reasoning in semantic wikis
TRANSCRIPT
![Page 1: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/1.jpg)
Reasoning in Semantic Wikis
Markus Krötzsch1 Sebastian Schaffert2 Denny Vrandecic1
1 Institut AIFB, Universität Karlsruhe, Germany2 Salzburg Research, Austria
Published under the terms of CC-BY-NC 3.0 (→ details)
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 1 / 39
![Page 2: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/2.jpg)
The Semantic Web Vision
[Berners-Lee, Hendler, Lassila, 2001]:
“A new form of Web content that is meaningful to computers willunleash a revolution of new possibilities”
“Computers will find the meaning of semantic data by followinghyperlinks to definitions of key terms and rules for reasoning aboutthem logically.”
“Naturally, you want to check this, so your computer asks the servicefor a proof of its answer, which it promptly provides by translating itsinternal reasoning into the Semantic Web’s unifying language.”
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 2 / 39
![Page 3: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/3.jpg)
The Reality of 2007
“This simple idea, however, remains largely unrealized.”[Shadbolt, Hall, Berners-Lee, 2006]
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 3 / 39
![Page 4: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/4.jpg)
The Reality of 2007
“This simple idea, however, remains largely unrealized.”[Shadbolt, Hall, Berners-Lee, 2006]
Wikipedia classifies “Semantic Web” into the category “Buzzwords”(together with Web 2.0)
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 3 / 39
![Page 5: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/5.jpg)
The Reality of 2007
“This simple idea, however, remains largely unrealized.”[Shadbolt, Hall, Berners-Lee, 2006]
Wikipedia classifies “Semantic Web” into the category “Buzzwords”(together with Web 2.0)
“The Semantic Web is a web of data” [W3C, 2007]
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 3 / 39
![Page 6: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/6.jpg)
The Reality of 2007
“This simple idea, however, remains largely unrealized.”[Shadbolt, Hall, Berners-Lee, 2006]
Wikipedia classifies “Semantic Web” into the category “Buzzwords”(together with Web 2.0)
“The Semantic Web is a web of data” [W3C, 2007]
“Spoken cynically, [. . . ] the Web has received toy engines that neithermeet its requirements nor scale to its size.”[Fensel, van Harmelen, 2007]
“The Contradiction of Web and Reasoning”
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 3 / 39
![Page 7: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/7.jpg)
Overview
1 Semantics in Wikis
2 Semantic MediaWiki and IkeWiki
3 Adding Expressivity
4 Query Languages
5 Performance
6 Closing Remarks
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 4 / 39
![Page 8: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/8.jpg)
Outline
1 Semantics in Wikis
2 Semantic MediaWiki and IkeWiki
3 Adding Expressivity
4 Query Languages
5 Performance
6 Closing Remarks
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 5 / 39
![Page 9: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/9.jpg)
From Websites to Wikis
WikisContent management systems for text (usually)
Web-based (usually)
Collaborative (usually)
Quick and simple (well, compared to XHTML)
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 6 / 39
![Page 10: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/10.jpg)
From Websites to Wikis
WikisContent management systems for text (usually)
Web-based (usually)
Collaborative (usually)
Quick and simple (well, compared to XHTML)
1994: Portland Pattern Repository (“Ward’s Wiki”). . .Jan 15, 2001: Wikipedia is launched (“to add a little feature toNupedia”)
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 6 / 39
![Page 11: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/11.jpg)
From Websites to Wikis
WikisContent management systems for text (usually)
Web-based (usually)
Collaborative (usually)
Quick and simple (well, compared to XHTML)
1994: Portland Pattern Repository (“Ward’s Wiki”). . .Jan 15, 2001: Wikipedia is launched (“to add a little feature toNupedia”)Feb 12, 2001: 1,000 articlesSep 7, 2001: 10,000 articlesJan 22, 2003: 100,000 articlesMar 1, 2006: 1,000,000 articlesToday: 1,987,473 articles, 10,000s of sites, 100s of engines
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 6 / 39
![Page 12: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/12.jpg)
From Wikis to Semantic Wikis
As wikis grow big, retrieving content becomes harder.
