Re: Linking HTML pages and data
On 16 Feb 2010, at 23:13, Pat Hayes wrote: On Feb 16, 2010, at 6:39 AM, Sean Bechhofer wrote: LODders A simple (possibly dumb) question. Is there a standard mechanism for linking an HTML page to the non-information resource that it describes? Um. OK, I have an equally dumb question in response. No dumber than mine I'm sure! :-) What does it (what can it possibly) mean to *link* to a non- information resource? I have been understanding the usage of link to mean that a link is a URI which both refers to the thing being linked to (the linkee) and also provides access to it when used in an HTTP GET. But this latter, of course, exactly what is impossible to do when the linkee is a non-information resource, pretty much by definition. Do you mean, a standard mechanism to *refer to* the resource? Because surely that is done simply by *using* the URI which names it. It requires no other 'mechanism'; indeed, I don't think that there possibly could be a mechanism for reference. You're right -- I don't really mean link, I mean refer to. For example, in the page http://dbpedia.org/page/Mogwai_(band) I see a number of link elements in the header that point me to alternate representations (rdf, json etc). There's nothing in the header that points me to *http://dbpedia.org/resource/Mogwai_ (band)* (as far as I can tell) though. But there is an owl:sameAs which links to http://mpii.de/yago/ resource/Mogwai_(band), which appears to be a use of a URI referring to the non-information resource. Is this an example of the kind of link you are looking for? Not quite. What I want to try and capture is the fact that the primary topic (to use the term suggested by others) of the page is http://dbpedia.org/resource/Mogwai_(band). Just using the URI doesn't seem to achieve this -- I could use lots of URIs in the page, and not all of them may be my intended primary topic. Why do I want to do this? We're publishing some information about things, and hoping to use a linked data friendly approach. So there are URIs for the things which will content negotiate to appropriate representations (RDF, HTML etc). What we were concerned about was when users end up bookmarking (or sending via email) the HTML pages. In that case, how might we *refer* :-) back to the resource that the page is describing. Clearly I can include human-readable text in the page: This page is about X, and we will do that, but I was wondering if there was a mechanism^H^H^H^H^H^H^H^ something that was in common usage (and which might then give me some advantages with existing tooling). Discussion above suggests that thing might be link with an appropriate rel attribute. Cheers, Sean -- Sean Bechhofer School of Computer Science University of Manchester sean.bechho...@manchester.ac.uk http://www.cs.manchester.ac.uk/people/bechhofer
Re: Linking HTML pages and data
On Wed, Feb 17, 2010 at 2:01 AM, Kingsley Idehen kide...@openlinksw.com wrote: I really don't believe we achieve much via: link rel=primarytopic href=http://education.data.gov.uk/id/school/56; / primarytopic isn't an IANA registered type link. Yes, I know. Nor is foaf:primaytopic :) I think there's a good chance of getting wide adoption for rel=primarytopic as a pattern / microformat / whatever. Having that very simple relation would be a massive boost for cross-linking the document web with the data web, important enough to warrant a special case IMHO. If you absolutely need to use foaf then its better to qualify it: link rel=foaf:primarytopic href=http://education.data.gov.uk/id/school/56; / Yes, its a PITA for the average HTML user/developer, but being superficially simpler doesn't make it a valid long term solution. There is a standard in place for custom typed links re. link/. The two are not exclusive. In an RDFa environment, I would suggest using foaf:primaryTopic (note case too - too easy for developers to mis-type) Ian
Re: Terminology when talking about Linked Data
PS. http://www.w3.org/DesignIssues/LinkedData.html On 17 February 2010 12:00, Danny Ayers danny.ay...@gmail.com wrote: For a definition of Linked Data I'd suggest anything that conforms to timbl's Linked Data expectations: 1. Use URIs as names for things 2. Use HTTP URIs so that people can look up those names. 3. When someone looks up a URI, provide useful information, using the standards (RDF, SPARQL) 4. Include links to other URIs. so that they can discover more things. While Tim only lists RDF SPARQL as the standards, pragmatically I reckon there's a bit of leeway here, e.g. a HTML document or Atom feed is likely to contain links and data that can be interpreted as RDF - in fact *any* hyperlink could be seen as an RDF statement (maybe docA dc:relation docB), so depending on the context a looser definition of linked data as linky stuff doesn't seem unreasonable. Cheers, Danny. -- http://danny.ayers.name -- http://danny.ayers.name
Re: Terminology when talking about Linked Data
For a definition of Linked Data I'd suggest anything that conforms to timbl's Linked Data expectations: 1. Use URIs as names for things 2. Use HTTP URIs so that people can look up those names. 3. When someone looks up a URI, provide useful information, using the standards (RDF, SPARQL) 4. Include links to other URIs. so that they can discover more things. While Tim only lists RDF SPARQL as the standards, pragmatically I reckon there's a bit of leeway here, e.g. a HTML document or Atom feed is likely to contain links and data that can be interpreted as RDF - in fact *any* hyperlink could be seen as an RDF statement (maybe docA dc:relation docB), so depending on the context a looser definition of linked data as linky stuff doesn't seem unreasonable. Cheers, Danny. -- http://danny.ayers.name
Re: Terminology when talking about Linked Data
