PeterYim: Welcome to the SemanticWiki mini-series Session-6 - Thu 5-Mar-2009
* Mini-series Title: Semantic Wikis: The Wiki Way to the Semantic Web
* Session-6 Topic: The Future of Semantic Wiki: Trends, Challenges and Outlook
* Session Chair: Prof. Dr. RudiStuder (FZI & Institut AIFB, Universitt Karlsruhe) & Dr. MarkGreaves (Vulcan)
* Panelists:
o Chairs of previous sessions in this mini-series to summarize the outcome from their sessions, and to make their short statements on today's topic (5 min. each)
+ Dr. SebastianSchaffert, Mr. HaroldSolbrig, Mr. MaxVoelkel, Mr. MarkusKroetzsch, Mr. MikeDean, Mr. PeterYim, Dr. LiDing & Dr. JieBao
o Speakers from previous sessions of this mini-series to each deliver short statements regarding the future of semantic wikis as they each see it (2 min. each)
+ Mr. ChristophLange, Mr. DanielHansch, Professor DanielSchwabe, Mr. HaroldSolbrig, Mr. JoelNatividad, Professor KeiCheung, Mr. MarkusKroetzsch, Mr. MikeDean, Professor Dr. PeterDolog, Mr. PeterYim, Dr. SebastianSchaffert, Mr. TobiasKuhn & Mr. YaronKoren
PeterYim: See details on the session page at: http://ontolog.cim3.net/cgi-bin/wiki.pl?ConferenceCall_2009_03_05
PeterYim: .
anonymous morphed into Daniel Schwabe
anonymous morphed into Daniel Schwabe
Daniel Schwabe: Hi Peter, just checking. I thought the conference would start at 3 (my time), but I just noticed it got pushed back 30m
anonymous morphed into Rudi Studer
Rudi Studer1 morphed into Markus Kroetzsch
anonymous morphed into Son Doan
Rudi Studer: We are having problems dialing in. The German telecon service says that the conference ID is not valid
Peter Dolog: hi all
Markus Kroetzsch: We are now using the UK line to dial in.
anonymous morphed into Yaron Koren
Markus Kroetzsch: Hi Yaron.
Yaron Koren: What's up, Markus.
Markus Kroetzsch: We are still waiting for Peter to dial in ...
Daniel (ontoprise): Hi everybody!
PeterYim: Sorry guys ... I am slightly delayed ... I will be over in a couple of minutes
PeterYim: slides just posted ... please refresh session page
Please change your name from 'anonymous' using the Settings button
anonymous morphed into EdDodds
Lars Ludwig: Hello there
Yaron Koren: Looking through the presentations, it looks like Markus and Max's is an old one.
anonymous morphed into Jesse Wang
Yaron Koren: From the 3rd session.
Sebastian Schaffert: Hi, I dialled in via the German line just fine
Markus Kroetzsch sees two people with their hands up. You can put them down agfain with the hand button at the bottom right, I think.
Markus Kroetzsch: We will try that too, back in a minnute.
Sebastian Schaffert: yes, I could if I had not loggen in again
Sebastian Schaffert: -n+d
Markus Kroetzsch: ok, we are back via the German line
anonymous1 morphed into Christoph Lange
Christoph Lange: hi, sorry for coming a bit late...
Markus Kroetzsch: Hi, we are just starting
anonymous morphed into Tim FInin
anonymous morphed into Mike Lang
Sebastian Schaffert morphed into SebastianSchaffert
SebastianSchaffert morphed into Sebastian Schaffert
anonymous3 morphed into DanielRedmon
Markus Kroetzsch morphed into Markus Krötzsch
Daniel Schwabe: Bullets for those w/o slides should are at http://ontolog.cim3.net/cgi-bin/wiki.pl?SemanticWiki/Future
anonymous morphed into JieBao
EdDodds: Thanks to the conveners!!!
anonymous1 morphed into John Pacheco
anonymous2 morphed into Bobbin Teegarden
anonymous morphed into Christoph Lange
Sebastian Schaffert: there is also the 4th SemWiki workshop at ESWC2009
Yaron Koren: I submitted a talk for the SemTech conference, but they never responded, which I guess means that they rejected me.
Yaron Koren:
Markus Krötzsch: You should inquire anyway.
Sebastian Schaffert: ero-training is the goal of every software developer, or should be
Yaron Koren: I did, actually, about a month ago - they didn't respond to that either.
Markus Krötzsch thinks they filter "Yaron" in email headers
Yaron Koren: I knew it!
Yaron Koren: ...or "forms".
EdDodds: Is there a knowledge engineer job posting resource, either on Ontolog or else place?
Tim FInin: we'll continue to need knowledge engineers just as we programmers and database specialists
Sebastian Schaffert: but hard to convince companies that they need one, at least that's my experience
Sebastian Schaffert: thanks Rudi
Markus Krötzsch: There will also be another SMW user meeting, maybe Daniel (ontoprise) can say something on that.