ChallengesWhere can I find the answer to my question?
What is the answer to my question?
Which interesting information can the wiki offer to me?
Text-based search mainly solves content location. Aggregating knowledge is a cumbersome manual task. More “structure” needed
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 7 / 39
![Page 13: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/13.jpg)
Structuring Data in MediaWiki
MediaWiki already has ways of structuring content:
Links: browsing and searching
Categories: classifying content pages for browsing
Redirects: treat synonyms efficiently (no unique names . . . )
Some more (namespaces, templates, disambiguation, . . . )
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 8 / 39
![Page 14: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/14.jpg)
Structuring Data in MediaWiki
MediaWiki already has ways of structuring content:
Links: browsing and searching
Categories: classifying content pages for browsing
Redirects: treat synonyms efficiently (no unique names . . . )
Some more (namespaces, templates, disambiguation, . . . )
Main structuring mechanism: pages!
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 8 / 39
![Page 15: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/15.jpg)
Many Approaches . . .
A number of semantic wikis appeared:
KendraBase
Rhizome
Platypus
Semantic MediaWiki (SMW)
WikSAR
KaukoluWiki
IkeWiki
MaknaWiki
OntoWiki
Design choices: Text-annotation, ontology creation, or data collection?Closed domain or open community? SemWeb experts or not? Smallor large scale?
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 9 / 39
![Page 16: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/16.jpg)
Outline
1 Semantics in Wikis
2 Semantic MediaWiki and IkeWiki
3 Adding Expressivity
4 Query Languages
5 Performance
6 Closing Remarks
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 10 / 39
![Page 17: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/17.jpg)
Semantic MediaWiki (SMW)
Semantic MediaWikiExtension module for MediaWiki (see Wikipedia)
Free (GPL)
Developed since 2005, 2–5 developers
Current version: SMW 1.0 (release pending)
Used
Documentation and download: http://ontoworld.org/wiki/SMWFurther Reading: [Kr. et al., 2007]
→ try it out SMW on ontoworld.org
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 11 / 39
![Page 18: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/18.jpg)
Properties and Types in SMW
SMW adds properties:
Associate pages with “values”
Introduce new properties as needed
Syntax: [[Property::value]]
Different kinds of values: numbers, other pages, strings, dates,locations, . . .
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 12 / 39
![Page 19: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/19.jpg)
Properties and Types in SMW
SMW adds properties:
Associate pages with “values”
Introduce new properties as needed
Syntax: [[Property::value]]
Different kinds of values: numbers, other pages, strings, dates,locations, . . .
Properties can have different types:
Properties are associated with types
Types are mostly built-in, but possibly customisable
Syntax: [[has type::Datatype]]
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 12 / 39
![Page 20: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/20.jpg)
Browsing and Querying
Why should anyone annotate pages?
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 13 / 39
![Page 21: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/21.jpg)
Browsing and Querying
Why should anyone annotate pages?
“Which interesting information can the wiki offer to me?” Browsing (→ online example)
Use semantics for finding “related” informationSimple interfaces: just clicking around
“What is the answer to my question?” Querying (→ online example)
Simple wiki-based query languageInline queries: dynamically embedd results into pages
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 13 / 39
![Page 22: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/22.jpg)
Formal Semantics: Mapping to OWL
Formal grounding of SMW annotations: OWL DL mapping
SMW DL OWLSimple Pages Individual names OWL-IndividualsCategory pages Class names OWL-ClassesProperty pages Roles OWL-PropertiesType pages Individual names OWL-Individuals
ObjectProperty, DatatypeProperty, or AnnotationProperty?Depends on type of SMW-property!
