SemanticWiki mini-series Session-5 - Thu 12-Feb-2009    (1PLC)

Conference Call Details    (1T95)

Attendees    (1TAW)

SemanticWiki mini-series Background    (1TBF)

The Semantic Wiki mini-series a 6-month mini-series comprising Talks, Panel Discussions and Online Discourse. The series is co-organized by FZI Karlsruhe, Mayo Clinic, Ontolog, RPI Tetherless World Constellation and Salzburg Research, Austria. This represents a collaborative effort between members from academia, research, software engineering, semantic web and ontology communities. The 6-month mini-series intends to bring together developers, administrators and users of semantic wikis, and provide a platform where they can conveniently share ideas and insights. Through a series of (mainly virtual) talks, panel discussions, online discourse and even face-to-face meetings, participants will survey the state-of-the-art in semantic wiki technology and get exposure to exemplary use cases and applications. Together, they will study trends, challenges and the outlook for semantic wikis, and explore opportunities for collaboration in the very promising technology, approach or philosophy which people has labeled "semantic wiki."    (1TBG)

This series of virtual events will dovetail into the face-to-face workshop: "Social Semantic Web: Where Web 2.0 Meets Web 3.0" at the AAAI Spring Symposium (March 23-25, 2009 at Stanford, California, USA - see: ).    (1TBH)

See: our SemanticWiki mini-series homepage and the developing program for the rest of the series.    (1TBI)

Agenda & Proceedings:    (1TBJ)

Session Abstract: Semantic Wiki Applications & Use Cases (2): horizontal applications    (1SYJ)

Session Abstract: by LiDing & JieBao - [ slides ]    (1SYK)

We will be focusing general purposed applications and use cases that can be applied across many domains. Our presenters at this session will take us through applications such as: Data oriented computing: privacy, protection, distributed data, inference (e.g., T1, T2); Supporting community and collaboration (e.g., T3, T4, T5, L1)þ; Usability: visualization and user interaction (e.g., T6);þ and Methodology: design pattern and quality control (e.g. L2).    (1TCW)

Lightning Talks:    (1T8O)

Q & A and Open Discussion:    (1TBS)

Questions and Discussion captured from the chat session:    (1TBU)

Transcript: (lightly edited for clarity only)    (1TBV)

