07:33:08 danbri has joined #swade 07:43:42 danbri_dna has joined #swade 07:48:28 dajobe-lap has joined #swade 07:48:53 logger_1 has joined #swade 07:48:53 Users on #swade: logger_1 dajobe-lap danbri_dna danbri CaptSolo Wack 07:50:09 hello 07:50:31 * danbri_dna waves 08:00:21 danbri has quit 08:09:57 danbri_dna has quit 08:13:08 JackRusher has joined #swade 08:13:23 AndyS has joined #swade 08:18:11 afs has joined #swade 08:19:05 AndyS has quit 08:19:14 afs is now known as AndyS 08:20:55 eikeon has joined #swade 08:25:57 bnoide introduction for datatypes - 2 08:26:04 infinte containership member props - 3 08:26:08 ^- daveR 08:26:41 jeen - mt is a formal spec of the semantics 08:26:57 assumed section on entailment rules was a reflection of the mt 08:27:16 why is bnode intro etc. in there. may be connect, but not useful 08:27:28 * dajobe-lap asks for volunteer typers 08:27:40 daver - rules are informative not normative 08:28:29 swh: (earlier) we aren't going to implement the 'obscure' things 08:29:13 swh has joined #swade 08:29:21 frankh - datatypes, who's not planning to impl them? 08:30:06 alberto plan to use them for search and a little bit of inferencing, espec for geo stuff 08:30:30 playing with the first level stuff near rdql support for dtypes in queries 08:31:42 arjon - sesame ... support for numeric, booleans, .. 08:31:50 .. support is comparison operators 08:32:08 .. that appear in SeRQL 08:32:40 jeen - datatype entailment in the spec is solved by our parser 08:32:45 such as rdfD2 08:33:02 frankv - litertals wit h xsd, store normalsied value in the databse 08:33:54 db - asks jena about int in jena 08:34:02 DaveR - ints in schema 08:34:08 subclassling? 08:34:10 xml datatypes? 08:34:28 alberto - ideas uscing xml 08:36:32 instance datatypes implementation not always available, but the parsing side of things can be found in xerces 08:36:54 users want datatypes in & out at the java level 08:37:08 don't want to do the datatype subclassing, at least so far - DaveR 08:37:18 jeen has joined #swade 08:37:33 comparison internal to the datatypes 08:37:44 arjon - received requests for integer comaprion ops, dates 08:37:57 daver - just the xsd builtins 08:38:20 ... in xsd you can define user defined but rdf can't use them since they have no uris 08:38:46 nmg has joined #swade 08:39:23 * nmg wondered why noone was on #swad-e 08:39:47 * AndyS did the same yesterday! 08:39:53 arjon 0- domain and range restr? 08:40:29 ... checking & rejecting things 08:41:04 daver - you can have a violation here, the only place in rdfs you can 08:41:14 .. bit of api that says validate()( 08:41:35 swh: we just try an accept it in those cases 08:41:51 .. we do fail a couple 08:41:56 daver - whitespace thing? 08:42:15 .. should reject ' 10.0 ' as an invalid xsd:integer 08:42:47 arjon - we do it 'prpoerly' :) 08:43:59 .. we don't have a parser input switch for this, but do trim on output 08:44:20 jeen - ignore, normalise and validate 3-way switch 08:47:33 http://db.bell-labs.com/galax/ 08:47:34 jeen has quit 08:49:22 RDF+Xquery: WWW2003 BOF: http://lists.w3.org/Archives/Public/www-rdf-rules/2003May/0020.html 08:53:03 discussion of OWL bits 08:53:17 owl tiny - transitive, symmetric, inverseof, sameas, different from, IFP, FP 08:53:24 soton - plan to do the above 08:53:33 advantages - no class defns, no restrictions 08:53:45 discssuion of sameas and diff frmo for classes or instances, some or ther other 08:54:04 * dajobe-lap needs volunteer to scribe 08:54:38 daver - the owl lite fragment of owl full 08:54:43 ... you an have clases as instances 08:54:54 not just parse but infer, sound but not complete 08:55:16 .. two extensions - havehasvalue rules forwads and backwards 08:55:29 haveValue - a restriction this property has that value 08:56:06 horrible interactions when you have inverseOf 08:56:30 2nd extension was for cardinality constraints beyond 1 08:56:47 i..e how many property values an instance has 08:56:55 ... owl lite has 0, 1 08:56:58 ^- frank 08:59:01 gets hard with eqaulity in combination 08:59:37 daver - reqeuest was for validating of datatype properties, needed >1 cardinality restrictions 08:59:58 ... 