Minutes for November 22 2008 SMW Users Meeting at Monitor Group
From Semantic MediaWiki Community Wiki
On a frosty morning in Cambride MA some 30 people gathered at the Monitor Group's office in 2 Canal Park.
[edit] Agenda
- Introductions and Welcome
- Global Outreach
- Academic issues
- Corporate usage
- Future developments
- SMW best practices
Here are the minutes, by section:
See also the chat log
[edit] Introductions and Welcome
Yaron introduces himself, smw hosting site, shameless plug for hosting site
Tom Eskeridge Institiue for human cognition etc, has ontology experience, ontology visualization
Molly Forbes, familysearch, using for genealogical research
Don Strigem, Software engineer, investigating smw for search and classification (gene's)
Mark Fieckert, learned about smw at last year, knowledge eng. comp. linguist, sold familysearch on SMW
Fabian Howahl, Univ. Karl. visiting MIT, will give talk about SMW and his research (exhibit project)
Dave, vulcan, halo,
Mike Axelrod, smw built large MW install for Blue cross blue shield, now at ontoprise working w/
Daniel Hansch, ontoprise, managing halo project, smw+, bringing smw into commercial
Henrik Oppermann, onto, managing department, PM, business side, pushes in research projects, funded by EU
Harold Solbrig, Mayo clinic, worked w/ smw since one month after inception, NCI thesaurus, NCI wiki ontology, Biomed GT (SMW, Halo, forms) allows federated ontology curation
Robert Freimuth, Mayo, new to SMW, director of vocab knowledge center, lex wiki
Patrick Sp. mindexchange, to substitute corp. intranet w/ SMW, telecom (tmobile), implemented social network based on smw, daimler also
Mike Cariaso, bio-team, builds smw db's for bio, medical research, genealogy, OLPC
Neil Davies, leads tech at eMonitor
Dave Forgianni, quality assurance at eMonitor
Guoqian Jiang, Mayo clinic, development of smw, collaborative ontology development
Clarence Dillon, science apps int. corp, defense contractor, DOD, looking at techs to use to stage db's for analysis of irregular warfare, analyze why terrorist choose that path, how they pick targets, social scientist
Rich Slezak, front-end dev, uses personally to track cycling trips etc
?Rui Huang, undergrad at RPI, developing google map extension
Zhenning Shangguan,
Jie Bau, RPI, uses smw for one year, test bed for semweb technologines, formalization of concept modeling, reasoning apps, uses everday for blog, task mgmt
Li Ding, research at RPI - starting to adeopt smw, trying to implement previously unimplenmentable ideas using SMW, uses of SMW to develop semweb - RPI, map system
Evan Patton, RPI student, devleops dist. query system to access multiple wikis and consoldiate
Mark Greaves, former DARPA, DAMO program (launched semweb, got RDF approved by w3c), now at vulcan, funds ontoprise, extensions include SMW+, programming practices, "rocket fuel for smw"
Markus and Denny via Audiolink
[edit] Global Outreach
View slide here (pdf)
(need to merge this info w/ that on the discussion page)
One possible forum,
In addition to this site (Referata) see also the SMW Forum [3]
Vulcan is working on supporting
- A community music site
- Corporate use
- Start ups
- Conferences
- Semtech [4]
- European Semantic Conference
- A gathering point
Monitor (LeveragePoint)
- Global Sustainability Exchange
- Global Learning Plan Exchange
Need For Quick Tutorials that go can on places like YouTube
Get O'Reilly involvement - publish official tutorials through them, either online or printed
Who are the audiences?
- General Business Users
- User Communities
- Developer Communities
- ICT Communities
- Academic
We need to drive this to all of the above communities.
But, the biggest break though would be to get Wikipedia to adopt SMW! There are some discussions on going about technical tests. But there are still some serious scalability issues and we need to understand what the end-user value is for Wikipedia. This would be a good topic for future discussion.
Partial Impressionistic and of course Incomplete Notes
Steven - should we have a talk on this at Semtech?
Mark - should we really focus so much on Wikipedia, which is a very high bar, or should we focus on the easier to achive areas?
Daniel - We need to provide a lot more clarity to potential adopters on where this is going.