Simple annotations of a page “p”:
[[Category:C]] 7→ p : C (or C(p))
[[R::v]] 7→ (p, v) : R (or R(p, v))
Also available as OWL DL export (OWL/RDF format)→ online example
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 14 / 39
![Page 23: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/23.jpg)
IkeWiki
IkeWikiStand-alone wiki implementation
Free (GPL)
Developed since 2005, 1 developer (more to come)
Current version: IkeWiki 1.99.39beta
Used in projects
Screencast and download:http://ikewiki.salzburgresearch.at/
Different focus than SMW:
URIs used as identifiers in the wiki
Free-form RDF editing possible
Some differences in features and interfaces
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 15 / 39
![Page 24: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/24.jpg)
Outline
1 Semantics in Wikis
2 Semantic MediaWiki and IkeWiki
3 Adding Expressivity
4 Query Languages
5 Performance
6 Closing Remarks
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 16 / 39
![Page 25: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/25.jpg)
Making Semantic Wikis more Expressive
So far only basic ABox statements. . .
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 17 / 39
![Page 26: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/26.jpg)
Making Semantic Wikis more Expressive
So far only basic ABox statements. . .
Issues that must be taken into account:
Expressivity vs. Performance:Web applications are performance critical
Usability/debuggability: users must understand logical features
Robustness: the system should be robust to vandalism anderrors
Standardisation: semantics well-documented, interchangepossible, simplifies implementation
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 17 / 39
![Page 27: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/27.jpg)
A Simple Example
“The problem is that the word ‘cook’ has no meaning, or semanticcontent, to the computer.” [Berners-Lee, Hendler, Lassila, 2001]
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 18 / 39
![Page 28: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/28.jpg)
A Simple Example
“The problem is that the word ‘cook’ has no meaning, or semanticcontent, to the computer.” [Berners-Lee, Hendler, Lassila, 2001]
Let’s consider a wiki of recipes:Pages about recipes and ingredientsRecipes with textual descriptions and list of ingredientsE.g., the page “Mango pickles” states that this recipe contains300g of green mango.
How can we represent this in terms of binary properties?
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 18 / 39
![Page 29: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/29.jpg)
A Simple Example
“The problem is that the word ‘cook’ has no meaning, or semanticcontent, to the computer.” [Berners-Lee, Hendler, Lassila, 2001]
Let’s consider a wiki of recipes:Pages about recipes and ingredientsRecipes with textual descriptions and list of ingredientsE.g., the page “Mango pickles” states that this recipe contains300g of green mango.
How can we represent this in terms of binary properties?
(pickles, 300g_gr_mango) : contains
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 18 / 39
![Page 30: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/30.jpg)
A Simple Example
“The problem is that the word ‘cook’ has no meaning, or semanticcontent, to the computer.” [Berners-Lee, Hendler, Lassila, 2001]
Let’s consider a wiki of recipes:Pages about recipes and ingredientsRecipes with textual descriptions and list of ingredientsE.g., the page “Mango pickles” states that this recipe contains300g of green mango.
How can we represent this in terms of binary properties?
(pickles, 300g_gr_mango) : contains(pickles, gr_mango) : contains (pickles, 300g) : amount
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 18 / 39
![Page 31: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/31.jpg)
A Simple Example
“The problem is that the word ‘cook’ has no meaning, or semanticcontent, to the computer.” [Berners-Lee, Hendler, Lassila, 2001]
Let’s consider a wiki of recipes:Pages about recipes and ingredientsRecipes with textual descriptions and list of ingredientsE.g., the page “Mango pickles” states that this recipe contains300g of green mango.
How can we represent this in terms of binary properties?
(pickles, 300g_gr_mango) : contains(pickles, gr_mango) : contains (pickles, 300g) : amount(pickles, cayenne) : contains (pickles, 2Tsp) : amount
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 18 / 39
![Page 32: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/32.jpg)
A Simple Example
“The problem is that the word ‘cook’ has no meaning, or semanticcontent, to the computer.” [Berners-Lee, Hendler, Lassila, 2001]
Let’s consider a wiki of recipes:Pages about recipes and ingredientsRecipes with textual descriptions and list of ingredientsE.g., the page “Mango pickles” states that this recipe contains300g of green mango.
How can we represent this in terms of binary properties?
