View
462
Download
0
Category
Preview:
DESCRIPTION
LWA 2008, Workshop Knowledge Management (FGWM)
Citation preview
Introduction Motivation Foundations Approach Implementation Evaluation Related Conclusion
Arguing on Issues with Mathematical Knowledge Itemsin a Semantic Wiki
LWA/FGWM 2008
Christoph Lange, Tuukka Hastrup, Stéphane Corlosquet
Jacobs University, Bremen, GermanyKWARC – Knowledge Adaptation and Reasoning for Content
Digital Enterprise Research Institute, National University of Ireland, GalwayThis work was supported by JEM-Thematic-Network ECP-038208.
October 7, 2008
Lange/Hastrup/Corlosquet (Jacobs U./DERI) Arguing on Issues with Math. Knowledge Items in a Wiki October 7, 2008 1
Introduction Motivation Foundations Approach Implementation Evaluation Related Conclusion
Resource Edit Discussion History
Issue Anna 2008–05–30 [Idea][Argument][Agree][Disagree][Decision]
It’s hard to find out how to improve content (= resources) in wikis /
Agree Bert 2008–05–31
Indeed, besides automated approaches it’s hard to get focused feedback from users.
Idea Carl 2008–06–01 [Argument][Agree][Disagree][Decision]
So let’s make wiki discussions semantic!
Argument Doro 2008–06–02 [Agree][Disagree]
We could take types from an argumentation ontology for the posts.
Argument Erwin 2008–06–03 [Agree][Disagree]
And every discourse should be connected to resources corresponding to the wikipage, and there should be domain-specific Idea and Issue subclasses.
Agree Anonymous 2008–06–04
That’s great, then the wiki could assist with the realisation of an approved idea.And old decisions would be documented.
Decision Christoph 2008–06–05
So let’s do it! (Available in the semantic wiki IkeWiki, domain-specific extensionfor mathematics in SWiM presented here)
Lange/Hastrup/Corlosquet (Jacobs U./DERI) Arguing on Issues with Math. Knowledge Items in a Wiki October 7, 2008 2
Introduction Motivation Foundations Approach Implementation Evaluation Related Conclusion
Resource Edit Discussion History
Issue Anna 2008–05–30 [Idea][Argument][Agree][Disagree][Decision]
It’s hard to find out how to improve content (= resources) in wikis /Agree Bert 2008–05–31
Indeed, besides automated approaches it’s hard to get focused feedback from users.
Idea Carl 2008–06–01 [Argument][Agree][Disagree][Decision]
So let’s make wiki discussions semantic!
Argument Doro 2008–06–02 [Agree][Disagree]
We could take types from an argumentation ontology for the posts.
Argument Erwin 2008–06–03 [Agree][Disagree]
And every discourse should be connected to resources corresponding to the wikipage, and there should be domain-specific Idea and Issue subclasses.
Agree Anonymous 2008–06–04
That’s great, then the wiki could assist with the realisation of an approved idea.And old decisions would be documented.
Decision Christoph 2008–06–05
So let’s do it! (Available in the semantic wiki IkeWiki, domain-specific extensionfor mathematics in SWiM presented here)
Lange/Hastrup/Corlosquet (Jacobs U./DERI) Arguing on Issues with Math. Knowledge Items in a Wiki October 7, 2008 2
Introduction Motivation Foundations Approach Implementation Evaluation Related Conclusion
Resource Edit Discussion History
Issue Anna 2008–05–30 [Idea][Argument][Agree][Disagree][Decision]
It’s hard to find out how to improve content (= resources) in wikis /Agree Bert 2008–05–31
Indeed, besides automated approaches it’s hard to get focused feedback from users.
Idea Carl 2008–06–01 [Argument][Agree][Disagree][Decision]
So let’s make wiki discussions semantic!
Argument Doro 2008–06–02 [Agree][Disagree]
We could take types from an argumentation ontology for the posts.
Argument Erwin 2008–06–03 [Agree][Disagree]
And every discourse should be connected to resources corresponding to the wikipage, and there should be domain-specific Idea and Issue subclasses.