Mike - There are many, many use cases that we need to address. We need to be precise about where SMW is strong. This is where the relations needed in the future are not known. We need to have a blend of both open, free form ontologies and formal ontologies.
Yaron - Referata and Wikia both provide hosting services.
Mark - At ISWC there was an embedded SMW on a memory stick, these are available from Ontoprise at request.
Clarence - Most people do not care about the technology, they care about the value provided.
Steven and Daniel - Both sell solutions that use SMW and not the application or technology. What is being sold is the benefit and the pain addressed.
Joel - Got interested through TBL, wrote a script to crawl a RDB and build a SMW.
Some things we need to do
- Have some high-profile open sites
- Have some big-name corporate customers, even if the sites are behind their firewalls
- Have some easily available high-powered apps
- Have a variety of guides, videos, etc.
- Speak, speak, speak at conferences
Let's try to keep this open and support open transfer of information and the open evolution of ontologies.
[edit] Academic issues
[edit] RPI talks
[edit] Li Ding, Jie Bao et al., "Toward Social Webtop using Semantic Wiki"
View presentation here (pdf)
Comments and Notes (needs gardening)
Tetherless World Group WIki
- using queries to generate pages that are useful and up-to-date (also nice design/look and feel), people page, publications
- mashup map, event, tag, considering merging map extensions (keep alignment on similar extension capabilities
- queries a wiki to embed content in a different wiki
- ? Next to ND: caching for this ext?
- ?MG: are these under GPL, where are they, is source available?
- is under GPL, no promotion as of yet
- source is available on RPI site
MG: we need to have alignment on publication of extensions in dev/available, suggests sourceforge Li Ding: semweb.org also
Simple export ext:
(see slide)
- n3 (embedded in wiki page), tabulator view
SMW best practices (Templates)
- How to share? export, then import, or copy and paste
- typed link, property doesn't really show up, templates provided by rpi expose this info
- YK: where is this?
- It is a template that uses existing extension, wiki template/markup used as a programming language (mentions 1MB cap?)
SUmmary of ONgoing Work
- Apps
- Extensions
- Templates
- Theory
Future Direction
- Academic research
- Good user interaction (50% of dev time spent on this)
- Scalability, temp table was full, had issue executing query (could be mySQL config issue)
DH: really is it a set of methods and best practice or extensions? if make category and give name, it has UID, in different language that won't work
- JB: different levels by page
DH: fears a missing technical layer, building an app on top of SMW, no IDE, how to debug
- JB: SMW provides a very good infrastructure, something missing that RPI and libraries can provide (java and c++), hope is that SMW templates in development can be reused by others
[edit] Evan Patton, "Distributing Queries: Sharing information in a world-wide wiki"
View presentation here (pdf)
Comments and Notes (needs gardening)
Sharing info in a ww wiki
AAAA Slogan
AskMany Extension
- have one wiki ask another wiki a question
- 2 parts request and receive
Why not use ExportRDF?
- efficiency, don't want whole triple store
Future Work
- communicate using rdf
- parallelize
- assert entries which are the same (different languages, similar property structure [e.g. name:, has name:])
- add support for partial queries
- access control (by triple/object/page?)
tw.rpi.edu/wiki/askmany
?YK: do both wiki's need to have it? can you use special:ask?
- yes, in future can use sparql
?MG: ACL's, what have you done on ACL?
- been using ACL, our own
MG: is it available? We need alignment on ACL extension (one here, one there..)
MG: MediaWiki will not use ACL
Yie B: e.g. task mgmt that they use, people don't want some others to see their tasks. protection on triple level, hard to code, seeks further collaboration.
