]