Daniel (ontoprise): Peter, would you be so kind and show my updated slide later? (v1.1)
Mark Greaves: Tim: I agree that we will need KEs in many cases; the question is the degree to which semantic wikis can socialize some of the lower-end schema design applications.
Daniel Schwabe: I don't believe there will be a single "user interface" that is universally "good" for all
Markus Krötzsch: Right, see item 2
Daniel Schwabe: therefore, we really need environments that make it easy to create customized interfaces
Lars Ludwig: we need one environment to customize, maybe
Daniel Schwabe: perhaps some communities may reach a consensus on some interface model that suits them
Markus Krötzsch: Wrongs slide!
Markus Krötzsch notes that slide numbers are not equal to topic numbers
Daniel Schwabe: Why access only from other wiki *systems*, and not from any data source?
Markus Krötzsch: Didn't he say this?
Daniel Schwabe: I heard Rudi say "accessing data from other wiki systems"...
Mark Greaves: DanielS: I agree about the multiple UIs; we don't expect a single best interface to a RDBMS, so why should we expect a single best interface to semantic data? Your M-V-C work is quite cool in this regard.
PeterYim: @DanielHansch: I've got your updated slide online
Markus Krötzsch: ok, we can check on the recording, I was more focussed on the slide
Daniel Schwabe: MarkG - ok, but the actual challenge of good UI design remains
Yaron Koren: No, he talked about other data sources before, including desktop-only data.
Daniel Schwabe: ok, my mistake...
Markus Krötzsch: The slides are here: http://ontolog.cim3.net/file/work/SemanticWiki/SWiki-06_Future-of-SemanticWiki_20090305/SemanticWiki-Future--RudiStuder_20090305.pdf
Yaron Koren: That was early on in the talk.
Daniel Schwabe: yes, there is the URL to all the slides earlier in the transcript
Daniel Schwabe: I think this definition is missing collectively produced content. To me this is one of the defining notions of "wiki"
Yaron Koren: Sure.
Sebastian Schaffert: we had a definition of "Wiki Philosophy" in the first session
Sebastian Schaffert: it included "everyone can edit"
EdDodds: Anyone using twitter here? What hash tags do you use for Ontolog Forum related tweets?
Mark Greaves: DanielS: I think there is enough diversity between SMW+, Knoodl, AceWiki, IkeWiki, SWiM, and the other semantic wikis that our community is not well served by drawing bright inclusion lines or debating terminological scope, even around a fundamental property like collaboration. The marketplace is redefining our term anyway. I'd rather see us be inclusive about the term "semantic wiki", leave it hazily defined, and let our various pieces of software speak for themselves.
Sebastian Schaffert:
Sebastian Schaffert: research has to be fun;.)
Markus Krötzsch: +1 to Mark
Yaron Koren: I would think a definition of semantic wikis that doesn't include collaboration is not a definition at all.
Daniel Schwabe: +1 to Yaron - and that's true for Wikis, not just Semantic Wikis...
Mark Greaves: YaronK: I use a semantic wiki in a noncollaborative way for my own personal information management, for example.
Markus Krötzsch: Actually, my hoempage is a semantic wiki, but I am the only editor
Sebastian Schaffert: BTW, first KiWi open source prerelease: http://www.schaffert.eu/2009/02/27/first-kiwi-open-source-release/ (sorry for advertisement, couldn't resist)
Daniel Schwabe: MarkG - I agree with the overall approach to the problem; I also don't believe in very strong categorizations that serve no purpose.
Yaron Koren: Yes, I'm aware of single-user wikis, but the tools are in place for collaboration.
Markus Krötzsch: Sure
Sebastian Schaffert: not necessarily
Daniel Schwabe: The issue is that if you really take away the collaboration infrastructure, the problem becomes much simpler.
Sebastian Schaffert: it always becomes simpler without cocurrency
Daniel Schwabe: exactly
Sebastian Schaffert: but still, wikis are not about collaboration primarily, they are about creating web content quickly
StephenDavies: (what slide are we on now?)
Mark Greaves: "Database" doesn't have a very tight definition, nor does "word processor" or other common classes of software -- they more have a family resemblence and hazy boundaries. No one has an issue with this. So I'd hope this approach is part of our semantic wiki community as well.
Daniel Schwabe: Hmmm, then wysywig HTML editors would be wiki tools!
StephenDavies: (ah, okay)
Sebastian Schaffert: there is tiddlywiki
Sebastian Schaffert: http://www.tiddlywiki.com/
Yaron Koren: Slide 11 - interestingly, it's about "what is a wiki".
Sebastian Schaffert: a kind of wysiwyg editor if you like
Sebastian Schaffert: but a wysiwyg editor does not create a website, it just creates HTML
Lars Ludwig: take a CMS
Daniel Schwabe: Ok, some CMSs or tools do that - create the page, publish right away. One of the really enabling factors in wikis is easy *linking* (not so much formatting, imho)
Tim FInin: FB and youtube don't seem to be wikis to me.