Wow Nathan, that's an interesting set of reactions - we could go off and discuss them, but I will give my 3 cents on the original question. I too have difficulty with customers on the Open word. Open can mean a few things, and some of the posters here seem to interpret it to mean open standards. My interpretation has been that the data is open; as it says at the start of the project page [1]: The Open Data Movement aims at making data freely available to everyone ... The goal of the W3C SWEO Linking Open Data community project is to extend the Web with a data commons by publishing various open data sets as RDF on the Web and by setting RDF links between data items from different data sources. So it is Linking Open Data, not something like Open Linked Data. So personally I have used Linked Data quite a lot, sometimes as Linked Data Technologies. I take it to mean the same thing as Linking Open Data, but where the data is not necessarily open - this is important for a customer that wants to use the (whole) technology stack, but does not want to make their data open. Open can really freak people out I avoid Semantic Web, as that is often received as primarily doing AI. More recently I have also badged as Web of Data; don't know if Michael started it, but you do see it around. Sort of a good capture of the ideas. I also talk about an application using the Unbounded Web of Data, if it actually goes out and fetches RDF on finding links. Finally, if I am pushed to use Semantic Web (ie that is what they come with), I always say I work in Semantic Web Technologies. As someone who works on the software, it can be very useful to append technologies to whatever phrase I use:- otherwise the assumption is that the work is primarily concerned with building ontologies or transforming datasets, rather than infrastructure development. I don't think that either Linked Data (Technologies) or Web of Data addresses your problems that customers think they already have it in Web Services; I usually talk about moving from point to point vocabularies towards widely agreed vocabularies at that stage, and through to unbounded. Best Hugh [1] http://esw.w3.org/topic/SweoIG/TaskForces/CommunityProjects/LinkingOpenData# head-277d7f68544ce1a9e252f5c0080b6402cd983a49 [2] http://www.webofdata.info/ [3] http://webofdata.wordpress.com/ On 17/02/2010 02:18, Nathan nat...@webr3.org wrote: Mike Bergman wrote: Hi Nathan, Though I assume not universally shared: On 2/16/2010 7:32 PM, Nathan wrote: Peter Ansell wrote: Hi Nathan, On 17 February 2010 11:18, Nathannat...@webr3.org wrote: Hi All, Other than the obvious - Linking Open Data = The name of W3C Community Project - I'm wondering which terminology to use where when talking about (what I'll term Linked Data for now). To me, Linked Data represents theuri uri uri triples; the thing at the core of it, which can be used behind the firewall in a silo with nothing open about it. So if I then term Linked Open Data as Linked Data which has been published properly, then what do I use to refer to the tech-stack and principals as a whole? If it is published internally to an organisation, it may still be Linked Data as the URI's may be resolvable internally by all people who have any need to see the information. It may violate privacy laws for example for the information to be publically available. I wouldn't so much refer to it as properly published, as publically published. Linked data is a set of best practices for publishing and deploying instance and class data using the RDF data model. Two of the best practices are to name the data objects using uniform resource identifiers (URIs), and to expose the data for access via the HTTP protocol. Both of these practices enable the Web to become a distributed database, which also means that Web architectures can also be readily employed. It is not an end in itself, a manifesto for open data, or a substitute for the semantic Web. It is a useful and recommended practice (technique), but nothing more [1]. ;) Mike [1] http://structureddynamics.com/linked_data.html would agree; so far all the responses have been different ways of saying what linked data is; which i agree with wholeheartedly; but further down the in-line comments you'll find the specific problem I'm facing. What is the context in which you need to make the distinction? The context is purely in discussion format; when I'm talking about Linked Data - if I first explain it to mean linked data; then talk about it being made public as linked open data (leaving the private/public what to publish bit out of it) then to what do I refer to the overall tech-stack as? everything that comes with it eg: - Linked Data, RDF, SPARQL, REST, Quad-Stores, REST, Ontologies, OWL2, EAV/CR, FOAF+SSL, HTTP, URIs etc A name for the above as a whole. Two people thus far have said semantic web with some extra words;
Re: Terminology when talking about Linked Data
On Wed, Feb 17, 2010 at 12:51 PM, Damian Steer d.st...@bristol.ac.uk wrote: Historical aside: On 17/02/10 11:20, Hugh Glaser wrote: More recently I have also badged as Web of Data; See [1], since 1998 :-) It's been used fairly regularly since then, although I'd highlight [2] as a particularly significant use of the term. Damian [1] http://www.w3.org/DesignIssues/Semantic.html [2] http://www.plasticbag.org/archives/2006/02/my_future_of_web_apps_slides/ Yes, any use of the phrase Web of data that excludes or sidelines work like Tom Coates' here ([2]) would be ... regrettable. There have already been unfortuate run-ins in blog land about whether you can do 'linked data' without using RDF in some LOD-approved manner. There is much much more to 'data' than RDF (or OWL, or triples, or W3C SemWeb). The Web's a big place and we have to be inclusive. RDF was originally standardised as a metadata system, a mechanism for finding stuff ... whether that stuff was photos, videos, HTML pages, excel spreadsheets, SQL databases, 3d models. It can also be used to provide summaries or normalisation of some of the information held in those data objects too. But we shouldn't forget the original use case, nor sideline it. Metadata about non-RDF documents is still linked data imho: all of those forms of Web information are 'linked data' if we use W3C information-linking technology to increase their findability. There's more information out there than fits comfortably in triples or quads; some of the best information is still in people's heads, after all. FOAF was always blurbed as an experimental linked information system; we should have been clearer that some of that info was in triples, some in human-oriented documents, and some ... critically ... was still in people's heads. The richness comes from the interplay between those three forms of information. But I guess that's why I still cling nostalgically to the word 'information' here, rather than just 'data'. BTW an early and important paper in the 'web of data' line, which tried to bring RDF and XML together as components of a larger ('Semantic Web') story is http://www.w3.org/1999/04/WebData ... it doesn't use the phrase explicitly (except in the url path maybe) but it is clear on the need for an inclusive approach. cheers, Dan