Agree Anonymous 2008–06–04
That’s great, then the wiki could assist with the realisation of an approved idea.And old decisions would be documented.
Decision Christoph 2008–06–05
So let’s do it! (Available in the semantic wiki IkeWiki, domain-specific extensionfor mathematics in SWiM presented here)
Lange/Hastrup/Corlosquet (Jacobs U./DERI) Arguing on Issues with Math. Knowledge Items in a Wiki October 7, 2008 2
Introduction Motivation Foundations Approach Implementation Evaluation Related Conclusion
Resource Edit Discussion History
Issue Anna 2008–05–30 [Idea][Argument][Agree][Disagree][Decision]
It’s hard to find out how to improve content (= resources) in wikis /Agree Bert 2008–05–31
Indeed, besides automated approaches it’s hard to get focused feedback from users.
Idea Carl 2008–06–01 [Argument][Agree][Disagree][Decision]
So let’s make wiki discussions semantic!
Argument Doro 2008–06–02 [Agree][Disagree]
We could take types from an argumentation ontology for the posts.
Argument Erwin 2008–06–03 [Agree][Disagree]
And every discourse should be connected to resources corresponding to the wikipage, and there should be domain-specific Idea and Issue subclasses.
Agree Anonymous 2008–06–04
That’s great, then the wiki could assist with the realisation of an approved idea.And old decisions would be documented.
Decision Christoph 2008–06–05
So let’s do it! (Available in the semantic wiki IkeWiki, domain-specific extensionfor mathematics in SWiM presented here)
Lange/Hastrup/Corlosquet (Jacobs U./DERI) Arguing on Issues with Math. Knowledge Items in a Wiki October 7, 2008 2
Introduction Motivation Foundations Approach Implementation Evaluation Related Conclusion
Resource Edit Discussion History
Issue Anna 2008–05–30 [Idea][Argument][Agree][Disagree][Decision]
It’s hard to find out how to improve content (= resources) in wikis /Agree Bert 2008–05–31
Indeed, besides automated approaches it’s hard to get focused feedback from users.
Idea Carl 2008–06–01 [Argument][Agree][Disagree][Decision]
So let’s make wiki discussions semantic!
Argument Doro 2008–06–02 [Agree][Disagree]
We could take types from an argumentation ontology for the posts.
Argument Erwin 2008–06–03 [Agree][Disagree]
And every discourse should be connected to resources corresponding to the wikipage, and there should be domain-specific Idea and Issue subclasses.
Agree Anonymous 2008–06–04
That’s great, then the wiki could assist with the realisation of an approved idea.And old decisions would be documented.
Decision Christoph 2008–06–05
So let’s do it! (Available in the semantic wiki IkeWiki, domain-specific extensionfor mathematics in SWiM presented here)
Lange/Hastrup/Corlosquet (Jacobs U./DERI) Arguing on Issues with Math. Knowledge Items in a Wiki October 7, 2008 2
Introduction Motivation Foundations Approach Implementation Evaluation Related Conclusion
Resource Edit Discussion History
Issue Anna 2008–05–30 [Idea][Argument][Agree][Disagree][Decision]
It’s hard to find out how to improve content (= resources) in wikis /Agree Bert 2008–05–31
Indeed, besides automated approaches it’s hard to get focused feedback from users.
Idea Carl 2008–06–01 [Argument][Agree][Disagree][Decision]
So let’s make wiki discussions semantic!
Argument Doro 2008–06–02 [Agree][Disagree]
We could take types from an argumentation ontology for the posts.
Argument Erwin 2008–06–03 [Agree][Disagree]
And every discourse should be connected to resources corresponding to the wikipage, and there should be domain-specific Idea and Issue subclasses.
Agree Anonymous 2008–06–04
That’s great, then the wiki could assist with the realisation of an approved idea.And old decisions would be documented.