 GuoqianJiang: @Yaron, for the input format, do you consider the RDF output from SMW itself as one of inputs for your extension?    (1TUZ)
 Michael Dale: and it gets really useful when you use it in templates : )    (1TV0)
 EricRLindahl: This seems like a way for a single-step rewrite. What about rewrites in general? Using other ground terms within 
               the page vs. URI resolution rewrite algorithms.    (1TV1)
 EricRLindahl: (also per Jiang's SPARQL type question)    (1TV2)
 EnricoDaga: data as wiki-page/csv... can be used as alternaive store for other extensions' data, flexible, I think...    (1TV3)
 MikeAxelrod: I think there will be uses for this approach...    (1TV4)
 MikeBennett: What about Genericode - this is a way of maintaining selection lists of things like ISO country codes. 
              Is that something that is supported by these mechanisms?    (1TV5)
 KeiCheung: nice presentation, Yaron. I just installed it for my wikineuron project. will let you know how it goes ...    (1TV6)
 PeterDolog: well, I guess quick integration of data from legacy web applications in enterprise environment could be 
             an application for the external page data    (1TV7)
 YaronKoren: Guoqian - no, I haven't really considered supporting RDF - it's too complicated...    (1TV8)
 PeterDolog: Yaron, I might be interested in more details of your extensions    (1TV9)
 PeterDolog: we need something similar in KIWI, in one use case, so we might learn something from you    (1TVA)
 YaronKoren: Ah, great.    (1TVB)
 YaronKoren: Well, you can look at the source code, for one thing - it's really not that long.    (1TVC)
 PeterDolog: ok    (1TVD)
 YaronKoren: (Assuming you're talking about External Data.)    (1TVE)
 PeterDolog: well, external to the wiki    (1TVF)
 YaronKoren: Right.    (1TVG)
 PeterDolog: but still intranet based in enterprise terminology    (1TVH)
 MarcFeickert: Phillip, can you maintain your own list of synonyms with this extension?    (1TVI)
 GuoqianJiang: Yaron, RDF output has more explicit semantics. in some way, you may not need interpret them by yourself.    (1TVJ)
 YaronKoren: Well, if nothing else, I haven't seen a use case for supporting RDF yet.    (1TVK)
 Michael Dale:  YaronKoren: probably would be good to support Xpath via some xml lib integration    (1TVL)
 Michael Dale: in case you need fine grain access to xml attributes and what not    (1TVM)
 MarcFeickert: good call    (1TVN)
 Michael Dale: for example the people.xml from gov track    (1TVO)
 YaronKoren: Maybe... it would only really be necessary if the XML had two different tags or attributes of the same name.    (1TVP)
 MarcFeickert: if you are not in charge of writing the XML, it could    (1TVQ)
 Michael Dale: it could also be a shorter expression / less logic to get at a given piece of data.    (1TVR)
 YaronKoren: Sure. I'm just afraid XPath support would add a lot of complexity.    (1TVS)
 MarcFeickert: for you or for the user?    (1TVT)
 YaronKoren: Marc - definitely for the programmer, probably for the user/admin as well.    (1TVU)
 MarcFeickert: complexity for advanced users is sometimes okay, but I understand the reluctance if it adds too much effort on your end as well    (1TVV)
 GuoqianJiang: Philipp, does autocompletion in searchbox support the query expansion e.g. synonym?    (1TVW)
 PhilippZaltenbach: the autocomplete shows only tile of wiki pages. wiki pages are separated in regular article pages (instances), 
                    category pages, property pages, template pages    (1TVX)
 MarcFeickert: and how can we expand the stemming dictionaries for multiple langauges?    (1TVY)
 RaviSharma: Philipp, How do you connect vocabularies or define related terms, are these a reasoning or inference engine sitting 
             behind the Tags or serach terms?    (1TVZ)
 PhilippZaltenbach: no we are not using external vocabularies at the moment for the query expansion. the only external vocabulary 
                    (for getting synonyms) is wordnet. but that is done by the MWsearch extension.    (1TW0)
 MarcFeickert: so by expanding the wordnet project, it would trickle down to expand this?    (1TW1)
 MarcFeickert: or would there be a way to import or tranform dictionaries?    (1TW2)
 MarcFeickert: actually, I need to talk to you in depth at another time. I'll email you to set something up    (1TW3)
 GuoqianJiang: query expansion for an external vocabulary would be very useful    (1TW4)
 MarcFeickert: agreed    (1TW5)
 MarcFeickert: we intend to cover 20-30 languages    (1TW6)
 PhilippZaltenbach: @marc: you can of course use different languages for annotating synoyms in your wiki. at the moment we
                    simply do not care what a users specifies in the wiki as synonym. so you are free to use different languages.    (1TW7)
 PhilippZaltenbach: at mark, sure feel free to mail me    (1TW8)
 RaviSharma:  PeterDolog, How is a single (?) Triplestoreservice related or delivered or choreographed with SOA Services in 
              the upper architecture layer such as COntentItemService?    (1TW9)
 PhilippZaltenbach: the connecting of external vocabularies (even in different languages) is an issue we have on our radar, 
                    but as i said, we have not investigated further on how to realize that    (1TWA)
 GuoqianJiang: how do you represent the internal vocabularies in Halo?    (1TWB)
