7. • Results are difficult to understand by the end
users
• Applications need to provide explanations
along with the flow of information
6
8. • Justification: metadata about why a
conclusion has been drawn
• Justifications themselves can be RDF data
distributed across the Web
7
9. Linked Justifications
• Applying the very approach of linked data to
publish justifications
– Use URIs as names of things (justifications and
their components)
– Use HTTP URIs
– Provide useful information on lookups
– Include links to the related URIs
(justifications/resources)
8
10. • Abstraction of justifications to provide
explanation; navigation between explanations
• Tracing the origin of inferences
– DBPedia live and the chains of dependent
inferences
9
13. AcadWiki:Scientist AcadWiki:ComputerScientist
rdfs:subClassOf
Data AcadWiki:Bob AcadWiki:ComputerScientist
rdf:type
AcadWiki:Bob GeoWiki:London
AcadWiki:birthPlace
AcadWiki:Bob Bob was born in London
AcadWiki:Scientist
rdf:type
r4ta:justifies r4ta:justifies
r4ta:justifies
Justifications
r4ta:antecedent
r4ta:antecedent
r4ta:justifies
AcadWiki London is part of UK Bob’s birthplace is UK
because Bob was born in
because London is part of
England and England is Londonand London is
part of UK part of UK
r4ta:antecedent
GeoWiki
r4ta:antecedent
r4ta:justifies
r4ta:justifies
GeoWiki:London GeoWiki:UnitedKingdom
GeoWiki:isPartOf Academician Locator
AcadWiki:Bob GeoWiki:UnitedKingdom
AcadWiki:birthPlace
12
14. • Named graphs
– Referencing triples/graphs
– Grouping justification related triples
J. J. Carroll, C. Bizer, P. Hayes, and P. Stickler. Named graphs, provenance and trust. In Proceedings of the 14th international conference on World
Wide Web, WWW ’05, pages 613–622, New York, NY, USA, 2005. ACM. 13
15. #graph for an inferred triple
aloc:t1 {
AcadWiki:BobAcadWiki:birthPlaceGeoWiki:UnitedKingdom.
}
#graph justifying the assertion of an inferred triple
aloc:j1 {
Link to the triple
aloc:j1 rdf:type Justification.
aloc:j1 r4ta:justifies aloc:t1.
aloc:j1 r4ta:antecedent AcadWiki:j4. Link to justifications
aloc:j1 r4ta:antecedent GeoWiki:j1.
aloc:t1 rdf:type r4ta:InferredAssertion.
aloc:t1 r4ta:inferredByRule aloc:pobRule. Link to rule
}
14
17. • Consuming Linked Justifications
– Transforming to human understandable
presentation for explanation
– Navigation between related explanation allowing
follow-your-nose principle
– Tracing the origins of chains of inferences
(reasoning/truth maintenance)
16
18. Related Work
• Laconic and precise justifications [Horridgeet
al., 2008]
– fine-grained subclasses of justifications (laconic
justifications and precise justifications)
– algorithm to compute laconic justifications
– our focus is on publishing and consuming distributed
justifications
• Proof Markup Language [McGuinneset al.,2007]
– provenance, information about information manipulation
steps and trust
– we have a narrower focus: justifications
•M. Horridge, B. Parsia, and U. Sattler. Laconic and precise justifications in owl. In Proceedings of the 7th International Conference on The Semantic Web,
ISWC ’08, pages 323–338, Berlin, Heidelberg, 2008. Springer-Verlag.
•D. McGuinness, L. Ding, P. Da Silva, and C. Chang. Pml 2: A modular explanation interlingua. In In AAAI 2007 Workshop on Explanation-aware Computing,
2007.
17
19. • Reasoning, explanation, reason maintenance and
semantic wikis [Kotowski and Bry, 2010]
– Explanation in incremental development of
knowledge bases
– reason maintenance:efficient knowledge base updates
– do not discuss representation
• Linked data and provenance in biological data
webs [Zhao et al.,2009]
– design patterns to encode provenance information
using named graphs
•J. Kotowski and F. Bry. A perfect match for reasoning, explanation and reason maintenance: Owl 2 rl and semantic wikis. In Proceedings of 5th Semantic
Wiki Workshop, Hersonissos, Crete, Greece (31st May 2010), 2010.
•J. Zhao, A. Miles, G. Klyne, and D. Shotton. Linked data and provenance in biological data webs. Briefings in bioinformatics, 10(2):139–152, 2009.
18
20. Provenance, Justification, Explanation
• Justifications: Why someone holds a belief,
explanation of why the belief is true, how one
knows what one knows. – Wikipedia
– Justifications justify why conclusions are drawn.
• Provenance: “sources of information, such as
people, entities, and processes, involved in
producing, influencing, or delivering a piece of
data or a thing in the world”. – W3C Provenance
ontology
19
21. Provenance, Justification, Explanation
• Explanation: sets of statements to describe a
set of facts in order to clarify the
causes, context, and consequences of those
facts. – Wikipedia
– Main purpose is better understanding through an
objective explanation.