Decision Christoph 2008–06–05
So let’s do it! (Available in the semantic wiki IkeWiki, domain-specific extensionfor mathematics in SWiM presented here)
Lange/Hastrup/Corlosquet (Jacobs U./DERI) Arguing on Issues with Math. Knowledge Items in a Wiki October 7, 2008 2
Introduction Motivation Foundations Approach Implementation Evaluation Related Conclusion
Resource Edit Discussion History
Issue Anna 2008–05–30 [Idea][Argument][Agree][Disagree][Decision]
It’s hard to find out how to improve content (= resources) in wikis /Agree Bert 2008–05–31
Indeed, besides automated approaches it’s hard to get focused feedback from users.
Idea Carl 2008–06–01 [Argument][Agree][Disagree][Decision]
So let’s make wiki discussions semantic!
Argument Doro 2008–06–02 [Agree][Disagree]
We could take types from an argumentation ontology for the posts.
Argument Erwin 2008–06–03 [Agree][Disagree]
And every discourse should be connected to resources corresponding to the wikipage, and there should be domain-specific Idea and Issue subclasses.
Agree Anonymous 2008–06–04
That’s great, then the wiki could assist with the realisation of an approved idea.And old decisions would be documented.
Decision Christoph 2008–06–05
So let’s do it! (Available in the semantic wiki IkeWiki, domain-specific extensionfor mathematics in SWiM presented here)
Lange/Hastrup/Corlosquet (Jacobs U./DERI) Arguing on Issues with Math. Knowledge Items in a Wiki October 7, 2008 2
Introduction Motivation Foundations Approach Implementation Evaluation Related Conclusion
Resource Edit Discussion History
Issue Anna 2008–05–30 [Idea][Argument][Agree][Disagree][Decision]
It’s hard to find out how to improve content (= resources) in wikis /Agree Bert 2008–05–31
Indeed, besides automated approaches it’s hard to get focused feedback from users.
Idea Carl 2008–06–01 [Argument][Agree][Disagree][Decision]
So let’s make wiki discussions semantic!
Argument Doro 2008–06–02 [Agree][Disagree]
We could take types from an argumentation ontology for the posts.
Argument Erwin 2008–06–03 [Agree][Disagree]
And every discourse should be connected to resources corresponding to the wikipage, and there should be domain-specific Idea and Issue subclasses.
Agree Anonymous 2008–06–04
That’s great, then the wiki could assist with the realisation of an approved idea.And old decisions would be documented.
Decision Christoph 2008–06–05
So let’s do it! (Available in the semantic wiki IkeWiki, domain-specific extensionfor mathematics in SWiM presented here)
Lange/Hastrup/Corlosquet (Jacobs U./DERI) Arguing on Issues with Math. Knowledge Items in a Wiki October 7, 2008 2
Introduction Motivation Foundations Approach Implementation Evaluation Related Conclusion
Issues with Knowledge Items
Terms used in this talk:Knowledge Item: A piece of knowledge about a distinct subject of interest
Issue: Any problem by which a knowledge item could be affected
ExampleA Wikipedia article is not written from a neutral point of viewA mathematical theorem is hard to understand or wrong
Lange/Hastrup/Corlosquet (Jacobs U./DERI) Arguing on Issues with Math. Knowledge Items in a Wiki October 7, 2008 3
Introduction Motivation Foundations Approach Implementation Evaluation Related Conclusion
Resolving Issues
1 Report: What is affected by what issue2 Propose a solution3 Argue about the solution4 Vote on the solution or on the arguments5 Take a decision6 Implement the decision
BenefitAfterwards, anybody can retrace these steps. Argumentations are partof the collective experience of the community.
Lange/Hastrup/Corlosquet (Jacobs U./DERI) Arguing on Issues with Math. Knowledge Items in a Wiki October 7, 2008 4
Introduction Motivation Foundations Approach Implementation Evaluation Related Conclusion
Solving Issues in Wikipedia (1)
{{Neutralität}}’’’Würzburg’’’ ist ein trostloses Nest in [[Unterfranken]],wo weder das Bier noch der Wein schmeckt.
Lange/Hastrup/Corlosquet (Jacobs U./DERI) Arguing on Issues with Math. Knowledge Items in a Wiki October 7, 2008 5
Introduction Motivation Foundations Approach Implementation Evaluation Related Conclusion
Solving Issues in Wikipedia (1)
{{Neutralität}}’’’Würzburg’’’ ist ein trostloses Nest in [[Unterfranken]],wo weder das Bier noch der Wein schmeckt.