Re: Terminology when talking about Linked Data
Nathan wrote: Hi All, Other than the obvious - Linking Open Data = The name of W3C Community Project - I'm wondering which terminology to use where when talking about (what I'll term Linked Data for now). To me, Linked Data represents the uri uri uri triples; the thing at the core of it, which can be used behind the firewall in a silo with nothing open about it. So if I then term Linked Open Data as Linked Data which has been published properly, then what do I use to refer to the tech-stack and principals as a whole? Will leave it there, Many Regards Nathan Nathan, As with all communications its depends on your audience. Describing Linked Data monotonically to all audiences is a fatal flaw. Client-Server Savvy Audience (folks that grok Distributed Data Objects realm as the final frontier re. Client-Server computing): Its HTTP based Data Access by Reference. Reference scope is the Data Item (a record) rather than Data Container (e.g. Table), and reference mechanism (Identifier) is delivered via Generic HTTP URI. Remember, you can't apply CRUD operations (local or across the wire) to things you can't Reference by Name or Access Data Representation by address. Database Audience: Use of generic HTTP URIs as Identifiers in EAV/CR graphs. This audience understands that we are all moving beyond SQL (i.e. Not Only SQL) and RDBMS as apex items within data access value pyramid (one defined by the ability to deliver agility via ad-hoc queries against homogeneous or heterogeneous data sources). Web Developer: RESTful data access where the atom is a Resource inextricably bound to a structured Information Resource that bears its description. The resource structured is EAV based (this community isn't interested in the CR part). TimBL's design issues document was basically a meme for how to make the above happen on the Web via a set of best practices. -- Regards, Kingsley Idehen President CEO OpenLink Software Web: http://www.openlinksw.com Weblog: http://www.openlinksw.com/blog/~kidehen Twitter/Identi.ca: kidehen
Re: Linking HTML pages and data
Ed Summers wrote: On Tue, Feb 16, 2010 at 5:51 PM, Ian Davis li...@iandavis.com wrote: You can see it in use on data.gov.uk: http://education.data.gov.uk/doc/school/56 contains: link rel=primarytopic href=http://education.data.gov.uk/id/school/56; / Wow, thanks Ian. I hadn't noticed this pattern in use at data.gov.uk. It seems like a worthwhile pattern to encourage people to follow, by adding it to the How to Publish Linked Data on the Web [1] ... or elsewhere? //Ed [1] http://www4.wiwiss.fu-berlin.de/bizer/pub/LinkedDataTutorial/ We need a document that covers the following: 1. Linked Data Auto Discovery Patterns 2. How to associate documents with the things they describe. -- Regards, Kingsley Idehen President CEO OpenLink Software Web: http://www.openlinksw.com Weblog: http://www.openlinksw.com/blog/~kidehen Twitter/Identi.ca: kidehen
Re: Linking HTML pages and data
Kingsley, Ed, We need a document that covers the following: 1. Linked Data Auto Discovery Patterns 2. How to associate documents with the things they describe. Agree. I've started a document at [1] now - please dump your ideas, thoughts, requirements, etc. there and I'll take care of getting it in a good shape ;) Cheers, Michael [1] http://esw.w3.org/topic/SweoIG/TaskForces/CommunityProjects/LinkingOpenData/ AutoDiscovery Cheers, Michael -- Dr. Michael Hausenblas LiDRC - Linked Data Research Centre DERI - Digital Enterprise Research Institute NUIG - National University of Ireland, Galway Ireland, Europe Tel. +353 91 495730 http://linkeddata.deri.ie/ http://sw-app.org/about.html From: Kingsley Idehen kide...@openlinksw.com Date: Wed, 17 Feb 2010 07:21:06 -0500 To: Ed Summers e...@pobox.com Cc: Linked Data community public-lod@w3.org Subject: Re: Linking HTML pages and data Resent-From: Linked Data community public-lod@w3.org Resent-Date: Wed, 17 Feb 2010 12:59:01 + Ed Summers wrote: On Tue, Feb 16, 2010 at 5:51 PM, Ian Davis li...@iandavis.com wrote: You can see it in use on data.gov.uk: http://education.data.gov.uk/doc/school/56 contains: link rel=primarytopic href=http://education.data.gov.uk/id/school/56; / Wow, thanks Ian. I hadn't noticed this pattern in use at data.gov.uk. It seems like a worthwhile pattern to encourage people to follow, by adding it to the How to Publish Linked Data on the Web [1] ... or elsewhere? //Ed [1] http://www4.wiwiss.fu-berlin.de/bizer/pub/LinkedDataTutorial/ We need a document that covers the following: 1. Linked Data Auto Discovery Patterns 2. How to associate documents with the things they describe. -- Regards, Kingsley Idehen President CEO OpenLink Software Web: http://www.openlinksw.com Weblog: http://www.openlinksw.com/blog/~kidehen Twitter/Identi.ca: kidehen
Re: Terminology when talking about Linked Data
Dan Brickley wrote: On Wed, Feb 17, 2010 at 12:51 PM, Damian Steer d.st...@bristol.ac.uk wrote: Historical aside: On 17/02/10 11:20, Hugh Glaser wrote: More recently I have also badged as Web of Data; See [1], since 1998 :-) It's been used fairly regularly since then, although I'd highlight [2] as a particularly significant use of the term. Damian [1] http://www.w3.org/DesignIssues/Semantic.html [2] http://www.plasticbag.org/archives/2006/02/my_future_of_web_apps_slides/ Yes, any use of the phrase Web of data that excludes or sidelines work like Tom Coates' here ([2]) would be ... regrettable. There have already been unfortuate run-ins in blog land about whether you can do 'linked data' without using RDF in some LOD-approved manner. There is much much more to 'data' than RDF (or OWL, or triples, or W3C SemWeb). The Web's a big place and we have to be inclusive. RDF was originally standardised as a metadata system, a mechanism for finding stuff ... whether that stuff was photos, videos, HTML pages, excel spreadsheets, SQL databases, 3d models. It can also be used to provide summaries or normalisation of some of the information held in those data objects too. But we shouldn't forget the original use case, nor sideline it. Metadata about non-RDF documents is still