Sebastian Schaffert: yes, linking is crucial
Sebastian Schaffert: and then there is versioning
Yaron Koren: You can't edit other people's contributions in FB, YouTube, etc.; that's the issue.
Sebastian Schaffert: and (you can debate that) everyone can edit
Daniel Schwabe: So that's why some of the social software sites/tools mentioned in slide 11 would not really count as wikis (from the easy linking pov). And, of course, editing other people's contents, sure.
Tim FInin: more like forums, then
Sebastian Schaffert: yes, but there are nowadays many corporate wiki installations where *not* everyone can edit
Sebastian Schaffert: but still they are wikis
Tim FInin: bossWiki
Daniel Schwabe: it's ok if you have some editorial control on who is allowed to publish...
Daniel Schwabe: functionally speaking, it's still collaborative content AND linking
Harold Solbrig: I think the history component is an important aspect as well.
Sebastian Schaffert: I often summarise the wiki characteristics as follows:
Sebastian Schaffert: - On a wiki, anyone can edit
Sebastian Schaffert: - Wikis are easy to use (buzzword!)
Sebastian Schaffert: - Wiki content is linkable
Sebastian Schaffert: - Wikis support versioning
Sebastian Schaffert: - Wikis support all media (that one is an extension of the old idea of web page)
Sebastian Schaffert: http://www.kiwi-project.eu/index.php/kiwi-vision/21-wiki-philosophy
Lars Ludwig: hm, I could think of a virtual wiki integrating personal statements without direct editing
Christoph Lange: I wouldn't call the feature "versioning" -- IIRC it was originally called "easy undo" = it's easier to undo a mistake than to mess up sth. (and versioning is one solution for that)
Sebastian Schaffert: true
Daniel Schwabe: my definition - easy content and linking; collaborative creation. Versioning is really just a way to overcome lack of concurrency control - leave to the users to undo inconsistent updates, simplifies implementation.
Sebastian Schaffert: @Lars: Twitter?
Sebastian Schaffert: Daniel: versioning is much more important
Sebastian Schaffert: it is about taking away fears
Sebastian Schaffert: the fear of breaking things
Lars Ludwig: why not integrating twitter messages into a wiki
Yaron Koren:
Daniel (ontoprise): The "rule knowledge in SMW"-prototype is online: http://ruledemo.ontoprise.com/index.php?title=Main_Page
Daniel Schwabe: ok, but I consider that as part of "collaborative content creation" support
Sebastian Schaffert: if I know that I can undo my changes in a Wiki, I feel much more easy to actually contribute
Sebastian Schaffert: @Lars: yes, of course (KiWi)
Yaron Koren: I was smiling at the "breaking things", BTW.
EdDodds: Isolated components will be available to link with electronic medical records and financial reporting increasingly done in extensible business reporting language (xbrl) as well. Anyone looking at the medical banking implications of this yet?
Harold Solbrig: The versioning isn't just being able to undo - it carries the evolution of how the idea was formed. Interestingly, discussions should probably be linear - time flows down the page, but core pages need history.
Markus Krötzsch: The "Ask The Wiki" demo is still found at http://semanticweb.org/wiki/Special:ATWSpecialSearch (URL not on my slides)
anonymous1 morphed into John McClure
Peter Dolog: just some thoghts based on above discussion. I think we probably need some kind of a metaphor for semantic wiki. Looking at what the other wrote: Sebastian - quickly update content -> I think this is historically wiki; Daniel: collaborative editing of content was probably added after when CSCW peaple entered, knowledge evolution goes towards semantics a bit (versioning vs. evolution? probably a discussion point too)
EdDodds: >>Lars Ludwig: why not integrating twitter messages into a wiki - Indeed, saw this a.m. http://www.kmworld.com/Articles/News/Breaking-News/Wikis-get-social-52891.aspx on a product (eTouch announces SamePage Version 4.1) that "functionally" attempts this, although not actually with twitter or identi.ca/laconi.ca
Daniel Schwabe: @PeterD - No, collaboration was there since the beginning in Ward Cunninghams first wiki
Peter Dolog: OK, I am not excluding it. I think we just probably miss some kind of methaphor we could all ground too.
Markus Krötzsch: A core aspect of Ward's Wiki was simplicity -- hardly any markup.
Harold Solbrig: As was the original HTML
Lars Ludwig: EdDodds: next step: integrate 'semantic' messages
Peter Dolog: ok, so what would be an equivalent simplicity to semantic wiki
Markus Krötzsch: As Ward said himself, Wikipedia and others have moved away from this part a lot, but I think he is okay with this
Peter Dolog: simplicity in editing beyond content - knowledge?
Daniel Schwabe: I translated these into "easy content creation and linking"
Daniel Schwabe: So I really don't care what is the underlying representation - if I have a tool that makes it very easy to create content, that's fine.