(pickles, 300g_gr_mango) : contains(pickles, gr_mango) : contains (pickles, 300g) : amount(pickles, cayenne) : contains (pickles, 2Tsp) : amount(pickles, gr_mango) : contains (gr_mango, 300g) : amount
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 18 / 39
![Page 33: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/33.jpg)
A Simple Example
“The problem is that the word ‘cook’ has no meaning, or semanticcontent, to the computer.” [Berners-Lee, Hendler, Lassila, 2001]
Let’s consider a wiki of recipes:Pages about recipes and ingredientsRecipes with textual descriptions and list of ingredientsE.g., the page “Mango pickles” states that this recipe contains300g of green mango.
How can we represent this in terms of binary properties?
(pickles, 300g_gr_mango) : contains(pickles, gr_mango) : contains (pickles, 300g) : amount(pickles, cayenne) : contains (pickles, 2Tsp) : amount(pickles, gr_mango) : contains (gr_mango, 300g) : amount(pickles, ing01) : contains(ing01, gr_mango) : ingredient (ing01, 300g) : amount
Semantic wikis should make this easy.M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 18 / 39
![Page 34: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/34.jpg)
Does Your Mum Really Need Logic?
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 19 / 39
![Page 35: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/35.jpg)
Does Your Mum Really Need Logic?
“What can I cook with 3kg of green mango?”
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 19 / 39
![Page 36: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/36.jpg)
Does Your Mum Really Need Logic?
“What can I cook with 3kg of green mango?” data value ranges, unit support
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 19 / 39
![Page 37: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/37.jpg)
Does Your Mum Really Need Logic?
“What can I cook with 3kg of green mango?” data value ranges, unit support
“Which recipes contain mango?”
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 19 / 39
![Page 38: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/38.jpg)
Does Your Mum Really Need Logic?
“What can I cook with 3kg of green mango?” data value ranges, unit support
“Which recipes contain mango?” generalisation to “mango”
taxonomic approach: ing01 : GrMango, GrMango ⊑ Mangorole composition:(gr_mango, mango) : kindOf, ingredient ◦ kindOf ⊑ ingredient
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 19 / 39
![Page 39: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/39.jpg)
Does Your Mum Really Need Logic?
“What can I cook with 3kg of green mango?” data value ranges, unit support
“Which recipes contain mango?” generalisation to “mango”
taxonomic approach: ing01 : GrMango, GrMango ⊑ Mangorole composition:(gr_mango, mango) : kindOf, ingredient ◦ kindOf ⊑ ingredient
“Which recipes do not contain meat?”
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 19 / 39
![Page 40: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/40.jpg)
Does Your Mum Really Need Logic?
“What can I cook with 3kg of green mango?” data value ranges, unit support
“Which recipes contain mango?” generalisation to “mango”
taxonomic approach: ing01 : GrMango, GrMango ⊑ Mangorole composition:(gr_mango, mango) : kindOf, ingredient ◦ kindOf ⊑ ingredient
“Which recipes do not contain meat?” closed world reasoning, role composition
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 19 / 39
![Page 41: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/41.jpg)
Does Your Mum Really Need Logic?
“What can I cook with 3kg of green mango?” data value ranges, unit support
“Which recipes contain mango?” generalisation to “mango”
taxonomic approach: ing01 : GrMango, GrMango ⊑ Mangorole composition:(gr_mango, mango) : kindOf, ingredient ◦ kindOf ⊑ ingredient
“Which recipes do not contain meat?” closed world reasoning, role composition
“Which recipes are unsuitable for mango allergics?”
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 19 / 39
![Page 42: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/42.jpg)
Does Your Mum Really Need Logic?
“What can I cook with 3kg of green mango?” data value ranges, unit support
“Which recipes contain mango?” generalisation to “mango”
taxonomic approach: ing01 : GrMango, GrMango ⊑ Mangorole composition:(gr_mango, mango) : kindOf, ingredient ◦ kindOf ⊑ ingredient
“Which recipes do not contain meat?” closed world reasoning, role composition
“Which recipes are unsuitable for mango allergics?” background ontology (What is a mango allergic? When canmango ingredients be substituted?)