MG: need one validated system to protect triples
ND and YB: always need to set up a new wiki to control access
[edit] Rui Huang, "RPI Map and Tetherless Map extension"
Links:
Comments and Notes (needs gardening)
Motivation
- limitations of the official campus map
Solution
- SMW + google map
RPI Map Architecture (slides)
-Form -Wikitext -Parsers -Template -SMW RDF db -Query -Result -Google Map api -customer
-Features (slide)
TetherlessMap ext
Two parser functions
Semantic Map layer
(slide) Building Search Integrated KML file
[edit] Jie Bao, "Concept Modeling on Semantic Wiki"
View presentation here (ppt)
Comments and Notes (needs gardening)
For research For fun and daily life
1. A formalization of SMW -(slide) modeling language -
2. Templage-based Rule Modeling
3. OWL Meta Modeling (by Zhenning Shuangguan)
(see slides)
Formal definition of query language for use on disparate DB systems
SMW-QL -logic programs
[edit] Zhenning Shangguan, "OWL Meta Modeling on SMW"
View presentation here (pdf)
Comments and Notes (needs gardening)
Goal: encoding OWL ontologty constructs in wiki
Features
Example Page
Example Class Description
Example Property Description
Ex. Ind. Description
Conclusion
URL: tw.rpi.edu/dev/cnl
JB: foraml semantics will allow more powerful queries
DV: extensions of current?
[edit] Fabian Howahl, "Semantic MediaWiki and Exhibit"
View presentation here (pdf)
Comments and Notes (needs gardening)
Building Bridges
Executive Summary
Integration of SMW and exhibit
Semantic Web and SMW have become islands (non-connected)
exhibit - faceted browsing (see simile project mit)
Three Steps
Exhibit has semantic results format
Passes json file from queried wiki to asking wiki
? can you browse the entire wiki content using exhibit ? json gives a syntax, but begs question of common labels and models, we need a community terminology (e.g. FOAF, DC, SKOS)
[edit] Mayo Clinic talks
[edit] Guoqian Jiang, "LexWiki Framework & Use Cases"
View presentation here (ppt)
Comments and Notes (needs gardening)
- Tools
- Framework
- Use Cases
- Discussion Points
- Demo
caBIG Vocabulary Knowledge Center
Goal: create IT program that redefines how research is conducted online
common infrastructure, common tools, vocab/terminology ISO 11179 standard for metadata
Not limited to biological research (need more info)
- SMW use facilitates ontology change requests from community
- Turnaround time, NCI - monthly updates to terminology, HL7 quarterly
- MG: is this BioMed GT? Yes.
Harold: LexWiki and BiomedGT: LexWiki is a shared set of templates based on a model, allows import, editing and export to Protege? (see wiki page)
- Parallels between goals of this meeting and goals of Vocabulary Knowledge Center, opportunities for synergism between SMW group and VKC?
LexWiki part
Motivation (see slides)
Ontology development becomes community based
Why Semantic Wikis?
[list slide titles and link to ppt and links to URL]
list three tools LexWiki/LexGrid/Protege
? DH: how many users?
- BMed GT, goto special:statistics to see number of writers (and readers?)
- recently introduced semantic graphs?
- searchbox at the top after usability study results
DH: did you customize?
- PR: yes, we changed the autocomplete function of halo
- Problem halo works best in article namespace, they are using mostly category namespace, where they don't get all the features of HALO. article-based. Large namespace support, a syntax for article names where first few chars of article name are the actual namespace...
[edit] Robert Freimuth, "caBIG Overview"
View presentation here (ppt)
(these notes may be merged w/ first Mayo presentation notes above, no further data available please add if have) (needs gardening)
[edit] Harold Solbrig, "Importing Terminology into Semantic MediaWiki"
View presentation here (pdf)
(these notes may be merged w/ first Mayo presentation notes above, no further data available please add if have) (needs gardening)
[edit] Lunch Discussion
(only the tail end caught here, if anyone has additional notes please add)
-Problems submitting bug reports/fixes to HALO extension, updating source code
-Yaron's model works well, svn access, send YK diffs
-NCI challenge does not allow rapid turnaround, if i change a template, is that a code change?
[edit] Corporate usage
[edit] Daniel Hansch, "Ontoprise and SMW" (title may be incorrect)
Comments and Notes (needs gardening)
Popular Usage
- callhome feature, info about install is sent back to Karlsrue to analyze installed base data
Applications on top of SMW
- like a virus, it can be installed for a small group in an organization and then grow to be used throughout
Team Collaboration - german lead innovation project
- project mgmt application, no workflow
- no blank page syndrome (pre-configured ontology)
- eliminates some of the pain of agreement on initial data models
- competes against CMS and PM tools (e.g. drupal, in re: to process metadata?)