linked data imho: all of those forms of Web information are 'linked data' if we use W3C information-linking technology to increase their findability. There's more information out there than fits comfortably in triples or quads; some of the best information is still in people's heads, after all. FOAF was always blurbed as an experimental linked information system; we should have been clearer that some of that info was in triples, some in human-oriented documents, and some ... critically ... was still in people's heads. The richness comes from the interplay between those three forms of information. But I guess that's why I still cling nostalgically to the word 'information' here, rather than just 'data'. BTW an early and important paper in the 'web of data' line, which tried to bring RDF and XML together as components of a larger ('Semantic Web') story is http://www.w3.org/1999/04/WebData ... it doesn't use the phrase explicitly (except in the url path maybe) but it is clear on the need for an inclusive approach. cheers, Dan Dan, I have a history tag on del.icio.us, would be nice if you tagged some of your precious historic links using this tag also :-) History is ultimately always the best teacher. Inclusiveness vs NIH has to be the dominant mindset in the emerging realm of Linked Data; nothing is new under the sun, bar context. If we improve on our historic mapping to related domains (as you do so well), we will be rewarded with a reduction in tutorial and definition burdens en route to the inevitable global epiphany re. Hyperdata. -- Regards, Kingsley Idehen President CEO OpenLink Software Web: http://www.openlinksw.com Weblog: http://www.openlinksw.com/blog/~kidehen Twitter/Identi.ca: kidehen
Re: Linking HTML pages and data
Michael Hausenblas wrote: Kingsley, Ed, We need a document that covers the following: 1. Linked Data Auto Discovery Patterns 2. How to associate documents with the things they describe. Agree. I've started a document at [1] now - please dump your ideas, thoughts, requirements, etc. there and I'll take care of getting it in a good shape ;) Cheers, Michael [1] http://esw.w3.org/topic/SweoIG/TaskForces/CommunityProjects/LinkingOpenData/ AutoDiscovery Cheers, Michael Okay, dropped a quick dump :-) -- Regards, Kingsley Idehen President CEO OpenLink Software Web: http://www.openlinksw.com Weblog: http://www.openlinksw.com/blog/~kidehen Twitter/Identi.ca: kidehen
Re: Terminology when talking about Linked Data
Hugh Glaser wrote: Wow Nathan, that's an interesting set of reactions - we could go off and discuss them, but I will give my 3 cents on the original question. I too have difficulty with customers on the Open word. Open can mean a few things, and some of the posters here seem to interpret it to mean open standards. My interpretation has been that the data is open; as it says at the start of the project page [1]: The Open Data Movement aims at making data freely available to everyone ... The goal of the W3C SWEO Linking Open Data community project is to extend the Web with a data commons by publishing various open data sets as RDF on the Web and by setting RDF links between data items from different data sources. So it is Linking Open Data, not something like Open Linked Data. So personally I have used Linked Data quite a lot, sometimes as Linked Data Technologies. I take it to mean the same thing as Linking Open Data, but where the data is not necessarily open - this is important for a customer that wants to use the (whole) technology stack, but does not want to make their data open. Open can really freak people out I avoid Semantic Web, as that is often received as primarily doing AI. More recently I have also badged as Web of Data; don't know if Michael started it, but you do see it around. Sort of a good capture of the ideas. I also talk about an application using the Unbounded Web of Data, if it actually goes out and fetches RDF on finding links. Finally, if I am pushed to use Semantic Web (ie that is what they come with), I always say I work in Semantic Web Technologies. As someone who works on the software, it can be very useful to append technologies to whatever phrase I use:- otherwise the assumption is that the work is primarily concerned with building ontologies or transforming datasets, rather than infrastructure development. I don't think that either Linked Data (Technologies) or Web of Data addresses your problems that customers think they already have it in Web Services; I usually talk about moving from point to point vocabularies towards widely agreed vocabularies at that stage, and through to unbounded. Best Hugh [1] http://esw.w3.org/topic/SweoIG/TaskForces/CommunityProjects/LinkingOpenData# head-277d7f68544ce1a9e252f5c0080b6402cd983a49 [2] http://www.webofdata.info/ [3] http://webofdata.wordpress.com/ Ahh ty, I can see Web of Data, and Linked Data Technologies both being thrown in to a conversation when discussing Linked Data in broad strokes. Also it had slipped my mind till now but there's always the Giant Global Graph reference too - Web of Data seems to set the tone and paint the ideal mental picture for further communications though (ie makes sense to me)! Regards
Re: Terminology when talking about Linked Data
Dan Brickley wrote: On Wed, Feb 17, 2010 at 12:51 PM, Damian Steer d.st...@bristol.ac.uk wrote: Historical aside: On 17/02/10 11:20, Hugh Glaser wrote: More recently I have also badged as Web of Data; See [1], since 1998 :-) It's been used fairly regularly since then, although I'd highlight [2] as a particularly significant use of the term. Damian [1] http://www.w3.org/DesignIssues/Semantic.html [2] http://www.plasticbag.org/archives/2006/02/my_future_of_web_apps_slides/ Yes, any use of the phrase Web of data that excludes or sidelines work like Tom Coates' here ([2]) would be ... regrettable. There have already been unfortuate run-ins in blog land about whether you can do 'linked data' without using RDF in some LOD-approved manner. There is much much more to 'data' than RDF (or OWL, or triples, or W3C SemWeb). The Web's a big place and we have to be inclusive. RDF was originally standardised as a metadata system, a mechanism for finding stuff ... whether that stuff was photos, videos, HTML pages, excel spreadsheets, SQL databases, 3d models. It can also be used to provide summaries or normalisation of some of the information held