Markus Krötzsch: Ward's wiki was not just "easy" (to use), its whole data model was extremely simple. There was hardly any structure in the data it contained. Mostly links.
LiDing: the notion of easy is hard to say
Daniel Schwabe: Hence, with wysiwyg HTML enable wiki content to be HTML, so long as people don't have to edit the source...
LiDing: even editing with wiki require training
Harold Solbrig: Interesting now that I think about it. HTML 1.0 was more semantic than syntax. The XML community argues that it "got off track" with the images and formatting information rather than the semantics of the message itself. XML, on the other hand, made it *too* easy for everyone to say whatever they said their own way. RDF & OWL etc. are attempts to agree on semantics, but WIKI, in a way, is a return to the original HTML principles.
Mark Greaves: LiD: very well said, "the notion of easy is hard to say"
LiDing: furthermore, editing in English is not that easy
Daniel Schwabe: Besides content itself with the simple markup, the real winner was the simple linking mechanism - reference by name
Markus Krötzsch: but maybe maintaining a basic data model with as little strucutre as possible is not actually our goal these days ...
Daniel Schwabe: at that time, this was the difficult part to achieve in a simple way
Harold Solbrig: Agreed - especially the link to a yet to be created page.
Markus Krötzsch: indeed
John McClure: Core & definitional to wikis -- which seems not to have received alot of attention -- are '''namespaces'''. I believe that public understanding of wikis needs to evolve, not be a revolution with too many concepts. Thus, I suggest that the concept of namespaces -- as a form of strong typing -- is the next crucial atom of info to be communicated.
Daniel Schwabe: @Markus - sure, I would not focus so much on a "data model"
Daniel Schwabe: @johnM - interesting point...
EdDodds: @Lars Ludwig: next step: integrate 'semantic' messages >> yes, I wonder if the open ontology registry (Ontolog Project) might be utilized to connect with tagging a la delicious, folksonomies, twit hash tags, etc. to help give context to these semantic messages...
Peter Dolog: @DanielS: so our ultimate goal is then, what is now difficult on the content in the wikis which semantics can make it simple
Markus Krötzsch: @Daniel: I mean the basic structural model of the wiki content. In spite of all freedom that users should have, technically there must be some basic "model".
Daniel Schwabe: that's part of it. Then there is the "consuming the information" part. No good to be easy to create if it is hard to consume! which leads us to customizable interfaces
Markus Krötzsch: s /Daniel/DanielS/
John McClure: I suggest that IF ontologies are calibrated with (inter)wiki namespaces, THEN ontology metadata can be reasoned, leading to interoperability
Daniel Schwabe: @MarkusK - sure
Markus Krötzsch: we see it!
Yaron Koren: SMW mostly uses categories and not namespaces.
Markus Krötzsch: Well, it uses both but for very different things.
Harold Solbrig: We had to splice namespaces in...
Lars Ludwig: document annotation is no solution
Harold Solbrig: e.g. RDF_type or WINE_cabernet
Markus Krötzsch: @PeterYim: my lightning slide is not linked online -- did you get it?
John McClure: Yaron, yes I agree, but that doesnt distress me. The essential point is that SMW needs to develop an ontology that describes NSs.
Yaron Koren: Markus - a minority of people use namespaces for data. Like Harold.
Markus Krötzsch: Oh, I would not encourage this, from a technical viewpoint
Yaron Koren: Talk to Harold. And John.
John McClure: because...?
Yaron Koren: John - I don't see why usage of namespaces is that important to you.
John McClure: Because it is a '''container'''. Because it's already in MW. Because it is not new
Yaron Koren: You could say the same for categories.
Harold Solbrig: Two different notions of namespace. One is the Mediawiki ns (Category, Template, Property, and talk analogs) and the second is ontology namespace (URI)
Markus Krötzsch: Ah, that is indeed different
Jesse Wang: Yaron: I believe one reason people use namespaces is to avoid name conflict: say, two pages: NS1:MembersList vs NS2:MembersList.
Daniel Schwabe: I already hit *2 on mine
John McClure: Harold, agreed. What is the difference my friend?
Markus Krötzsch: @Harold: Of course, namespaces in ontologies often have no semantic significance. They are specific to some serializations, but not part of the ontological content.
Sebastian Schaffert: and they don't say anything about the ontology
Sebastian Schaffert: they are about the URIs and not suitable for grouping ontologies
Markus Krötzsch: (opening a file in an ontology editor and saving it again may sometimes change the namespaces that are used)
Markus Krötzsch: The notion of "URI" does not involve namespaces either.
Markus Krötzsch: They only come in in XML and some other file formats.
Daniel Schwabe: I guess what is meant is some facility to help "distributed vocabulary management"
LiDing: isn't wiki a place forcing people to converge?
Markus Krötzsch: @DanielS: Yes, the problem is that semantic technology standards are pretty poor on this task.