2 sets of users - logiicans ndpeople using it as a near-schema validation language 09:02:08 owl lite 09:02:18 no big pressures for some/all values from to implement it 09:04:34 jeen has joined #swade 09:11:17 eikeon has left #swade 09:16:25 dirkx has joined #swade 09:19:33 alberto has joined #swade 09:22:56 alberto has left #swade 09:27:55 eikeon has joined #swade 09:28:36 JackRusher has quit 09:28:57 owl discussion, no record 09:31:00 jeen has quit 09:49:16 jeen has joined #swade 09:49:32 * nmg scribes 09:50:02 JackRusher has joined #swade 09:50:05 dajobe: session concentrating on query languages rather than APIs 09:50:40 dajobe: libby found > 12 rdf query languages 09:50:55 dajobe: use cases for query as part of swad-e 09:51:15 zool has joined #swade 09:51:33 alberto: bof at sardinia (iswc) for query, delivered page of use cases (uri someone?) 09:52:00 http://rdfstore.sourceforge.net 09:52:13 arjohn has joined #swade 09:52:29 http://rdfstore.sourceforge.net/2002/06/24/rdf-query/query-use-cases.html 09:52:31 alberto: comparison of query languages as well 09:52:42 (thanks, arjohn) 09:52:47 ^^ query use cases by Andy and Alberto 09:53:44 alberto: agreement on basics - conjunctive queries, constraints, context/provenance 09:55:24 alberto: use acse page gives syntax comparison for queries in different QLs (also, return formats) 09:56:29 http://www.w3.org/2003/03/rdfqr-tests/recording-query-results.html 09:56:32 libby has joined #swade 09:56:42 alberto: continuation of effort through series of irc meetings 09:57:21 http://www.w3.org/2003/03/rdfqr-tests/ - query test data 09:57:30 alberto: meeting at www2003 in budapest, feed into proposals for rdf query wg 09:58:13 alberto: (our) trend for query langs is sql-friendly 09:58:27 (lot of background noise :( ) 09:58:55 alberto has joined #swade 10:00:13 jeen/arjohn: (on SeRQL) main inputs for defn were other rdf query languages - didn't explicitly look at sql 10:00:32 jeen: oql was an influence 10:00:38 for tests see http://swordfish.rdfweb.org/rdfquery/tests/ 10:00:44 SeRQL user manual: http://sesame.aidministrator.nl/publications/SeRQLmanual.html 10:00:45 thanks 10:01:25 The reason for using RDQL is two fold - reuse the exisitng knowledge at the developer and reusre existing ODBC and JDBC infrastructure - and be able to exploit a lot of 'extra's in the SQL world (transactions, views, what not). 10:01:48 dajobe: recursive queries and compositionality (cf. SeRQL CONSTRUCT) 10:01:57 Plus of course - you are overwhelming the developer/customer already with this totally new RDF thing - so going careful on the query is something to be considered. 10:02:11 dirkx: agreed 10:02:46 dajobe: also, path-based query languages 10:02:47 see also http://sourceforge.net/projects/rdftests for RDF (query) tests Web site - still empty :) 10:03:22 libby: eg. damien steer's treehugger, also work by sbp and graham klyne. XPath-like syntax for RDF query 10:03:54 dajobe: (query spec'd as path through a graph) 10:04:07 dirkx is now known as Dirk-Willem 10:04:32 alberto: mapping between rdql and ath queries quite clear 10:05:21 alberto: xpath can fit on rdql 10:05:41 Dirk-Willem has quit 10:05:55 andys: generating webpages from rdf is an important (future) use case 10:06:10 arjonh: some work from CWI on this, augmenting XSLT 10:06:45 that work is part of the Cuypers project 10:06:49 dirkx has joined #swade 10:07:24 dajobe: graph stylesheet proposals as well (GSS) 10:08:04 dajobe: query over XML in