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 19 / 39
![Page 43: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/43.jpg)
Outline
1 Semantics in Wikis
2 Semantic MediaWiki and IkeWiki
3 Adding Expressivity
4 Query Languages
5 Performance
6 Closing Remarks
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 20 / 39
![Page 44: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/44.jpg)
Semantics in SMW
SMW has configurable support for
Class hierarchies
Property hierarchies
Datatypes
N-ary data values
Equality reasoning
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 21 / 39
![Page 45: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/45.jpg)
The SMW Query Language
[[Category:City]] <q> [[population::>500,000]] ||[[located in::
<q>[[Category:Country]] [[member of::EU]]</q>]] </q>
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 22 / 39
![Page 46: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/46.jpg)
The SMW Query Language
[[Category:City]] <q> [[population::>500,000]] ||[[located in::
<q>[[Category:Country]] [[member of::EU]]</q>]] </q>
Expressive features:
Category (class) and property statements
Conjunction and disjunction
Data ranges
Nesting
No variables, universal statements, negation, number restrictions. Queries can be mapped to DL:
City ⊓ (∃population.ge(500000) ⊔∃located_in.(Country ⊓ ∃member_of.{EU}))
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 22 / 39
![Page 47: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/47.jpg)
Complexity of SMW queries (1)
SMW queries use only ⊓, ⊔, ∃, {a} (nominals), ⊤, and data values
Instance retrieval possible in polynomial time
Ideamaterialise answers inside out,computing result sets for subqueries first
City ⊓ (∃population.ge(500000) ⊔∃located_in.(Country ⊓ ∃member_of.{EU}))
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 23 / 39
![Page 48: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/48.jpg)
Complexity of SMW queries (2)
Can we generalise the above result?
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 24 / 39
![Page 49: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/49.jpg)
Complexity of SMW queries (2)
Can we generalise the above result?
The DL EL++
Tractable description logic, featuring ⊓, ∃, {a} (nominals), rolecomposition R ◦ S ⊑ T , data values (concrete domains, if “convex”)
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 24 / 39
![Page 50: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/50.jpg)
Complexity of SMW queries (2)
Can we generalise the above result?
The DL EL++
Tractable description logic, featuring ⊓, ∃, {a} (nominals), rolecomposition R ◦ S ⊑ T , data values (concrete domains, if “convex”)
but SMW also has ⊔ and non-convex concrete domains generally EXPTIME-complete! (even for logic FEL: ⊓, ⊔, ∃)
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 24 / 39
![Page 51: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/51.jpg)
Complexity of SMW queries (2)
Can we generalise the above result?
The DL EL++
Tractable description logic, featuring ⊓, ∃, {a} (nominals), rolecomposition R ◦ S ⊑ T , data values (concrete domains, if “convex”)
but SMW also has ⊔ and non-convex concrete domains generally EXPTIME-complete! (even for logic FEL: ⊓, ⊔, ∃)
ObservationDisjunctions in SMW do not add real complexity:
(A ⊔ B)? 7→ A? ∪ B?
general approach: Horn Description Logics source for theoretical results . . .
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 24 / 39
![Page 52: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/52.jpg)
Semantics in IkeWiki
IkeWiki has configurable support for
class and property hierarchies
domain and range restrictions
OWL DL inferencing if installed
experimental Prolog binding (unstable)
SPARQL as a query language
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 25 / 39
![Page 53: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/53.jpg)
Queries in IkeWiki
“SPARQL” 7→ “SPARQL Protocol And RDF Query Language”
Almost a W3C standard (hopefully)
Based on graph pattern matching + filters
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 26 / 39
![Page 54: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/54.jpg)
Queries in IkeWiki
“SPARQL” 7→ “SPARQL Protocol And RDF Query Language”
Almost a W3C standard (hopefully)
Based on graph pattern matching + filters
PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>PREFIX ex: <http://www.example.org/>SELECT ?resultWHERE { ?result rdf:type ex:City .
{ ?result ex:locatedIn ?place .?place rdf:type ex:Country .?place ex:memberOf ex:EU . } UNION
{ ?result ex:population ?number .FILTER (?number >= 500000) }
}
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 26 / 39
![Page 55: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/55.jpg)
Semantics and Complexity of SPARQL
Operational semantics based on SPARQL algebra:
Join(BGP(?result rdf:type ex:City.),Union(BGP(?result ex:locatedIn ?place .