Daniel Schwabe: @MarkusK - sure, I was just trying to give my interpretation of the remark on using NSs
Markus Krötzsch: yes, and I think it is a valid remark, just hard to implement properly using standards
Daniel Schwabe: @LiDing - yes, but converge != merge
John McClure: What is meant is a hook to define context, the 'frame' through which a particular wiki, or namespace within a wiki, is constructed. Where else to define such context *in the existing framework*?
LiDing: @daniel, right we do forced-converge or merged by mapping
John McClure: Frankly I am most concerned with making swikis palatable as possible
- Start with existing concepts, like namespaces (aka context) and pages (aka resources)
Daniel Schwabe: @LiDing - I believe merged by mapping is more acceptable, hence the use of NSs to help that
John McClure: we can all agree semantically what the 'talk' ns is for. Why stop there?
Am suggesting some additional ''foundational namespaces'' in addition to the
fourteen (14) standard ns's.
Tim FInin morphed into Tim Finin
Daniel Schwabe: Shameless plug - partial answer to point 2 (interface between dbpedia and semantic wiki) - Explorator: http://www.tecweb.inf.puc-rio.br/explorator. First step, exploring data; second step, create/add/edit content (coming)
Yaron Koren: @KeiCheung - well, some might say that a semantic Wikipedia might replace DBPedia entirely.
John McClure: e.g., put one's formal ontology in a 'term' ns, controlled by KEs
Leave folksonomies in the 'category' ns -- potential terms adoptable by KEs
Daniel Schwabe: @yaronK - sorry, I really don't believe that. They serve entirely different purposes. I could see Semantic Wikipedia being based on top of dbpedia...
Sebastian Schaffert: @Yaron: which would make sense, why are they separate in the first place ...
Sebastian Schaffert: well, at the moment, DBPedia is based on Wikipedia
Daniel Schwabe: @YaronK (perhaps that's what you meant?)
Sebastian Schaffert: adding the DBPedia-way of querying data to Wikipedia would make sense
Yaron Koren: No, I'm talking about adding SMW (for instance ) to Wikipedia.
Yaron Koren: Wikipedia could then be queried directly.
Sebastian Schaffert: same result, isn't it (or even better)
Sebastian Schaffert: that's what I meant
Daniel Schwabe: @sebastian - one of the advantages of dbpedia is having a sparql endpoint. It enables powerful interfaces such as the Explorator I mentioned above
Markus Krötzsch hears echo
Sebastian Schaffert: yes, but Wikipedia could offer the SPARQL endpoint
Daniel Schwabe: In addition, you could always have the (Semantic) Media Wiki interface, as you said
Sebastian Schaffert: instead of having it separate
Sebastian Schaffert: I disagree with Mike - OWL must prove that it is useful or otherwise we dump it
Sebastian Schaffert: (my challenge ... )
Christoph Lange: one more thought about namespaces: I think they are needed if links should remain easy to author. LinkByName actually requires one flat namespace, wich is not practical for structuring knowledge, but full URIs are hard to author
Daniel Schwabe: +1 to ChristophL
Sebastian Schaffert: @Christoph: link lookup can be done differently
Harold Solbrig: @Markus (wrt namespaces) - indeed, namespaces have no semantic significance, which is why they work out ok as a part of the name itself. The key, however, is disambiguation - especially when you are referencing outside resources that have not coordinated their names. The classic example is the NCI has a class called "Agent", which includes drugs and other delivery mechanisms. NCI_Agent is needed...
Sebastian Schaffert: I'll maybe discuss with you separately how we now do it in KiWi
Markus Krötzsch: I agree with Sebastian, but I don't think OWL is just useful if it is useful in wikis; actually, we would have very advanced swikis indeed if they are able to leverage a technology as powerful as OWL. Maybe we are not there yet.
KeiCheung: Yaron, my chat room page was blocked by other windows so I didn't see your comment. wikipedia started with free text, so it's not a natural fit to semantic web even dbpedia addresses some of the issues. If we start right using semantic mediawiki (instead of mediawiki), we might be able to create a better dbpedia (neurodbpedia in my case).
Yaron Koren: Well, as far as I know DBPedia just uses Wikipedia's infobox data, which could be relatively straightforwardly semantic-ized.
Sebastian Schaffert: ... or create a flop because people are reluctant to use it
Sebastian Schaffert: if it is too complex
Sebastian Schaffert: difficult act of balancing
Markus Krötzsch: @Harold: yes, I agree; I am not sure how well this is supported by current tools, though
Harold Solbrig: @Markus I disagree. OWL, while it may not appear directly, is an important component when defining the intended meaning of the semantic components. With OWL, RDF, ... we have "Category" and "Property". That said, OWL should be under the covers.
Daniel Schwabe: I think we can build special purpose interfaces to create, edit and navigate data for which we KNOW the (meta) schema.