literals using XPath, in addition to making queries look 'XPath friendly' 10:08:15 http://homepages.cwi.nl/~media/cuypers/ 10:08:19 ta 10:09:12 andys: different query syntaxes support differnet usecases (in response to comment from zool) 10:10:04 And I guess an implementors lesson is - be careful with the special chars <, ?, # which are hard to escape in URI's or carry certain semantics outside the SemWeb world 10:10:11 andys: path expressions subset of graph matching, syntaxes for each of these uses 10:10:38 dajobe: XML syntax not primary goal, cf. XPath/XQuery 10:11:03 dajobe: again, compositionality (RDF output from queries) 10:11:12 XML syntax for XQuery (not supporrted by Galax) : http://www.w3.org/TR/xqueryx 10:11:24 dajobe: modularity an issue too 10:11:51 swh: if native language is acceptable, can translate other syntaxes into it 10:12:42 libby: what we did for usecases - small subset of functionality in rdf, then map to specific syntaxes 10:13:11 swh: difficult bits - substrings matching, quads... 10:14:13 dirkx: first step in rdql often good enough, later move to more expressive languages 10:14:39 dirkx: 80% of problem with 20% (rdql) 10:15:12 andys: compactness of queries, wrt to namespace declarations 10:16:18 swh: defining standard nss helps here 10:16:30 dirkx: or add level of indirection 10:16:44 swh: stateful protocol to set these things up in advance? 10:17:28 [not serious suggestion] 10:17:33 :-) 10:17:35 :) 10:17:44 dirkx: Have reference to apoplication profiles which 10:17:55 have sets of common prefixes 10:18:43 (discussion of changes to namespaces - dc given as example of one which has) 10:19:11 dajobe: contexts and grouping in queries - is this imprtant for query languages? 10:19:29 andys: shouldn't be a query-only solution 10:19:38 Applications MAY define certain namespaces but MUST override those when they are re-defined in the USING block. 10:19:53 libby: queries over multiple models? 10:21:11 alberto: fourth component in triple (quad) may be against rdf model, but it is useful 10:21:24 swh: just a way of collecting triples 10:22:21 are SQL views outside the original SQL model? :) 10:22:24 Note to self: Also discuss return set format: RDF/triples, a table, defined at query time, negotiable. 10:23:24 I mean, as soon as you move down to real-world data sets and application you need to find/make compromises to make your software working 10:23:52 (discussion about use of contexts in queries, 3store as example, scribe fails to capture what he said himself) 10:24:32 dajobe: need use cases for these to give a basis for discussion about what contexts are in the context of queries 10:25:24 arjohn: needs not be in query itself, but in context' of query 10:26:19 * AndyS agrees with dirk's suggestion 10:29:36 (again, more discussion about nature of contexts in the 3store view - as resources corresponding to the (rdf/xml) files in which triples appear - that the scribe fails to properly record) 10:30:20 dajobe: relation between querying and inferencing - "evaluate this query assuming RDFS semantics" 10:31:29 dajobe,dirkx: a priori (service) description of capabilities of query engines 10:32:06 dirkx: commonality between return formats - table of bindings versus graph 10:32:16 jeen: resultset format has been defined 10:32:41 http://www.w3.org/2003/03/rdfqr-tests/recording-query-results.html 10:32:45 dirkx: future interoperability wrt soap/wsdl? 10:32:52 thanks, andys 10:33:43 dirkx: lessons to be learned from odbc/jdbc 10:34:27 andys, could you list those options for return formats here? 10:34:59 Main ODBC/JDBC nightmare; Its an API - not a protoocl; so your -client- needs to have libraries for each vendor you talk to. And your firewall needs to understand proprietary protocols. 