20
22. Future Work
• How other provenance vocabularies can be used together
– Combine with provenance vocabularies
• Provide context
• Better understanding and trust [McGuinnes et al., 2007]
• W3C Provenance Model
– A core data model for provenance
– Ratio4TA as a specialization of the W3C PROV Ontology (PROV-O)
• r4ta:Assertion, r4ta:InferredAssertion – prov:Entity
• r4ta:Justification – prov:Trace, prov:Account,
prov:ProvenanceContainer
• r4ta:InferenceRule – prov:Plan
• r4ta:inferredByRule – prov:hadPlan, prov:Activity
• r4ta:justifies – prov:traceTo, prov:qualifiedTrace
• r4ta:antecedent – prov:specializationOf
21
23. • Summarization of justifications
Rules for finding components in the graphs of justification graphs
22
24. • Level of granularity
– Single triple
– Set of triples
• how to deal with overlapped triples in this level
• Efficient storage and query
– Huge number of graphs
23
25. Thank You
Ratio4TA http://ns.inria.fr/ratio4ta/v1/
http://kolflow.univ-nantes.fr/
http://wimmics.inria.fr/
24
Editor's Notes
Producecontinuously changing and interlinked Semantic Web dataTake for example DBpedia
Each of this changes may trigger to addition or deletion of many triplesOpen Archives Initiative Protocol for Metadata Harvesting (OAI-PMH)DBpedia-Live can on average process about 105 pages per minute on average.
Imagine a scenario where applications use the data produced by these collaborative platformsDeriving new information from these continuously changing interlinked datasets which are distributed across the webRESULTS IN: EVER CHANGING INTERLINKED DATASPACENew information derived from these data continuously need updates changes must be reflected in the dependent applicationit is important to allow tracing back the origins of inferences in order to understand if a given inferred statement is up-to-date with regard to all the changes.
The knowledge created by the collaborative applications are continuously changing and evolving. Therefore, the inferences that these applications make change and evolve. The chains of applications that are depended on these collaboratively created knowledge continuously need to update the inferences they make. In such a scenario, it is important to allow tracing back the origins of inferences in order to understand if a given inferred statement is up-to-date with regard to all the changes.Furthermore, a user in a collaborative knowledge management application such as a semantic wiki might be more knowledgeable about the schema, the facts and the inferences in the semantic wiki in which he is working. The user might be less knowledgeable about such information from the external data sources. He might encounter pieces of information that have been inferred from the information from external data sources.
Linked Data is a set of principles for publishing of structured data in the form of RDF graphs, so that they can be explored and navigated in a manner analogous to the HTML WWW and be understandable by computers. These principles are:1. Using URIs as names of things2. Using HTTP URIs, so that people can look up those names3. Providing useful information when someone looks up a URI4. Including links to other URI, so people can discover more thingsThe linked data concept is an enabling factor for the realization of the Semantic Web as a global web of structured data.
A truth maintenance platform that enables tracing the origin of inferences. Linked justifications will provide the basis for such a platform by enabling to determine the triples that might be affected if a triple is modified or removed, and thus avoid the inefficient incremental approach of removing and recomputing all the inferred triples
Each wiki include triples representing data and triples representing justification metadataGroup together triples that represent a justificationLink related justifications from the justifications
Named graphs enable data and metadata to be decoupled in a modular way
Justifications can be used to determine the triples that might be affected if a triple is modified or removed thus avoid the inefficient incremental approach of removing and recomputing all the inferred triples~\\cite{PMS-FB-2010-5}. Linked justifications will provide the basis for a truth maintenance platform such as the one we discuss.
Horridge \\textit{et al.} present two fine-grained subclasses of justifications called laconic justifications and precise justifications~\\cite{Horridge:2008:LPJ:1483155.1483182}. Laconic justifications are the justifications whose axioms do not contain any superfluous parts. Precise justifications are derived from laconic justifications and each of whose axioms represents a minimal part of the justification. The authors also present an optimisedalgorithm to compute laconic justifications showing the feasibility of computing laconic justifications and precise justifications in practice. In contrast to this work, we focus on a platform for justifications in a distributed environment. We do not focus on the theoretical aspects of the justifications such as the minimal parts of axioms in a justification which are required to hold an entailment. Rather, we focus on the aspects related to providing a platform for publishing and consuming justifications in a distributed environment.PMLPML supports capturing provenance, information about information manipulation steps and trust.We have a narrower focus as we do not consider encoding manipulation steps, trust related information, or the generic provenance related information. Our focus is on the representation of inference dependencies between triples in form of justifications in a distributed environment.
Semantic WikiKotowski and Bry~\\cite{PMS-FB-2010-5} argue that explanation complements the incremental development of knowledge bases in the frequently changing wiki environments.reason maintenance: efficient knowledge base updates (using the justifications stored for the derivations)Do not discuss how justifications are represented, publishedDo not discuss how this can be done in a distributed settingBiological dataThe authors provide design patterns to encode provenance information of the mapping linksusing named graphsOur encoding approach is inspired by these design patterns. However, we do not consider a broader notion of provenance as Zhao \\textit{et al.} do. We focus on encoding justifications for the assertions of triples.
Causes: justificationContext: provenance
- Using other provenance vocabularies to provide additional information for explanation- Missing information -> factual information modified by users in a particular revision of a wiki page.If you want to understand a justification, you may also want to know the provenance ( e.g. where the data come from, when they were retrieved etc)[SelftNote] r4ta:antecedent – prov:specializationOf, might need to keep the links of antecedent triples in addition to the antecedent justifications, rethinkThe idea that a single way of representing and collecting provenance could be adopted internally by all systems does not seem to be realistic today. Instead, a pragmatic approach is to consider a core data model for provenance that allows domain and application specific representations of provenance to be translated into such a data model and exchanged between systems. Heterogeneous systems can then export their provenance into such a core data model, and applications that need to make sense of provenance in heterogeneous systems can then import it, process it, and reason over it.
Two dimensionsTraversing links for related justificationsZooming in for details