Sebastian Schaffert: people absolutely don't care about semantics
Sebastian Schaffert: they have to be added in a natural way
Sebastian Schaffert: forms are one way
Markus Krötzsch: in addition, people absolutely don't care about wikis
Sebastian Schaffert: tagging might be one
Sebastian Schaffert: exactly
Yaron Koren:
Yaron Koren: So what are we doing here?
KeiCheung: faviki?
Daniel Schwabe: @MarkusK, unless you are saying this "tongue in cheek", people DO care about wikis...
Harold Solbrig: @Markus: We had to roll our own for the time being. Not only do we have the namespace issue, but, one way or another, we have to know that "Wine" and "Wein" map to the same core resource, so we need a notion of identity.
KeiCheung: owl:sameAs?
Sebastian Schaffert: @Yaron: we care about positioning semantic wikis as a tool that people really like to use without noticing that they are using a wiki or semantics
Harold Solbrig: I want to get to the same page in the end.
Sebastian Schaffert: @Harold: this becomes very difficult
Sebastian Schaffert: think of "Snow"
Markus Krötzsch: @Yaron: That is why I think we should step back and consider the goal we have. We gather valuable experience in supporting structured and unstructured content, for trained and untrained users, in groups or alone -- we can define "CMS" or at least contribute significantly to its future definition.
EdDodds: Folks don't care about "semantics" but they do care about "context" -- they just don't realize they can be the same thing
Sebastian Schaffert: in Bavaria, we have about 5-10 notions of snow, in Iceland they have about 15
Daniel Schwabe: Good user interface are crucial to ANY interactive application...
Sebastian Schaffert: and in Saudi Arabia they probably have only 1
Sebastian Schaffert: +1 to Daniel
Sebastian Schaffert: user centred design
Daniel Schwabe: @SebastianS (Ha, in Brazil we have may .5 notion of snow, none real! :-Q)
Markus Krötzsch: @DanielS: The people from the street care about the label "wiki" as they care about "Web 2.0", but the cateogrization as one or the other type of CMS is not essential to them as long as it works
Daniel Schwabe: @MarkusK - ah, ok
Harold Solbrig: @Sebastian: What we've done is created a 3 part identifier (NS_designation(code)). NS is namespace ID, designation is language specific and possibly changeable name and code is immutable. Note that links don't just come from other wiki pages (!). If we don't find NS_designation(code), we look up NS(Code). If found, we build a redirect page.
EdDodds: A simple case is job matching a la hr-xml. HR still refuses to use "context" to match "job description" with "resume" -- job seekers, however, do wish they would
Markus Krötzsch: @Harold: I think I understand what kind of problems you would encounter there. You want to work on syntax, when all SemWeb standards work on semantics (i.e. letting you identify the entities you model, but not the names that you use for them).
Markus Krötzsch wonders how many parallel side chats one can have while still being an attentive listener ...
EdDodds: Another case is the White House Forum on Health Reform Event on now http://www.whitehouse.gov/live2/ where about of the jawing is about differing definitions and cross talk
EdDodds: alot of
John McClure: Special purpose interfaces (@Daniel) could be bound each to a namespace.
People 'get' that pretty easily, and understand that
Sebastian Schaffert: @Harold: but how does it solve the problem of owl:sameAs redirecting to the same page?
John McClure: place:White House is SURELY different than article:White House and talk:White House
Sebastian Schaffert: concepts are slightly different in different cultures
Daniel Schwabe: @JohnM - sure, that's one of the primitive mechanisms we use in HDEWiki (and more general in HyperDE tool)
John McClure: each requiring a wholly different set of tools to make the goal (completing content) faster, better, cheaper
Daniel Schwabe: @JohnM - there is a builtin notion of "context", and a way to customize the interface depending on the context.
Joel Natividad: Following up on my point about word processors - has anybody looked into using the new document formats - Open XML and ODF in particular as a jumping point
Joel Natividad: for capturing semantically annotated data when creating documents
Joel Natividad: ODF has a metadata Technical Comittee
smishra morphed into Sunil Mishra
Sebastian Schaffert: yes, we even have an open bug tracker issue on this
Sebastian Schaffert: still open
Daniel Schwabe: @YaronK, can you post the URL to the website you mentioned?
Yaron Koren: Sure - http://opencongress.org/wiki
anonymous1: How is the UI problem related to the non-specific spirit of the wiki paradigm? For instance, from my experience working within the biomedical domain I have see that tools such as WIKI-Proteins do not facilitate any specific tool for the kind of information they are meant to support. How could having more specific UIs help solving the UI problem?
Bobbin Teegarden: Has anyone tried to move the wiki ui from words and bullets to an interactive graph (with GIS overlay) more like Gelernter's Mirror World?
Sebastian Schaffert: because specific is always better for the user than generic
Markus Krötzsch: anonymous1, you can use "Settings" at the top panel to get a name.
Daniel Schwabe: @anonymous1 - that's precisely the point I was making in my earlier interventions!