10:35:21 Returns of queries 10:35:21 1/ single subgraph 10:35:21 2/ results 10:35:21 3/ result set (2 in RDF) 10:35:21 4/ template (c.f. SeRQL) 10:35:22 5/ sequence of graphlets 10:35:24 Returns of queries 10:35:26 1/ single subgraph 10:35:28 2/ results 10:35:30 3/ result set (2 in RDF) 10:35:32 4/ template (c.f. SeRQL) 10:35:34 5/ sequence of graphlets 10:35:36 Returns of queries 10:35:38 * AndyS has cut and paster error 10:36:20 dajobe: access protocols 10:36:36 andys: semantic web server written by graham moore - soap interface to rdf store 10:36:52 andys: addresses how to access large rdf stores remotely 10:37:01 2: results: driven by the select (?foo,?bar) and in that order - or 'declared' in a header prefix of the reply.. 10:37:05 andys: query is natural paradigm for accessing 10:37:09 http://www.w3.org/Submission/2003/SUBM-rdf-netapi-20031002/ 10:37:55 andys: rdf netapi submitted to w3c 10:38:25 andys: query language is triple with holes in 10:38:37 andys: maps onto existing toolkits 10:41:13 zool: service description issues wrt query languages 10:41:43 andys: different interfaces for different query languages? 10:45:00 nmg: marta sabou has been working on web service descriptions of things like Sesame 10:45:58 (paper about it on ISWC'03) 10:48:31 (service discovery, brokerage, matchmaking, service choreography and composition) 10:49:57 Paper by Marta c.s.: http://www.cs.vu.nl/~marta/papers/iswc_paper56.pdf 10:51:52 (semantic web services - DAML Services (-> OWL-S), SWSI (Semantic Web Services Initiative?)) 10:53:17 andys: jim hendler's message on rdf query, focussed query wg to reduce barrier to providing access to sw data 10:54:06 andys: straw poll on interest in wg on access protocols? 10:54:45 dirkx: syntax of query 10:55:02 we are definitively interested also in access protocols - but something easy/small to implement i.e. DNS like 10:55:52 andys: reqs for language important, since some features may present implementation issues (< constraints on integers, eg) 10:56:20 arjohn has quit 10:57:46 alberto has left #swade 10:57:48 AndyS has quit 10:57:48 Next trains: 04, 16, 28, 34, 46 and 58 11:16:11 eikeon has quit 11:24:43 JackRusher has quit 11:33:46 jeen has quit 12:01:21 dirkx has left #swade 12:08:11 dirkx has joined #swade 12:08:14 * zool volunteers 12:08:30 dajobe: what kind of sw related web serviecs might we want 12:08:45 and what kind of protcols they might use - looking up to OWL-S etc 12:08:54 mars needs use cases 12:09:07 ontology mapping... 12:09:33 (mapping translation inside an ontomatic sort of thing) 12:10:01 * zool wonders what use it all is 12:10:10 transform the data or translate the query 12:10:21 (arjohn) 12:10:43 jeen describes something that sounds like p2p negotiation between ontologies 12:10:56 "mapping the terms in the query to terms in your own framework" 12:10:57 eikeon has joined #swade 12:11:21 thesaurus services 12:11:39 multilingual concept maps - fuzzy 12:11:59 similarly with classification schemes 12:12:34 DW: query expanders, e.g. oil spill gives you subtypes which add more properties 12:13:03 request-reply interaction... an interactive reasoning process 12:13:08 jeen has joined #swade 12:13:12 (is that either or) 12:13:51 dajobe: a reasoning service 12:14:14 like jenas plugin reasoner api - expose each one as a web service 12:14:21 rules engines, trust, up the stack 12:14:26 or not at all 12:14:57 (you'd need to have trust in all these inference engines like in aggregators) 12:15:29 making up hypothetical web services 12:15:55 DW: a 'trust expander' a common kind of reasoner in industrial banking apps 12:16:05 (like credit refernce expert systems?) 