in those data objects too. But we shouldn't forget the original use case, nor sideline it. Metadata about non-RDF documents is still linked data imho: all of those forms of Web information are 'linked data' if we use W3C information-linking technology to increase their findability. There's more information out there than fits comfortably in triples or quads; some of the best information is still in people's heads, after all. FOAF was always blurbed as an experimental linked information system; we should have been clearer that some of that info was in triples, some in human-oriented documents, and some ... critically ... was still in people's heads. The richness comes from the interplay between those three forms of information. But I guess that's why I still cling nostalgically to the word 'information' here, rather than just 'data'. BTW an early and important paper in the 'web of data' line, which tried to bring RDF and XML together as components of a larger ('Semantic Web') story is http://www.w3.org/1999/04/WebData ... it doesn't use the phrase explicitly (except in the url path maybe) but it is clear on the need for an inclusive approach. cheers, Dan I'd say you're pretty much living testament to the fact that some of the best information is still in people's heads - thanks for the valuable history links, a great example of an aside (which is being debated over in html land), and to delve OT for a minute - do you have any papers or even books written on the history of the web / semantic web / linked data - I've noted several rather good informative posts like this, from yourself, throughout my travels through the mailing list archives. Many Regards, Nathan
Re: Terminology when talking about Linked Data
On Feb 17, 2010, at 6:37 AM, Dan Brickley wrote: ... . RDF was originally standardised as a metadata system, a mechanism for finding stuff ... whether that stuff was photos, videos, HTML pages, excel spreadsheets, SQL databases, 3d models. ... Really? That was not the impression I got when I first got involved with it. In fact, I asked explicitly for clarification, at the first F2F in Sebastopol: is RDF intended to be metadata for Web 'objects', or is it supposed to be a notation for describing **things in general**? And the resounding chorus from the WG was the latter, most definitely not the former. (Which is also what Guha told me right after the very first RDF speclet was first released.) And that is why I designed the semantics based on a logical model theory rather than a computational annotation system. If RDF was supposed to be primarily a mechanism for finding stuff, then we designed it wrong. Pat IHMC (850)434 8903 or (650)494 3973 40 South Alcaniz St. (850)202 4416 office Pensacola(850)202 4440 fax FL 32502 (850)291 0667 mobile phayesAT-SIGNihmc.us http://www.ihmc.us/users/phayes
Re: Terminology when talking about Linked Data
Nathan wrote: Hugh Glaser wrote: Wow Nathan, that's an interesting set of reactions - we could go off and discuss them, but I will give my 3 cents on the original question. I too have difficulty with customers on the Open word. Open can mean a few things, and some of the posters here seem to interpret it to mean open standards. My interpretation has been that the data is open; as it says at the start of the project page [1]: The Open Data Movement aims at making data freely available to everyone ... The goal of the W3C SWEO Linking Open Data community project is to extend the Web with a data commons by publishing various open data sets as RDF on the Web and by setting RDF links between data items from different data sources. So it is Linking Open Data, not something like Open Linked Data. So personally I have used Linked Data quite a lot, sometimes as Linked Data Technologies. I take it to mean the same thing as Linking Open Data, but where the data is not necessarily open - this is important for a customer that wants to use the (whole) technology stack, but does not want to make their data open. Open can really freak people out I avoid Semantic Web, as that is often received as primarily doing AI. More recently I have also badged as Web of Data; don't know if Michael started it, but you do see it around. Sort of a good capture of the ideas. I also talk about an application using the Unbounded Web of Data, if it actually goes out and fetches RDF on finding links. Finally, if I am pushed to use Semantic Web (ie that is what they come with), I always say I work in Semantic Web Technologies. As someone who works on the software, it can be very useful to append technologies to whatever phrase I use:- otherwise the assumption is that the work is primarily concerned with building ontologies or transforming datasets, rather than infrastructure development. I don't think that either Linked Data (Technologies) or Web of Data addresses your problems that customers think they already have it in Web Services; I usually talk about moving from point to point vocabularies towards widely agreed vocabularies at that stage, and through to unbounded. Best Hugh [1] http://esw.w3.org/topic/SweoIG/TaskForces/CommunityProjects/LinkingOpenData# head-277d7f68544ce1a9e252f5c0080b6402cd983a49 [2] http://www.webofdata.info/ [3] http://webofdata.wordpress.com/ Ahh ty, I can see Web of Data, and Linked Data Technologies both being thrown in to a conversation when discussing Linked Data in broad strokes. Also it had slipped my mind till now but there's always the Giant Global Graph reference too - Web of Data seems to set the tone and paint the ideal mental picture for further communications though (ie makes sense to me)! Regards Here are issue against Web of Data : 1. most people assumed a Web of Data from the onset of the Web 2. the fact that a document may or may not host structured data doesn't invalidate it as a unit of data albeit compound in nature (re. innards). Thus, based on the items above, whether its a Web of Documents or a Web of Data, we don't end up with immediate clarity re. what the new Web interaction dimension is all about. I use Web of Linked Data because its easy for juxtaposition re. Web of Documents or Web of Data since neither convey implicit linkage of the kind delivered by generic HTTP URIs :-) -- Regards, Kingsley Idehen President CEO OpenLink Software Web: http://www.openlinksw.com Weblog: http://www.openlinksw.com/blog/~kidehen Twitter/Identi.ca: kidehen