- working domain is different than PM domain, smw fuses this leading to easier adoption
Compliance - PriceWaterhouse
- ensure accounting practices are adhered to globally/varied departments
- multi-language support
UNESCO Repository for resources about oceanography
- ww.oceanteacher.net
- wisywig nice but doesn't handle annotations, users said wysiwig important
- Approval workflow
- ?DF: does this exemplify some kind of ACL power? port this functionality to lower level objects?
Vocab mgmt Volkswagen AG
- Problem statement
- Success Factors
Siemens AG - Employee something...
How does Ontoprise bring SMW to the market?
- examples of why easy
[edit] Open Forum for Discussion
What is the path for corporate adoption?
- Advantages
- Disadvantages
- Common Objections
- What is the business model for companies that want to do this (companies that want to make a business out of supporting SMW) ?
Comments:
Mike C.: he's built at least 10 SMW's for people. Uses PerlWikipedia and PyWikipediahe's
- What are competing apps?
- .xls, sharepoint
- integration w/ sharepoint
Mike A.: conservative businesses understand the need for a controlled vocabulary. front-line mgrs terminology is constantly changing. unstructured vs. structured data. get permission from top, then talk to ground workers and see how to implement their changing techniques and vocabs.
NCI-Mayo guys: tried to enforce, jimmy-style, a rigid ontology that lost a lot of the dialog that was valuable. 'the money is in the expert, it doesn't matter if what they say doesn't fit into ontology/vocabulary'.
MG: key sale is a line of business sale, not an IT/data center sale. wiki installations come in as 'line of business' sale.
?: perlscript used to create semantic MW instantly based on data model? used to bridge structured and unstructured
TRAC? subversion + wiki. post-commit hook, harvest metadata from artifact upon submission into svn
MG: What is the intent behind OWL/DL?
SF to MG: How do VC's look at opensource companies?
- MG: very critically. working on models for this. ecosystem will look like the following: there will be "Tool Co.", a business whic his concerned w/ open source, community mgmt, evangelism around a "platform" (e.g. SMW), then there will be App Co's, selling apps for the platform (?).. build an app w/ a lot of IP on top of it. Also "Visualization" companies.
[edit] Future Developments
[edit] Yaron Koren, "Future Developments"
SMW
- Category format
- Support for true n-ary relations (e.g. X assumed the title of president in X year)
Sematic Forms
- FCK Editor
- add Halo support into FCK editor
- SF API, would help w/ integrating w/ other apps, also could be used in workflow editing (change a field in form will auto-update category etc)
- support for SMW's #declare funciton
- combobox input
- editable table input (.xls) replace multiple instance templates (could be used w/ n-ary relations)
- YB more graphical support, routines added (e.g. create toolbar, easier for people for creating forms etc)
- Solicit style-sheets for better looking forms
- Category tree in auto-complete?
Semantic Drilldown
- free input of text and dates
- alternate formats for results
- ui improvements
Semantic Layers
- extension fully released
- mapping extension that uses openlayers
- can map out any image
Data Transfer
- import xml
- suggest xml transform 'begin/end'?
Halo Extension (see slides) {insert ppt) (insert/combine slide titles/main points)
- support for smw data store ineroperation, can store in both jena and smw triple store
- allows to retrieve query results much more quickly
- advisable to have both (redundancy), triples stores tend to break down on occasion
- consider rendering technical metadata tied to article? not the content, only the triples
- intercepting the query
- if have jena that contains triples that do not exist in smw store, and you query jena, you will see red links?
- good question from Li Ding that didn't get captured !!
Web services comments:
- comment: need to not only fetch a page (using web service), but need to update 'one link away'
- should be handled by smwupdate engine
- is it also for 'inserting page', not 'materialization' of new content via web service call, if time 'materialization' of web services will be implemented in core. e.g. send an email to smw and have it make an article. e.g. automated workflow, if use triple/sparql to populate a table based on a query, if red links exist want to go and search wiki for pages that match red links and convert red link to a link to an existing page.
- what web services are you going to provide?
- we'll provide bugzilla web service (already provided), amazon web service to retrieve items and prices
- need wordnet support, can lead to light inferencing
- how to map vocabulary into web service call?
- define mapping which maps individual parts of a dataset into a template/newly created article
Data API comments:
Use case: have content mgmt system, want to send content to smw and write it when being created in SMW, also support full-text retrieval
- interface roadmap?