Lars Ludwig: name space? -- I call it mind space
Harold Solbrig: @Sebastian: Even with owl:sameAs, we still have to get folks from whatever historical or language specific hyperlink they've got to the SMW page that defines the category or property. The advantage of the identifier approach is we don't have to carry a history of all names that have been used.
Markus Krötzsch: oops, I misinterpreted the text field next to the hand button
Sebastian Schaffert: response:
Sebastian Schaffert: namespaces are not for semantics
Daniel Schwabe: @YaronK - is there some sort of schema underlying the "structured" portion of this website? Is there a way to access the "raw" underlying data?
Sebastian Schaffert: @Harold: that's fine - my concern is only that it is not so easy to use the same concept in multiple languages
Sebastian Schaffert: we had this in IkeWiki
Yaron Koren: @DanielS - not really, no.
Sebastian Schaffert: but didn't to it in KiWi after a long discussion
Sebastian Schaffert: did the conference break? I was kicked out
Sebastian Schaffert: and cannot log in again
Markus Krötzsch: we are still on
anonymous1: I suppose that also has to do with being able to define atomic components of a wiki page, the page is the atomic unit of the wiki.
PeterYim: @Sebastian ... was that for voice or for the chat room?
Harold Solbrig: @Sebastion: agreed that it isn't. There are attempts and claims to be able to do this, however, in Medical "ontologies" such as Gene Ontology and SNOMED-CT. Also, what of Dublin Core and good ol' RDF?
Sebastian Schaffert: @peter: voice, but back again
PeterYim: glad you are able to get back in .. sorry for the mishap
Sebastian Schaffert: @anonymous1: not necessarily, we also discussed heavily in the community how to annotate parts of pages
Sebastian Schaffert: take the Wikipedia page about the computer mouse
Sebastian Schaffert: it would be useful to annotate each section differently, because it is about many historical mice
Joel Natividad: @Tim: Great stuff! Would love to see what your team put together.
Joel Natividad: @Tim: Great stuff! Would love to check out the work that your team did
anonymous1: can u track changes over anything else but pages?
Sebastian Schaffert: @anonymous1: yes, in KiWi we implemented versioning and transactions for metadata
Sebastian Schaffert: not 100% solved, but mostly
Joel Natividad: Actually, I'm currently working on embedding semantic metadata in the SVG XML files that SRF-Ploticus produces.
anonymous1: is there any literature on that?
John McClure: Markus' concern is that an ns organization conflicts with
one's ability to 'categorize' things in multiple ways.
There are several responses to this. First, don't overlook the
impace of REDIRECTS. Second, consider closely whether the structure
of the ontology is ill-fitting wrt the implemented namespaces.
Christoph Lange: @anonymous1: the page can remain the atomic unit, but then we need good refactoring workflows in case a page grows more complex (and subconcepts emerge within a page)
Sebastian Schaffert: on the versioning? we are working on an article
Sebastian Schaffert: should be finished next week
Yaron Koren: Sorry, what was the question? I couldn't hear it.
Markus Krötzsch: @John: My point is: categories already do that, are well known, and have suppor by advnaced UIs even for searching. So why try to use MW namespaces?
Sebastian Schaffert: @John: never forget the user
Sebastian Schaffert: who would find this useful? I would find it awkward
Markus Krötzsch: @John: I agree that it could be conceived, even though it would not play well with the technical use of namespaces now (e.g. namespaces cannot be added from the wiki but only server-side)
Daniel Schwabe: @anonymous1 - there is a whole series of work on model-based interface specification. If you want to get a flavor of what's possible, some is exemplified in our HDEWiki demo: http://ontolog.cim3.net/file/work/SemanticWiki/SWiki-02_Technology-1_20081120/HDEWiki--DanielSchwabe_20081120.html
John McClure: I don't believe there is any ontology for defining a category.
(Unless you wanna raise Topic Maps v Ontology debate)
Sebastian Schaffert: SKOS
Markus Krötzsch: well, I think we are talking of different things here
Joel Natividad: In my ideal world, users will work on their documents using familiar interfaces and then the knowledge is published on Semantic Wiki
Sebastian Schaffert: +1 to Joel
Sebastian Schaffert: intuitiveness means known patterns of use
John McClure: yep, skos.
Further, mw searching does understand ns's and categories.
Certainly I acknowledge that there's a break with categories in some sense.
But this is called 'emerging ke' right
anonymous1: Hi Christoph, how could the re-factoring be done?
Joel Natividad: and other users can further annotate on Wiki, and if it can be done, the annotations are round-tripped down to the document
Daniel Schwabe: +1 to Joel too
Joel Natividad:
Guoqian Jiang: @Markus, We are using RDF output from SMW for processing proposals generated from LexWiki
Peter Dolog: @Joel: And this is exactly also a big challenge for us as researchers
Peter Dolog: i.e. how to make it simpler to make
Jesse Wang: We are doing something at Vulcan.