Re: Terminology when talking about Linked Data
Kingsley Idehen wrote: Nathan wrote: Hugh Glaser wrote: Wow Nathan, that's an interesting set of reactions - we could go off and discuss them, but I will give my 3 cents on the original question. I too have difficulty with customers on the Open word. Open can mean a few things, and some of the posters here seem to interpret it to mean open standards. My interpretation has been that the data is open; as it says at the start of the project page [1]: The Open Data Movement aims at making data freely available to everyone ... The goal of the W3C SWEO Linking Open Data community project is to extend the Web with a data commons by publishing various open data sets as RDF on the Web and by setting RDF links between data items from different data sources. So it is Linking Open Data, not something like Open Linked Data. So personally I have used Linked Data quite a lot, sometimes as Linked Data Technologies. I take it to mean the same thing as Linking Open Data, but where the data is not necessarily open - this is important for a customer that wants to use the (whole) technology stack, but does not want to make their data open. Open can really freak people out I avoid Semantic Web, as that is often received as primarily doing AI. More recently I have also badged as Web of Data; don't know if Michael started it, but you do see it around. Sort of a good capture of the ideas. I also talk about an application using the Unbounded Web of Data, if it actually goes out and fetches RDF on finding links. Finally, if I am pushed to use Semantic Web (ie that is what they come with), I always say I work in Semantic Web Technologies. As someone who works on the software, it can be very useful to append technologies to whatever phrase I use:- otherwise the assumption is that the work is primarily concerned with building ontologies or transforming datasets, rather than infrastructure development. I don't think that either Linked Data (Technologies) or Web of Data addresses your problems that customers think they already have it in Web Services; I usually talk about moving from point to point vocabularies towards widely agreed vocabularies at that stage, and through to unbounded. Best Hugh [1] http://esw.w3.org/topic/SweoIG/TaskForces/CommunityProjects/LinkingOpenData# head-277d7f68544ce1a9e252f5c0080b6402cd983a49 [2] http://www.webofdata.info/ [3] http://webofdata.wordpress.com/ Ahh ty, I can see Web of Data, and Linked Data Technologies both being thrown in to a conversation when discussing Linked Data in broad strokes. Also it had slipped my mind till now but there's always the Giant Global Graph reference too - Web of Data seems to set the tone and paint the ideal mental picture for further communications though (ie makes sense to me)! Regards Here are issue against Web of Data : 1. most people assumed a Web of Data from the onset of the Web 2. the fact that a document may or may not host structured data doesn't invalidate it as a unit of data albeit compound in nature (re. innards). Thus, based on the items above, whether its a Web of Documents or a Web of Data, we don't end up with immediate clarity re. what the new Web interaction dimension is all about. I use Web of Linked Data because its easy for juxtaposition re. Web of Documents or Web of Data since neither convey implicit linkage of the kind delivered by generic HTTP URIs :-) :-) Even better! Creating a Web of Linked Data - Linked Data - The Web of Linked Data - Linked Data Technologies - etc Nice and neat, I like it. Thanks, Nathan
Re: Terminology when talking about Linked Data
Nathan wrote: Dan Brickley wrote: On Wed, Feb 17, 2010 at 12:51 PM, Damian Steer d.st...@bristol.ac.uk wrote: Historical aside: On 17/02/10 11:20, Hugh Glaser wrote: More recently I have also badged as Web of Data; See [1], since 1998 :-) It's been used fairly regularly since then, although I'd highlight [2] as a particularly significant use of the term. Damian [1] http://www.w3.org/DesignIssues/Semantic.html [2] http://www.plasticbag.org/archives/2006/02/my_future_of_web_apps_slides/ Yes, any use of the phrase Web of data that excludes or sidelines work like Tom Coates' here ([2]) would be ... regrettable. There have already been unfortuate run-ins in blog land about whether you can do 'linked data' without using RDF in some LOD-approved manner. There is much much more to 'data' than RDF (or OWL, or triples, or W3C SemWeb). The Web's a big place and we have to be inclusive. RDF was originally standardised as a metadata system, a mechanism for finding stuff ... whether that stuff was photos, videos, HTML pages, excel spreadsheets, SQL databases, 3d models. It can also be used to provide summaries or normalisation of some of the information held in those data objects too. But we shouldn't forget the original use case, nor sideline it. Metadata about non-RDF documents is still linked data imho: all of those forms of Web information are 'linked data' if we use W3C information-linking technology to increase their findability. There's more information out there than fits comfortably in triples or quads; some of the best information is still in people's heads, after all. FOAF was always blurbed as an experimental linked information system; we should have been clearer that some of that info was in triples, some in human-oriented documents, and some ... critically ... was still in people's heads. The richness comes from the interplay between those three forms of information. But I guess that's why I still cling nostalgically to the word 'information' here, rather than just 'data'. BTW an early and important paper in the 'web of data' line, which tried to bring RDF and XML together as components of a larger ('Semantic Web') story is http://www.w3.org/1999/04/WebData ... it doesn't use the phrase explicitly (except in the url path maybe) but it is clear on the need for an inclusive approach. cheers, Dan I'd say you're pretty much living testament to the fact that some of the best information is still in people's heads - thanks for the valuable history links, a great example of an aside (which is being debated over in html land), and to delve OT for a minute - do you have any papers or even books written on the history of the web / semantic web / linked data - I've noted several rather good informative posts like this, from yourself, throughout my travels through the mailing list archives. Many Regards, Nathan Nathan, Del.icio.us is our friend, bookmark em :-) -- Regards, Kingsley Idehen President CEO OpenLink Software Web: http://www.openlinksw.com Weblog: http://www.openlinksw.com/blog/~kidehen Twitter/Identi.ca: kidehen