Lange/Hastrup/Corlosquet (Jacobs U./DERI) Arguing on Issues with Math. Knowledge Items in a Wiki October 7, 2008 5
Introduction Motivation Foundations Approach Implementation Evaluation Related Conclusion
Solving Issues in Wikipedia (2)
Das ist ja wohl eine sehr verkürzte Betrachtung.--[[Benutzer:Alfred|Alfred]] 13:41, 27. Aug. 2008 (CEST):Stimmt. Sollte man nicht noch was zum Weltkulturerbeschreiben? --[[Benutzer:Bernd|Bernd]] 13:47, 27. Aug.2008 (CEST)
::Bin ich sehr dafür. --~~~~
Lange/Hastrup/Corlosquet (Jacobs U./DERI) Arguing on Issues with Math. Knowledge Items in a Wiki October 7, 2008 6
Introduction Motivation Foundations Approach Implementation Evaluation Related Conclusion
Solving Issues in Wikipedia (2)
Das ist ja wohl eine sehr verkürzte Betrachtung.--[[Benutzer:Alfred|Alfred]] 13:41, 27. Aug. 2008 (CEST):Stimmt. Sollte man nicht noch was zum Weltkulturerbeschreiben? --[[Benutzer:Bernd|Bernd]] 13:47, 27. Aug.2008 (CEST)
::Bin ich sehr dafür. --~~~~
Lange/Hastrup/Corlosquet (Jacobs U./DERI) Arguing on Issues with Math. Knowledge Items in a Wiki October 7, 2008 6
Introduction Motivation Foundations Approach Implementation Evaluation Related Conclusion
Solving Issues in Wikipedia (3)
Does that work? – Sufficiently, because:
Quote [Brändle]Zu wenige Köche verderben den Brei
Wikipedia has a large user base. . . and deals with topics of general interest
Our system does neither!
Lange/Hastrup/Corlosquet (Jacobs U./DERI) Arguing on Issues with Math. Knowledge Items in a Wiki October 7, 2008 7
Introduction Motivation Foundations Approach Implementation Evaluation Related Conclusion
Semantic Wikis
Generally: wikis that employ semantic web technologies like RDF orontologiesMost commonly: 1 article = 1 knowledge itemArticles and links have types (taken from ontologies)
ApplicationsIn Würzburg: Problem solving using expert knowledge (KnowWE)In Karlsruhe: Semantic MediaWiki ; Semantic Wikipedia
Lange/Hastrup/Corlosquet (Jacobs U./DERI) Arguing on Issues with Math. Knowledge Items in a Wiki October 7, 2008 8
Introduction Motivation Foundations Approach Implementation Evaluation Related Conclusion
Semantic Discussions
SIOC (Semantically Interlinked Online Communities)ontology for user-generated content, particularly online discussionsapplied in forums, blogs, content management systems, . . .