Improved semantic forms: 'schema last' assumption = start annotating and as it evolves some parts become stable and thus formalized (build it up first, see what comes to the top/frequently appears, and thus is the data model)
- missed Li Ding's comment here regarding some kind of catch-22
Improved Retrieval I Improved Retrieval II
Improved Semantic tree view
- believes this is a core component
Project mgmt bundle
DF: see segetech http://oss.segetech.com/bugzilla-svn-wiki.html
[edit] SMW best practices
1- Note these two notes sections need to be merged
- Mike A.: users add references for data
- Michael: automate addition of data
- Many people: usage of SMW for collaboration on documents?
- Steven: bottom-up vs. top-down creation of semantic data (e.g., annotations vs. forms)
- Harold: good to be able to assert that two properties equal one another
- Steven: no consensus; community should be sensitive to the two different philosophies
- An emerging issue is how to build and model mutlilingual semantic sites, we are looking for best practices here
- Li: how to handle n-ary data, until a true solution comes in SMW - use subpages? For further discussion see [6] and [7]
- Another role for this community is to share information on how we are implementing RDF and OWL standards such as DC, FOAF and SKOS (and SIOC). Over time we need to converge on how we implement standards inside SMW as this will help with federation and integration.
- And maybe what we need is a CNS (like the DNS) a Concept Naming System. An analog is MOAT (Meaning of a Tag) [8]
- Mayo will post a link to examples of how RDF, RDFS, SKOS, etc. tags might be referenced from a wiki. (For a rather broken preview, see: http://www.wiktolog.com/agrowiki/index.php?title=Main_Page)
2 - Note these two notes sections need to be merged
Mike Axelrod: please need notes, missed, was patterns something
Yaron: how to achieve technically
Need Area on SMW forum for best practices
Collaborative papering?
- word2mediawiki macro
- google docs is suggested as a good option for this
Inclusionists vs. Deletionists
Bottom-up vs. Top-down
- Mike A. - sideways method, data in .xls, want it on the wiki, had to meet with them about how to input the data (via form, template. property fields = admin responsible, business person responsible, roles. normalizing on the fly, takes months.
- Mike C. - worst practices.. break everything up into smallest stuff, atomic components, make them parameters in a template. 96-well plate, all lumped into info on a single page (uses perl script to make 96 pages out of that one page). unstructured content goes into template that doesn't get seen.
Templates grow increasing complex
- always write a testpage e.g. template-testpage to check that it is viable as it grows
- don't include and include sections, puts only category info in that area
- templates used as a path to a file
Do people add semantics using just the page instead of templates/forms?
Does halo grab curly's or just square brackets?
On SMW Referata there is a page for links to free template template etc
IN lexwiki, goal is take components of lexwiki and model using RDF/standards, skos community, overlapping DC terms
Standardize how we do standards?!
Ping-o-Matic
MOAT meaning of a tag
Sharing of NCI working draft models?
[edit] Some General Questions
1. Can we get a list of workflow models and/or templates and forms, assign to business types, then review and find best parts of each?
2. Why can't SMW compete directly w/ Sharepoint? Can we scrape a SP site for structure and document names (convert document names to links that would then force employees to locate files and upload to wiki [enabling better version control and collaboration/document building?)
3. Formal procedure for external code contributions. Installers, continuous integration in re: to code submissions from outside sources would need to 1) have a use case/story? 2) not break the build in re: to version control and integration w/ other extensions etc and required software (mysql, php etc, backwards compat?), and then 3) be vetted by appropriate peoples for inclusion given satisfied 1 and 2. Look at mozilla and other open source models in this regard.
[edit] Dinner
Photos from the post-meeting dinner at the Sonesta Hotel, courtesy of Joel Natividad:
[edit] Topics for Sunday
1. Access Control
2. Workflow implementation semantically
3. Halo, notion of a common RDF/triple store via Jena/smw data store integrations?
- group called open ontology repository, wants shared RDF/semantics
Livescribe of Sunday's meeting Note: contact Joel if you need access (smw2008 -at- natividads.com) Treat the ink as Visual Bookmarks. Click on the ink and the audio will jump to what was being said while the corresponding inking was being written.