Re: Terminology when talking about Linked Data
Pat Hayes wrote: On Feb 17, 2010, at 6:37 AM, Dan Brickley wrote: ... . RDF was originally standardised as a metadata system, a mechanism for finding stuff ... whether that stuff was photos, videos, HTML pages, excel spreadsheets, SQL databases, 3d models. ... Really? That was not the impression I got when I first got involved with it. In fact, I asked explicitly for clarification, at the first F2F in Sebastopol: is RDF intended to be metadata for Web 'objects', or is it supposed to be a notation for describing **things in general**? And the resounding chorus from the WG was the latter, most definitely not the former. (Which is also what Guha told me right after the very first RDF speclet was first released.) Yes, and I think you've inadvertently unveiled a subtle distinctin between Linked Data and the earlier Semantic Web Project goals. At the heart of HTTP based Linked Data lies the use of the generic HTTP URIs duality ( Identity/Access ) to actually deliver metadata for Data Objects on an HTTP network. And that is why I designed the semantics based on a logical model theory rather than a computational annotation system. If RDF was supposed to be primarily a mechanism for finding stuff, then we designed it wrong. I think you kinda attested to some of that in your blogic presentation :-) Kingsley Pat IHMC (850)434 8903 or (650)494 3973 40 South Alcaniz St. (850)202 4416 office Pensacola(850)202 4440 fax FL 32502 (850)291 0667 mobile phayesAT-SIGNihmc.us http://www.ihmc.us/users/phayes -- Regards, Kingsley Idehen President CEO OpenLink Software Web: http://www.openlinksw.com Weblog: http://www.openlinksw.com/blog/~kidehen Twitter/Identi.ca: kidehen
Re: Terminology when talking about Linked Data
On Feb 17, 2010, at 11:36 AM, Kingsley Idehen wrote: Pat Hayes wrote: On Feb 17, 2010, at 6:37 AM, Dan Brickley wrote: ... . RDF was originally standardised as a metadata system, a mechanism for finding stuff ... whether that stuff was photos, videos, HTML pages, excel spreadsheets, SQL databases, 3d models. ... Really? That was not the impression I got when I first got involved with it. In fact, I asked explicitly for clarification, at the first F2F in Sebastopol: is RDF intended to be metadata for Web 'objects', or is it supposed to be a notation for describing **things in general**? And the resounding chorus from the WG was the latter, most definitely not the former. (Which is also what Guha told me right after the very first RDF speclet was first released.) Yes, and I think you've inadvertently unveiled a subtle distinctin between Linked Data and the earlier Semantic Web Project goals. At the heart of HTTP based Linked Data lies the use of the generic HTTP URIs duality ( Identity/Access ) to actually deliver metadata for Data Objects on an HTTP network. Fine, I understand. But then you can't link to a non-Web object. But I was chiefly commenting on DanB's historical recollections. And that is why I designed the semantics based on a logical model theory rather than a computational annotation system. If RDF was supposed to be primarily a mechanism for finding stuff, then we designed it wrong. I think you kinda attested to some of that in your blogic presentation :-) Ah no, what I was talking about there were just plain bugs and bad design decisions. But this is a more fundamental split, between RDF as a description format and RDF as a purely metadata tool intended to aid Web searching. Maybe these are compatible, but AFAIK the original, intended, design goal for RDF was the former. Pat Kingsley Pat IHMC (850)434 8903 or (650)494 3973 40 South Alcaniz St. (850)202 4416 office Pensacola(850)202 4440 fax FL 32502 (850)291 0667 mobile phayesAT-SIGNihmc.us http://www.ihmc.us/users/phayes -- Regards, Kingsley Idehen President CEO OpenLink Software Web: http://www.openlinksw.com Weblog: http://www.openlinksw.com/blog/~kidehen Twitter/Identi.ca: kidehen IHMC (850)434 8903 or (650)494 3973 40 South Alcaniz St. (850)202 4416 office Pensacola(850)202 4440 fax FL 32502 (850)291 0667 mobile phayesAT-SIGNihmc.us http://www.ihmc.us/users/phayes
Re: Terminology when talking about Linked Data
Pat Hayes wrote: On Feb 17, 2010, at 11:36 AM, Kingsley Idehen wrote: Pat Hayes wrote: On Feb 17, 2010, at 6:37 AM, Dan Brickley wrote: ... . RDF was originally standardised as a metadata system, a mechanism for finding stuff ... whether that stuff was photos, videos, HTML pages, excel spreadsheets, SQL databases, 3d models. ... Really? That was not the impression I got when I first got involved with it. In fact, I asked explicitly for clarification, at the first F2F in Sebastopol: is RDF intended to be metadata for Web 'objects', or is it supposed to be a notation for describing **things in general**? And the resounding chorus from the WG was the latter, most definitely not the former. (Which is also what Guha told me right after the very first RDF speclet was first released.) Yes, and I think you've inadvertently unveiled a subtle distinctin between Linked Data and the earlier Semantic Web Project goals. At the heart of HTTP based Linked Data lies the use of the generic HTTP URIs duality ( Identity/Access ) to actually deliver metadata for Data Objects on an HTTP network. Fine, I understand. But then you can't link to a non-Web object. But I was chiefly commenting on DanB's historical recollections. And that is why I designed the semantics based on a logical model theory rather than a computational annotation system. If RDF was supposed to be primarily a mechanism for finding stuff, then we designed it wrong. I think you kinda attested to some of that in your blogic presentation :-) Ah no, what I was talking about there were just plain bugs and bad design decisions. But this is a more fundamental split, between RDF as a description format and RDF as a purely metadata tool intended to aid Web searching. But, don't structured resource descriptions aid resource discovery, courtesy of relations? Maybe these are compatible, but AFAIK the original, intended, design goal for RDF was the former. Methinks they are compatible, even if we arrived here via unintended effects, of the positive kind :-) Kingsley Pat Kingsley Pat IHMC (850)434 8903 or (650)494 3973 40 South Alcaniz St. (850)202 4416 office Pensacola(850)202 4440 fax FL 32502 (850)291 0667 mobile phayesAT-SIGNihmc.us http://www.ihmc.us/users/phayes -- Regards, Kingsley Idehen President CEO OpenLink Software Web: http://www.openlinksw.com Weblog: http://www.openlinksw.com/blog/~kidehen Twitter/Identi.ca: kidehen IHMC (850)434 8903 or (650)494 3973 40 South Alcaniz St. (850)202 4416 office Pensacola(850)202 4440 fax FL 32502 (850)291 0667 mobile phayesAT-SIGNihmc.us http://www.ihmc.us/users/phayes -- Regards, Kingsley Idehen President CEO OpenLink Software Web: http://www.openlinksw.com Weblog: http://www.openlinksw.com/blog/~kidehen Twitter/Identi.ca: kidehen