. . . and in one (!) semantic wiki – which ours is based on
Did you know?Discussions in semantic wikis are rarely semantic /
Lange/Hastrup/Corlosquet (Jacobs U./DERI) Arguing on Issues with Math. Knowledge Items in a Wiki October 7, 2008 9
Introduction Motivation Foundations Approach Implementation Evaluation Related Conclusion
Wicked Problems
WarningIssues can be wicked problems
A wicked problemdoes not allow for a definitive formulationhas solutions that are not true-or-false but good-or-badThere is no immediate and ultimate test of a solution
IBIS: model for structured argumentation about wicked problems
Lange/Hastrup/Corlosquet (Jacobs U./DERI) Arguing on Issues with Math. Knowledge Items in a Wiki October 7, 2008 10
Introduction Motivation Foundations Approach Implementation Evaluation Related Conclusion
Argumentation Ontologies
DILIGENT: IBIS applied to ontology engineering
DILIGENT workflow1 Issue (“some concept is not covered by our ontology”, “some
requirement is not satisfied”)2 Idea (how to conceptualise/formalise it)3 Arguments challenge or justify ideas4 Positions: opinions/votes on ideas5 Decision: points to Issue, Idea, and supporting votes
Lange/Hastrup/Corlosquet (Jacobs U./DERI) Arguing on Issues with Math. Knowledge Items in a Wiki October 7, 2008 11
Introduction Motivation Foundations Approach Implementation Evaluation Related Conclusion
Domain-Specific Argumentation Ontology
Agree
Position
Disagree
Decision
subClassOf
Issue
Inappropriatefor Domain
Wrong Incomprehensible
subClassOf
Idea
ProvideExample
Keep asBad Example
Delete
subClassOfresponseTo
positionOn positionOn
onIssue onIdea
withPositions
OntologyEntity
resolvesInto
Math. Know-ledge Item
Theorem Example
subClassOf
subClassOf
DILIGENTargumentationontology
Domain-specificargumentationclasses (partly shown)
OMDoc ontology
……
Applicable Issue/Idea types depend on type of knowledge item
Lange/Hastrup/Corlosquet (Jacobs U./DERI) Arguing on Issues with Math. Knowledge Items in a Wiki October 7, 2008 12
Introduction Motivation Foundations Approach Implementation Evaluation Related Conclusion
Issues in Mathematics
50 participants, experienced in mathematical knowledge managementSurvey is still open: http://tinyurl.com/5qdetd
Common granularities of knowledge: course unit, mathematicaltheory, mathematical statementLittle software support for issue tracking and solving knownKnowledge item type most frequently affected by issues: definitionsCommon issues: wrong, incomprehensible, truth uncertain,underspecified, redundantCommon solutions: directly improve affected knowledge item, split itIssues remain unresolved due to insufficient restructuring support
Lange/Hastrup/Corlosquet (Jacobs U./DERI) Arguing on Issues with Math. Knowledge Items in a Wiki October 7, 2008 13
Introduction Motivation Foundations Approach Implementation Evaluation Related Conclusion
Automated Assistance
1 Find legitimate issuesnot yet decided, more agreement than disagreement
2 Find best solution proposal (idea)highest agreement/disagreement ratio
3 Assist with implementationsystem knows common idea and issue types
Lange/Hastrup/Corlosquet (Jacobs U./DERI) Arguing on Issues with Math. Knowledge Items in a Wiki October 7, 2008 14
Introduction Motivation Foundations Approach Implementation Evaluation Related Conclusion
Implementation
SWiM (Semantic Wiki for Mathematical Knowledge Management)one page holds one mathematical statement
DILIGENT ; SIOC Argumentation Module (work in progress)SPARQL queries for finding the winning ideas
Lange/Hastrup/Corlosquet (Jacobs U./DERI) Arguing on Issues with Math. Knowledge Items in a Wiki October 7, 2008 15
Introduction Motivation Foundations Approach Implementation Evaluation Related Conclusion
Structure of a Use Case
discussion page
knowledgeitems
(OMDoc ontology)on wiki pages
theoremforum1
example
post1: Issue(Incomprehensible)
post6: Decision
post3: Agree
post2: Idea(ProvideExample)
post4: Disagree
post5: Agree
exemplifies
hasDiscussion(IkeWiki ontology)
has_container
has_replyresponseTo
resolvesInto
positionOn
onIdea
withPositionsonIssue
physical structure(SIOC)
argumentativestructure
(DILIGENT)
Lange/Hastrup/Corlosquet (Jacobs U./DERI) Arguing on Issues with Math. Knowledge Items in a Wiki October 7, 2008 16
Introduction Motivation Foundations Approach Implementation Evaluation Related Conclusion
Structure of a Use Case
discussion page
knowledgeitems
(OMDoc ontology)on wiki pages