12:16:22 and business rules related stuff 12:17:05 auditing 12:17:38 dajobe - library service models or at least component models 12:17:49 'wandering round a remote conceptual map' service 12:18:14 frank - remote 'concept explosion' services 12:18:25 pathfinders 12:18:41 (but couldnt that be also subsumed in a QL as dajobe is now pointing out) 12:19:07 frank - including rdfs semantics in pathfinding - sounds more like the prolog appserver 12:19:26 steve - akt OKBC/WS work 12:19:45 effetcively dropped now in favour of sending RDQL 12:20:11 thats a trend that's interesting to note, says dajobe 12:20:44 arjohn mentions the sesame client library 12:21:10 visualisation services 12:21:17 dajobe thinks this is out of our terrain 12:22:19 arguments over naviagting graphs from ontologies 12:22:26 is this something users want access to 12:22:50 users are keen on using the browsing software in 3store 12:23:07 * zool keen to steal this browsing software 12:25:14 http://triplestore.aktors.org 12:25:32 oh it's a frame-based (html) browser 12:26:08 shows co-referent sameAs instances at the top, with the union of all their properties 12:26:15 extract provenance 12:28:14 dajobe: "there are some times when you do want to look at the graph" 12:28:49 similar: brownsauce: http://brownsauce.sourceforge.net/ 12:29:14 seealso foafnaut: http://www.foafnaut.org 12:30:01 whwhwhwh has joined #swade 12:30:14 * libby has some RDF 'webservices' 12:30:29 !nearby bristol,0.2 12:31:06 hm 12:31:11 !pic dirk 12:31:19 Dirk-Willem van Gulik http://www.apache.org/images/asf_logo.gif 12:31:21 steve: not throwing around rgaphs but references to them inside of 3store 12:31:23 yay 12:31:44 passing around the uri of a store and uri of a concept 12:31:52 "sending service A to look at service B" 12:31:55 :-) voiled 12:31:58 !nearto brisghton,0.2 12:31:58 no results found, sorry 12:32:02 !nearto brighton,0.2 12:32:02 -0.15 51 Mid Sussex United Kingdom 12:32:02 -0.17 50.83 Brighton United Kingdom 12:32:10 queries with triggers to take another request 12:32:22 interconnected services in semiautomatic chains 12:32:39 pub/sub for a queryable aggregator 12:32:56 (send a query, get new result sets when they drop into the slots) 12:33:29 asynchronous messaging 12:33:34 !wn query 12:33:35 an instance of questioning; "there was a question about my training"; "we made inquiries of all those who were present" 12:33:37 waiting especially for reasoners 12:34:05 another use case, monitoring frequently updating sources 12:35:55 http://www.w3.org/2001/sw/Europe/reports/dissemination_plan_2/ 12:36:07 Nay - how about that list.... :-) 12:36:10 # Internet, Web and Open Source developer communities 12:36:14 darn 12:36:20 Ta 12:36:27 Now all we need is a pen. 12:36:29 # Internet, Web and Open Source developer communities # Academic and Research Community # Content and Tool Producers # Industry and Commerce 12:36:33 one or more of those. 12:36:36 :) 12:36:47 thanks very much and sorry to be a nuisance! 12:38:20 stabnards for WS descriptions 12:38:32 even simple ones to do with use of utf8 etc 12:39:20 dajobe on language agnosticism and impossiblity of giving good advice 12:39:30 pointing people at web services might help 12:39:44 jeen - do we need a common underlying API for the tools 12:39:54 or are the abstractions higher up 12:40:22 overlapping and inter-seeding between jena, sesame etc 12:40:59 jena gives the impression of activity, sesame has more mature schema support 12:41:08 (dajobe) 12:41:30 kaon [sp?] switched to sesame 12:42:17 DW/frank, this seems to be an argument for a common abstraction layer 12:42:24 whether that's at the model interface 12:42:38 or at a http one... 12:43:11 jeen: you want to be able to run everything local and standalone.; this argues against a common interface being http based 12:43:52 DW: "you are already postulating there there is a common language style API... which could also function over a netwrok" : ideally, both 12:44:42 arjohn: SOAP, RMI and HTTP supported; the latter the most lightweight 