Contd: Terminology when talking about Linked Data
Pat Hayes wrote: On Feb 17, 2010, at 11:36 AM, Kingsley Idehen wrote: Pat Hayes wrote: On Feb 17, 2010, at 6:37 AM, Dan Brickley wrote: ... . RDF was originally standardised as a metadata system, a mechanism for finding stuff ... whether that stuff was photos, videos, HTML pages, excel spreadsheets, SQL databases, 3d models. ... Really? That was not the impression I got when I first got involved with it. In fact, I asked explicitly for clarification, at the first F2F in Sebastopol: is RDF intended to be metadata for Web 'objects', or is it supposed to be a notation for describing **things in general**? And the resounding chorus from the WG was the latter, most definitely not the former. (Which is also what Guha told me right after the very first RDF speclet was first released.) Yes, and I think you've inadvertently unveiled a subtle distinctin between Linked Data and the earlier Semantic Web Project goals. At the heart of HTTP based Linked Data lies the use of the generic HTTP URIs duality ( Identity/Access ) to actually deliver metadata for Data Objects on an HTTP network. Fine, I understand. But then you can't link to a non-Web object. But I was chiefly commenting on DanB's historical recollections. I assume we agree that: you can Reference a non-Web object via a Generic HTTP URI, which -- courtesy of Linked Data pattern -- then gives you a Link to a representation of its structured description, at some Web Address. Basically the duality thing in action etc.. Kingsley [SNIP] Pat Kingsley Pat IHMC (850)434 8903 or (650)494 3973 40 South Alcaniz St. (850)202 4416 office Pensacola(850)202 4440 fax FL 32502 (850)291 0667 mobile phayesAT-SIGNihmc.us http://www.ihmc.us/users/phayes -- Regards, Kingsley Idehen President CEO OpenLink Software Web: http://www.openlinksw.com Weblog: http://www.openlinksw.com/blog/~kidehen Twitter/Identi.ca: kidehen IHMC (850)434 8903 or (650)494 3973 40 South Alcaniz St. (850)202 4416 office Pensacola(850)202 4440 fax FL 32502 (850)291 0667 mobile phayesAT-SIGNihmc.us http://www.ihmc.us/users/phayes -- Regards, Kingsley Idehen President CEO OpenLink Software Web: http://www.openlinksw.com Weblog: http://www.openlinksw.com/blog/~kidehen Twitter/Identi.ca: kidehen
Re: Terminology when talking about Linked Data
On 17 Feb 2010, at 18:14, Pat Hayes pha...@ihmc.us wrote: On Feb 17, 2010, at 6:37 AM, Dan Brickley wrote: ... . RDF was originally standardised as a metadata system, a mechanism for finding stuff ... whether that stuff was photos, videos, HTML pages, excel spreadsheets, SQL databases, 3d models. ... Really? That was not the impression I got when I first got involved with it. In fact, I asked explicitly for clarification, at the first F2F in Sebastopol: is RDF intended to be metadata for Web 'objects', or is it supposed to be a notation for describing **things in general**? And the resounding chorus from the WG was the latter, most definitely not the former. (Which is also what Guha told me right after the very first RDF speclet was first released.) And that is why I designed the semantics based on a logical model theory rather than a computational annotation system. If RDF was supposed to be primarily a mechanism for finding stuff, then we designed it wrong. The original use cases were various flavours of 'metadata'; however that concept melts on closer inspection. We did the right thing by going with a general system; but we did lose touch a bit with some of the original scenarios which motivated W3C to standardise RDF in '97. MCF and RDF were never themselves technologies with a built-in scope of 'describing only data', and that was all fine and good. Whenever you dig into 'metadata' requirements you soon find that the whole world is soon in-scope. The gamble of course with a highly general standard is that it can be used in-principle for *everything* but risks in practice being used for nothing. It took us a while to find that niche... Dan Pat IHMC (850)434 8903 or (650)494 3973 40 South Alcaniz St. (850)202 4416 office Pensacola(850)202 4440 fax FL 32502 (850)291 0667 mobile phayesAT-SIGNihmc.us http://www.ihmc.us/users/phayes
Re: Terminology when talking about Linked Data
Danny Ayers wrote: For a definition of Linked Data I'd suggest anything that conforms to timbl's Linked Data expectations: 1. Use URIs as names for things 2. Use HTTP URIs so that people can look up those names. 3. When someone looks up a URI, provide useful information, using the standards (RDF, SPARQL) 4. Include links to other URIs. so that they can discover more things. While Tim only lists RDF SPARQL as the standards, pragmatically I reckon there's a bit of leeway here, e.g. a HTML document or Atom feed is likely to contain links and data that can be interpreted as RDF - in fact *any* hyperlink could be seen as an RDF statement (maybe docA dc:relation docB), so depending on the context a looser definition of linked data as linky stuff doesn't seem unreasonable. Cheers, Danny. Danny, Yes, and basically watch the Microsoft OData [1] space, it will basically accentuate your point re. other data representations for HTTP based Linked Data using the baseline Entity-Attribute-Value graph model via extensions to the Atom+Feed format. Basically, they are picking up where GData stopped etc.. Links: 1. http://www.odata.org/ -- Regards, Kingsley Idehen President CEO OpenLink Software Web: http://www.openlinksw.com Weblog: http://www.openlinksw.com/blog/~kidehen Twitter/Identi.ca: kidehen