?place rdf:type ex:Country .?place ex:memberOf ex:EU . ),
Filter( ?number >= 500000,BGP(?result ex:population ?number . )
))
)
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 27 / 39
![Page 56: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/56.jpg)
Semantics and Complexity of SPARQL
Operational semantics based on SPARQL algebra:
Join(BGP(?result rdf:type ex:City.),Union(BGP(?result ex:locatedIn ?place .
?place rdf:type ex:Country .?place ex:memberOf ex:EU . ),
Filter( ?number >= 500000,BGP(?result ex:population ?number . )
))
)
SPARQL is hard:
Basic graph pattern matching is already NP-complete
Matching SPARQL graph patterns is PSPACE-complete
Filters are also tricky
However: data complexity is LOGSPACE
Note: no schema, just RDF with simple entailmentM. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 27 / 39
![Page 57: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/57.jpg)
SPARQL beyond RDF
Operational SPARQL algebra vs. model theoretic OWL semantics What does SPARQL for OWL mean?
Conjunctive Queries (CQs) for DLsConjunctive queries similar to Datalog, e.g.
Person(x) ∧ livesIn(x , y) ∧ bornIn(x , y)
more expressive than DL querying
Remark: still not quite the same as SPARQL!
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 28 / 39
![Page 58: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/58.jpg)
Complexity of Conjunctive Querying
Conjunctive queries stricly harder than DL reasoning alone:
2-EXPTIME for SHOQ and SHIQ[Glimm, Horroks, Lutz, Sattler, 2007; Lutz, 2007]
PSPACE for (restricted) EL++ [Kr., Rudolph, Hitzler, 2007]
NP if TBox is empty (plain RDF)
Complexity/decidability for SHOIN (OWL DL) unknown!
Difficult in a Web application . . .
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 29 / 39
![Page 59: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/59.jpg)
Outline
1 Semantics in Wikis
2 Semantic MediaWiki and IkeWiki
3 Adding Expressivity
4 Query Languages
5 Performance
6 Closing Remarks
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 30 / 39
![Page 60: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/60.jpg)
Practical Performance
Complexity 6= Performance:
“Hard tasks” might be feasible in practice
“Easy tasks” might still perform poorly
ExampleKB-complexity (and data complexity) for CQs on EL++ is P.
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 31 / 39
![Page 61: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/61.jpg)
Practical Performance
Complexity 6= Performance:
“Hard tasks” might be feasible in practice
“Easy tasks” might still perform poorly
ExampleKB-complexity (and data complexity) for CQs on EL++ is P.Bounded by some multiple of kqq
(k size of KB, q size of query)!
“Polynomial” is often still too bad for practice
On a real site
SMW takes less than 9% of overall processing time, but
40% − 50% of this time is query answering.
(http://ontoworld.org/profileinfo.php)
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 31 / 39
![Page 62: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/62.jpg)
Reasoning Implementations in SMW and IkeWiki
SMW:
built-in semantic store
based on MySQL
binding via storage abstraction layer
IkeWiki:
Jena store as backend
Support for RDFS and OWL inherited from Jena
Binding via DIG interface
Clean interfaces enable easy extension. Try it.
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 32 / 39
![Page 63: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/63.jpg)
Practical Results
Basic evaluations and experiences:
IkeWiki for OWL: up to around 10,000 statements
IkeWiki for RDFS: up to around 50,000 statements
SMW with simple KB: tested with about 13,800,000 statements
Disclaimer:Performance greatly depends on schema/queries!Even a hundred statements are too much if you can choose the query!
Sensible restrictions on queries needed
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 33 / 39
![Page 64: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/64.jpg)
Reasoners Wanted!
A wiki-reasoner must . . .. . . tolerate large datatsets, even if reasoning is too slow
(Jena in-memory crashes quite instantly with 1.3Mio triples),
. . . support basic datatype operations,
. . . support basic retrieval options (order by, limit, offset),
. . . support incremental updates,
. . . not impose non-local syntactical constraints on axioms,
. . . crash only when truly necessary
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 34 / 39
![Page 65: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/65.jpg)
Reasoners Wanted!