 GuoqianJiang: for external vocabulary, we mapped them to wiki pages    (1TWC)
 PhilippZaltenbach: internally we use the special relations such as "also known as" or "broader term". these pre-defined 
                    properties are actually mapped to SKOS properties.    (1TWD)
 GuoqianJiang: I mean we may use the same mechanism for external voca as long as we observe the rule    (1TWE)
 MarcFeickert: is it possible to use those properties to define synonyms for a single page? Like asking the contributor 
               to input them into a semantic form?    (1TWF)
 PhilippZaltenbach: we didn't want to force the not-technical user to use properties such as "skos:altLabel" or "skosrefLabel". 
                    but when you export your ontology these skos properties will be serialized too    (1TWG)
 MarcFeickert: And here I think also of text in the page that may need its own synonym separate of the page title    (1TWH)
 MarcFeickert: You've hit on my professional area, Philipp    (1TWI)
 GuoqianJiang: Is this query expansion a separate extension?    (1TWJ)
 PeterDolog: ok, give me a second to recollect the thought and read your questions    (1TWK)
 PhilippZaltenbach: the synonyms actually relate to the source article where you put in your semantic annotations such as "[[also known as:blabla]]"    (1TWL)
 PhilippZaltenbach: marc, i have to leave soon, but i am looking forward to hear from you via email!    (1TWM)
 MarcFeickert: I have asked Daniel Hansch for your contact info.    (1TWN)
 PhilippZaltenbach: the query expansion is just a "feature" of the enhanced retrieval extension which i highlighted    (1TWO)
 GuoqianJiang: Philipp, you mean enhanced retrieval extension could be separated from Halo?    (1TWP)
 MarcFeickert: aber alles ist ausgezeichnet!    (1TWQ)
 PhilippZaltenbach: yes, it will be available as a separate extension which you can plug in into an SMW (without having to use HALO extension)    (1TWR)
 GuoqianJiang: great    (1TWS)
 PhilippZaltenbach: ok, i have to leave guys, cu!    (1TWT)
 MarcFeickert: Tschuss!    (1TWU)
 GuoqianJiang: thanks, Philipp    (1TWV)
 PeterDolog:  RaviSharma: any store has to be connected to the entitymanager, so there is always a bit of work to connect it    (1TWW)
 PhilippZaltenbach:    (1TWX)
 PeterDolog: content item is an entity in the data model    (1TWY)
 PeterDolog: accessible through hibernate like interface    (1TWZ)
 PeterDolog: that you can actually query it directly    (1TX0)
 YaronKoren: Peter: are these tags editable by everyone, or are they user-specific?    (1TX1)
 PeterDolog: Yaron, which things?    (1TX2)
 YaronKoren: The tags you were talking about.    (1TX3)
 PeterDolog: Whether any person can add tags? Yes    (1TX4)
 PeterDolog: Any user can tag    (1TX5)
 YaronKoren: No, the question is, if one person adds a tag, can another person edit/remove it.    (1TX6)
 PeterDolog: it depends how you set the rights control    (1TX7)
 PeterDolog: in principle yes, if you want    (1TX8)
 PeterDolog: but you can restrict it    (1TX9)
 YaronKoren: I see, okay.    (1TXA)
 PeterDolog: but if you look at the screenshot, I did not have any "-" symbol for me    (1TXB)
 PeterDolog: so it was restricted    (1TXC)
 PeterDolog: it was simple to make recommendations because the core datamodel is basically a triple (Content Item x Tags x Users)    (1TXD)
 PeterDolog: similarly, access control is then simple    (1TXE)
 PeterDolog: and then you can put a type system over it through semantic web annotations    (1TXF)
 GuoqianJiang: good talk, Mike    (1TXG)
 MikeAxelrod: thx    (1TXH)
 GuoqianJiang: we did meet many scenarios you described in your zoo    (1TXI)
 YaronKoren: Peter - is this (tagging) the semantic content of KIWI, or is it in addition to the semantic content?    (1TXJ)
 PeterDolog: it is a bit tricky    (1TXK)
 PeterDolog: actually tag is a special kind of content item    (1TXL)
 PeterDolog: i.e. kind of a resource    (1TXM)
 PeterDolog: which link to another "media/text" based content item    (1TXN)
 PeterDolog: and it is cartesian    (1TXO)
 PeterDolog: i.e you can see the link bidirectionally    (1TXP)
 PeterDolog: I do not know if it answered your question    (1TXQ)
 YaronKoren: Okay, I think I understand...    (1TXR)
 YaronKoren: Basially, unlike SMW, KIWI doesn't store semantic data on the page - it's all separate "resources".    (1TXS)
 PeterDolog: yes    (1TXT)
 PeterDolog: also concepts in an ontology are currently separate content items    (1TXU)
 PeterDolog: and you can use them as labels for tags    (1TXV)
 YaronKoren: Hm, okay.    (1TXW)
 PeterDolog: also user is a content item    (1TXX)
 PeterDolog: so content item is a base entity    (1TXY)
 PeterDolog: and then you type it    (1TXZ)
 PeterDolog: and link instances around    (1TY0)
 YaronKoren: Okay.    (1TY1)
 PeterDolog: actually to clarify    (1TY2)
 PeterDolog: we distinguish between rdf/owl annotations and tags    (1TY3)
 PeterDolog: we allow both    (1TY4)
 PeterDolog: and we can link them    (1TY5)
 MarcFeickert: I'm out of breath...    (1TY6)
 LiDing: thank marc    (1TY7)
 YaronKoren: Peter: people type in RDF directly?    (1TY8)
 PeterDolog: well, they can, if they have an editor in their configuration    (1TY9)
 PeterDolog: but that is not what usually happen to normal user    (1TYA)
 PeterDolog: this is why we actually introduced tagging in the KIWI project    (1TYB)