Peter Dolog: because program we can always
Jesse Wang: Mark is talking on that.
Peter Dolog: but how everybody can do that?
Christoph Lange: @anonymous1: in Wikipedia it is done manually. Whenever a subsection (e.g. history of Italy) grows too large, somebody first puts a warning there (this should be rolled out to an article of its own), then somebody else does that, and replaces the former section by a short summary, and fixes links on other pages pointing to the subtopic
Markus Krötzsch: we are also working on solutions for inter-wiki data exchange and integration -- I would like to collect input on what people need
Christoph Lange: @anonymous1: Now assume semantic structures on pages (e.g. saying this paragraph is a subconcept of type T), then a semantic wiki could assist with that
Tim Finin: gotta go. Thanks for all the fish
John McClure: @Sebastian - Users find it useful that [[place:White House]] involves its own set of tools
Markus Krötzsch: we plan to release software for exchanging data between wikis (instead of copying it from one wiki to the other by duplicating pages); use cases could affect our desing choices
Guoqian Jiang: @Markus, we are really interested in the future plan for RDF/OWL backend of SMW
Markus Krötzsch: well, feel free to write an email
Markus Krötzsch: (I think the session stops rather soon)
Guoqian Jiang: @Markus, yes, talk to you later by email
Sebastian Schaffert: @John: how about the page about the place which is called "White House" involving their own tools instead of placing it on the link?
Markus Krötzsch: ok, just bear with me being slow processing mails; quite some of them these days ...
Guoqian Jiang: OK, I see.
Daniel Schwabe: @MarkusK - why not use LOD as the underlying basis to share content between wikis?
anonymous1: @ Joel: having a semantic structure supporting the generation of documents could make it possible to produce scientific papers, for instance, fully annotated. This could deliver an OLD environment over the paper. I think that when supporting the generation of documents the annotation should happen naturally and without any effort, the document being generated should be immerse in the web by means of relationships over those data types contained in the paper.
Markus Krötzsch thanks Mark and Rudi for chairing this session, and Peter for setting this up
Joel Natividad: plaudits to the all the conveners, Peter in particular!
Sebastian Schaffert: @Daniel: done by KiWi and planned for extension
John McClure: Users find it useful that [[place:*]] is in effect, a geo database
That [[:White House]] is an article ABOUT something (like the [[place:White House]]
that [[place_talk:White House]] is an article ABOUT [[place:White House]], etc etc
Of course, you can see metastatements abound
anonymous1: Thanks everybody
Sebastian Schaffert: yes, thanks!
Guoqian Jiang: Thanks all
Peter Dolog: thanks everybody
Daniel Schwabe: @SebastianS - great, will take a closer look!
Daniel Schwabe: Thanks all!
Christoph Lange thanks the organizers and all participants
Joel Natividad: any news about 2nd User Group Meeting of SMW?
Daniel Schwabe: Bye all, thanks for the interesting exchange!
Markus Krötzsch: bye
Peter Dolog: bye to all
Sebastian Schaffert: @Daniel: essentially, if you access the KiWi system with a client that sends "Accepts: application/rdf" it redirects to http://showcase.kiwi-project.eu/KiWi/linkeddata.seam?http://showcase.kiwi-project.eu/KiWi/content/FrontPage
PeterYim: Great session ... Mark, Rudi and everyone! Thank you all for a most wonderful mini-series!
Sebastian Schaffert: or something similar
Yaron Koren: @Joel - I guess there's no news.
Joel Natividad:
Yaron Koren: The plan was to have one in Germany in May or so...
Sebastian Schaffert: @Daniel: the key is to use URIs that are "local" to the server domain
Sebastian Schaffert: which can be problematic if you want at the same time use ontologies
Yaron Koren: We could have one instead in the U.S., if there's interest.
Joel Natividad: Count me in, perhaps we can time it with SemTech
Yaron Koren: Yeah... although I'm not going to that one.
Yaron Koren: Is anyone from SMW presenting there, do you know?
Sebastian Schaffert: Markus apparently tried to submit a presentation
Yaron Koren: That was me. Unless Markus did too.
Joel Natividad: I also submitted one but the passed on it as well
Yaron Koren: Oh. Maybe it's a trend.
Joel Natividad: like Markus said
Joel Natividad: mail filters
Yaron Koren: So, are you still planning to go?
Joel Natividad: I am. Particularly since Mark said they were planning to host some SMW sessions
Yaron Koren: Hm.
Yaron Koren: Well, I guess that's where the 2nd SMW user meeting will be, then.
Joel Natividad: Great! Maybe we should continue the planning on semediawiki-user mailing list
Yaron Koren: Sure, feel free to send an email. I still don't think I'm going.
Joel Natividad: k. bye all!
PeterYim: bye everyone! a big THANK YOU to all, once again!
anonymous1: bye everyone
Recent Comments