A wiki-reasoner must . . .. . . tolerate large datatsets, even if reasoning is too slow
(Jena in-memory crashes quite instantly with 1.3Mio triples),
. . . support basic datatype operations,
. . . support basic retrieval options (order by, limit, offset),
. . . support incremental updates,
. . . not impose non-local syntactical constraints on axioms,
. . . crash only when truly necessary
A wiki-reasoner should . . .. . . support some syntactical KB inspection (string-matching?),
. . . provably implement a well-documented semantics,
. . . have some approach towards non-unique names,
. . . be free software (no, not as in beer)
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 34 / 39
![Page 66: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/66.jpg)
Outline
1 Semantics in Wikis
2 Semantic MediaWiki and IkeWiki
3 Adding Expressivity
4 Query Languages
5 Performance
6 Closing Remarks
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 35 / 39
![Page 67: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/67.jpg)
Summary
ConclusionIn 2007, semantic wikis bear great challenges and opportunities forreasoning reasearch.
SMW and IkeWiki need and use reasoning
More expressivity is needed
Complexity results are only the first step
Extending SMW/IkeWiki is easy:software is free, we cooperate, and we hire ;-)
(and, of course, you can also just use them)
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 36 / 39
![Page 68: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/68.jpg)
Outlook
[Berners-Lee, Hendler, Lassila, 2001]:
“Knowledge representation, as this technology is often called, iscurrently in a state comparable to that of hypertext before the advent ofthe Web:”
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 37 / 39
![Page 69: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/69.jpg)
Outlook
[Berners-Lee, Hendler, Lassila, 2001]:
“Knowledge representation, as this technology is often called, iscurrently in a state comparable to that of hypertext before the advent ofthe Web:”
“it is clearly a good idea,
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 37 / 39
![Page 70: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/70.jpg)
Outlook
[Berners-Lee, Hendler, Lassila, 2001]:
“Knowledge representation, as this technology is often called, iscurrently in a state comparable to that of hypertext before the advent ofthe Web:”
“it is clearly a good idea,
and some very nice demonstrations exist,
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 37 / 39
![Page 71: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/71.jpg)
Outlook
[Berners-Lee, Hendler, Lassila, 2001]:
“Knowledge representation, as this technology is often called, iscurrently in a state comparable to that of hypertext before the advent ofthe Web:”
“it is clearly a good idea,
and some very nice demonstrations exist,
but it has not yet changed the world.”
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 37 / 39
![Page 72: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/72.jpg)
Further reading I
Tim Berners-Lee, James Hendler, and Ora Lassila.The Semantic WebScientific American, May 2001. → Available online.
Birte Glimm, Ian Horrocks, Carsten Lutz, and Uli Sattler.Conjunctive Query Answering for the Description Logic SHIQProc. IJCAI 2007 (→ Download)
Markus Krötzsch, Sebastian Rudolph, and Pascal Hitzler.Conjunctive Queries for a Tractable Fragment of OWL 1.1Proc. of ISWC’2007 (→ Download)
Markus Krötzsch, Denny Vrandecic, Heiko Haller, Max Völkel, andRudi Studer.Semantic WikipediaJournal of Web Semantics, 2007, to appear. (→ Download)
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 38 / 39
![Page 73: Reasoning in Semantic Wikis](https://reader031.vdocuments.us/reader031/viewer/2022020703/61fb43362e268c58cd5c1736/html5/thumbnails/73.jpg)
Further reading II
Carsten LutzInverse Roles Make Conjunctive Queries HardDL-Workshop 2007 (→ Download)
Semantic MediaWikiOnline Dokumentation and Installation GuideAvailable at http://ontoworld.org/wiki/SMW, 2007.
Nigel Shadbolt, Wendy Hall, and Tim Berners-Lee.The Semantic Web RevisitedIEEE Intelligent Systems, May/June 2006. (→ Download)
World Wide web ConsortiumSemantic WebOnline at http://www.w3.org/2001/sw/ (Sept. 2007)
M. Krötzsch (AIFB Karlsruhe) Semantic Wiki Reasoning Reasoning Web 2007 39 / 39