 PeterDolog: I cannot imagine a project manager - one of the use cases we are targeting    (1TYC)
 PeterDolog: who would write RDF    (1TYD)
 YaronKoren: Right.    (1TYE)
 PeterDolog: some people in the project are working on some simple editor for kind of rdf annotations    (1TYF)
 PeterDolog: they are actually text analysis people    (1TYG)
 PeterDolog: and are currently making some drag and drop highliting plugin    (1TYH)
 PeterDolog: to get those derived in writing directly    (1TYI)
 PeterDolog: but user in fact does not see that it is RDF    (1TYJ)
 PeterDolog: but we would like to have a bit more than just tags    (1TYK)
 PeterDolog: so we (meaning the whole project) are thinking how to support that    (1TYL)
 YaronKoren: Right, interface is important.    (1TYM)
 PeterDolog: But tagging seems to be acceptable    (1TYN)
 PeterDolog: simple tagging is actually not semantic    (1TYO)
 PeterDolog: so one way to extend it is to somehow support editing of links between ontologies and tags    (1TYP)
 MarcFeickert: can Poticus work for a non-semantic wiki? We'd love to play with it now    (1TYQ)
 PeterDolog: Yaron, both architecture and some details on the whole KIWI data model is described 
             here:    (1TYR)
 PeterDolog: in case, you'd like to know more    (1TYS)
 YaronKoren: Okay, thanks.    (1TYT)
 YaronKoren: Ploticus=awesome.    (1TYU)
 JoelNatividad: Thanks Yaron    (1TYV)
 MikeBennett: @Joel: thanks for your kind mention of the EDM Council Semantics Repository    (1TYW)
 JoelNatividad: It would not have been possible without all the help and feedback    (1TYX)
 JoelNatividad: you gave during the dev process    (1TYY)
 JoelNatividad: Hi Mike!    (1TYZ)
 YaronKoren: Hey, I'm glad I did.    (1TZ0)
 JoelNatividad: So glad to see you here    (1TZ1)
 MikeBennett: Hi Joel - good to see you also, the Ploticus stuff looks really exciting.    (1TZ2)
 MarcFeickert: I almost wet my pants    (1TZ3)
 MarcFeickert: very cool    (1TZ4)
 YaronKoren: I'm looking forward to the time-series stuff - animated graphs? Damn.    (1TZ5)
 YaronKoren: Jennifer: that's too bad. That's the first I've heard of people not being able to find pages after they create them.    (1TZ6)
 YaronKoren: (Although maybe people just don't tell me about it.)    (1TZ7)
 YaronKoren: ...created using forms, that is.    (1TZ8)
 RaviSharma: Joel: Excellent semantic plugin through Queries, it does have general value such as in BI OLAP Reports, 
             my own opinion, you can probably offer to Excel or other databses this as a add-on similar to what 
             financial analysts use as excel @risk, but very useful and is this likelty to be open source?    (1TZ9)
 JoelNatividad: @Ravi: it is open source and the good thing about CSV is that it opens in Excel without a problem    (1TZA)
 JoelNatividad: @Ravi: perhaps, we can create Excel Bundles too in addition to the PDF Bundle    (1TZB)
 JenniferVendetti: Yaron:  sorry - I was perhaps speaking too fast during my 3 minute talk.  What I meant to say 
                   is that sometimes users enter something slightly incorrect in a form and then their plug-in 
                   is listed under an area that they didn't anticipate.  The fix would be to correct the category 
                   or property that they entered, but they don't always seem to know what they have done wrong 
                   because they are unaware of the strict ontological structure in place for the plug-ins library.    (1TZC)
 YaronKoren: Oh, okay, that makes sense.    (1TZD)
 JenniferVendetti: ... even though we have documented this.    (1TZE)
 YaronKoren: Jennifer: I think, having seen your forms to some extent, that there might be ways to make them a little more usable.    (1TZF)
 YaronKoren: Like, for instance, having a dropdown for the plugin name using the "values from category" parameter.    (1TZG)
 JenniferVendetti: Yaron: Great - I am very happy for your suggestions.  I will look at this values from category parameter. 
                   Sounds like it would be great for solving data entry problems.    (1TZH)
 YaronKoren: You can also have the title of the new page be automatically-generated, if the name is meant to just fit a format based on the page's data.    (1TZI)
 YaronKoren: These are both relatively new features.    (1TZJ)
 JenniferVendetti: Oh - I didn't know about the auto-generation of new page names.  Sounds like we should use this as well.  Thank you for this suggestion!!    (1TZK)
 YaronKoren: Sure.    (1TZL)
 RaviSharma: to speakers: is there a uniform tagging and query standard emerging other than GRRDLE such as agreed Meta-tags etc?    (1TZM)
 MikeAxelrod: bye all!    (1TZN)
 Duane Searsmith: thanks!    (1TZO)
 MarcFeickert: thanks for all the fish    (1TZP)
 EnricoDaga: thank you all    (1TZQ)
 PeterDolog: thanks, bye all    (1TZR)
 MikeBennett: Thanks all, great talks    (1TZS)
 TomEskridge: thanks    (1TZT)
 PeterYim: Thank you all for participating ... another great session!    (1TZU)
 JoelNatividad: thanks to ontolog for hosting the session!    (1TZV)
 PeterYim: thanks to JieBao and LiDing for putting this together ... and to all our wonderful speakers for sharing their insights with us!    (1TZW)
 PeterYim: bye! ... full proceedings, along with audio archives will be out shortly!    (1TZX)

Audio Recording of this Session    (1TBY)

For the records ...    (1TEU)

How To Join (while the session is in progress)    (1TC8)