12:44:54 ('less marshalling and unmarshalling'- dajobe) 12:44:56 we like http 12:45:33 * libby loves http 12:45:46 yay http 12:45:49 !wn sesame 12:45:49 East Indian annual erect herb; source of sesame seed or benniseed and sesame oil 12:45:55 !wn sesame-2 12:45:55 no results found, sorry 12:46:15 Dirks lovs http so much that there is a bit of despair creeping in... 12:46:30 aw 12:47:00 dajobe talking abotu similar cross-lang API 12:47:19 the core spec is basically sitting in this room 12:47:29 Get-if-Modified, Byterange, 30x redirect, Keep-Alive, Pragma: close, E-Tag, Accept header - and all of a sudden http is stateful and things like firewalls, proxies and content delivery netweork start fiddling with your payload en-route. 12:47:38 but too much cross-language consistency isn't a fruitful expectation suggests jan 12:48:31 overengineering in the DOM api a cautionary lesson - daniel / arjohn 12:49:33 essential simplicity of RDF / divergence of different peoples data manipulation extensions 12:49:51 sesame+jena some api convergence in motion 12:50:22 isn't it easy to emulate the one on top of the other, suggests jan cheerfullly 12:50:48 not just API mapping but translation between object types in the respective APIs 12:51:29 "some sort of graph algebra", common modelling of data perhaps 12:51:34 inside out 12:52:17 most of the APIs are all triple-y says dajobe 12:52:38 also a node-centric approach, walking along the arcs, rather than between the conceptual triples 12:53:34 graph pathfinding traversal in jena - sesame ppl would like it to be in a separate package - perhaps both could slowly modularise towards fusion 12:53:57 can you get partial graphs from a node oriented API asks dajobe 12:54:35 after you add the first node you have to connect new ones with an arc to something 12:54:46 3store only does triples, redland does both 12:55:01 "where is that node, it should be in a graph" 12:55:37 steve: a sideeffect of context is to cause the model to have nodes that arent in the graph, momentarily 12:56:26 jena's OWL api using the triple API... 12:56:47 jeen in planning to implement this in sesame rsn 12:57:24 that's Sean Bechhofer's OWL api, not Jena's. 12:57:33 ah okay sorry 12:57:38 np :) 12:58:00 nmg pulls a funny face 12:58:29 I did? 12:58:57 arjohn: generating statements upon statements, or geenrating rdf/xml data and importing it 12:59:10 or both 12:59:40 which of these tactics should we recommend asks dajobe - the triple api seems intuitive but is easy to get lost in 13:01:05 oops 13:01:29 jeen talks about people finding migration to and fro jena pretty straightforward 13:02:18 frank wants to define the shape of our fuzzy cloud of knowledge currently assembled 13:03:02 * zool doesnt write down what dajobe just said 13:03:23 kevin wants portability per store 13:03:41 debate tending towards complete standardisation or complete translatability 13:04:07 also code maturity / bitrot concerns (DW) 13:04:35 "this is a healthy ecosystem" 13:04:55 Healthy ecosystem - or are they still duking it out and should I wait a wee bit longer for things to stabilize. (And be 'safe'). 13:05:11 arjohn: modularisation of code, model, parser, storage, query, suggesting 13:05:42 i suppose we are still duking it out 13:06:00 we might be running out of things to say 13:06:14 Jan grant's framework: http://www.w3.org/2001/sw/Europe/reports/rdf_api_comparison_report/ 13:06:36 comparative overview would be interesting 13:07:06 describing web services in daml-s or owl-s again 13:08:50 arjohn has joined #swade 13:18:23 swh has quit 13:20:13 eikeon_ has joined #swade 13:20:17 dajobe-ibook has joined #swade 13:20:27 libby has quit 13:20:34 zool_ has joined #swade 13:21:40 libby has joined #swade 13:21:54 nmg has quit 13:21:54 dajobe-lap has quit 13:21:54 eikeon has quit 13:22:16