theoremforum1
example
post1: Issue(Incomprehensible)
post6: Decision
post3: Agree
post2: Idea(ProvideExample)
post4: Disagree
post5: Agree
exemplifies
hasDiscussion(IkeWiki ontology)
has_container
has_replyresponseTo
resolvesInto
positionOn
onIdea
withPositionsonIssue
physical structure(SIOC)
argumentativestructure
(DILIGENT)
Lange/Hastrup/Corlosquet (Jacobs U./DERI) Arguing on Issues with Math. Knowledge Items in a Wiki October 7, 2008 16
Introduction Motivation Foundations Approach Implementation Evaluation Related Conclusion
Demo (1)
Lange/Hastrup/Corlosquet (Jacobs U./DERI) Arguing on Issues with Math. Knowledge Items in a Wiki October 7, 2008 17
Introduction Motivation Foundations Approach Implementation Evaluation Related Conclusion
Demo (2)
Lange/Hastrup/Corlosquet (Jacobs U./DERI) Arguing on Issues with Math. Knowledge Items in a Wiki October 7, 2008 18
Introduction Motivation Foundations Approach Implementation Evaluation Related Conclusion
Demo (3)
Lange/Hastrup/Corlosquet (Jacobs U./DERI) Arguing on Issues with Math. Knowledge Items in a Wiki October 7, 2008 19
Introduction Motivation Foundations Approach Implementation Evaluation Related Conclusion
Critical Questions
Coverage of the model: No arguments so far (will make it harder to findthe winning idea)
Structure required: Knowledge items already need to have types (butDILIGENT-like approach to conceptualising/formalising fromscratch would be possible)
Wicked problems? Standard problems with standard solutions are notreally wicked.But we give the user enough freedom to discussnon-standard problems as well.
Lange/Hastrup/Corlosquet (Jacobs U./DERI) Arguing on Issues with Math. Knowledge Items in a Wiki October 7, 2008 20
Introduction Motivation Foundations Approach Implementation Evaluation Related Conclusion
OpenMath Case Study
OpenMath Content Dictionary Wiki: http://wiki.openmath.orgsmall group of knowledge engineersadding and revising semiformal definitions of math. symbolsintensive discussions, formerly on a mailing list
Lange/Hastrup/Corlosquet (Jacobs U./DERI) Arguing on Issues with Math. Knowledge Items in a Wiki October 7, 2008 21
Introduction Motivation Foundations Approach Implementation Evaluation Related Conclusion
Practical Evaluation
Idea:Given: a knowledge base where we intentionally break someknowledge itemsTask: fix it!One group has argumentation feature enabledThe other group doesn’tFind out which group is more effective
Lange/Hastrup/Corlosquet (Jacobs U./DERI) Arguing on Issues with Math. Knowledge Items in a Wiki October 7, 2008 22
Introduction Motivation Foundations Approach Implementation Evaluation Related Conclusion
Related Work
Foucault@Wiki: How to edit summaries and discussion posts relate tochanges made to Wikipedia articles?
Lekapidia: developing a new ontology with a DILIGENT-based wiki:detect inconsistent arguments, foster consensus
Cicero: Arguing and voting on problem solutions using SemanticMediaWiki
Collaborative Protégé: collaborative ontology editorDrupal: Versatile Content Management System, could port our
implementation therepanta rhei: interactive and collaborative reader for mathematical
documents; discussion forums and ratings
Lange/Hastrup/Corlosquet (Jacobs U./DERI) Arguing on Issues with Math. Knowledge Items in a Wiki October 7, 2008 23
Introduction Motivation Foundations Approach Implementation Evaluation Related Conclusion
Summary
Knowledge bases need structured discussions for better problemsolving and documentationWe have integrated an argumentation ontology into a semantic wiki. . . offering domain-specific extensions and assistance
Lange/Hastrup/Corlosquet (Jacobs U./DERI) Arguing on Issues with Math. Knowledge Items in a Wiki October 7, 2008 24
Introduction Motivation Foundations Approach Implementation Evaluation Related Conclusion
Scale it to the web
Users can have accounts on multiple social media sitesSIOC allows for modeling thisThere are crawlers and indexers for RDF data on the web⇒ we can find about arguments and actions of the same user onother sites. . . and use it for the purpose of our own site
ExampleA user on site A proposes lots of solutions to issues which get positivefeedback and are then implemented ⇒ This is a competent user!Site B, dealing with a related topic, might be interested in him.
Lange/Hastrup/Corlosquet (Jacobs U./DERI) Arguing on Issues with Math. Knowledge Items in a Wiki October 7, 2008 25
Recommended