Date post: | 11-Aug-2015 |
Category: |
Technology |
Upload: | enrico-daga |
View: | 57 times |
Download: | 2 times |
A Bottom-Up Approach for Licences Classification and Selection
1
Enrico Daga (KMi-OU)Mathieu d’Aquin (KMi-OU)Aldo Gangemi (CNR & Universite’ Paris13) Enrico Motta (KMi-OU)
June 1st, 2015 Interna0onal Workshop on Legal Domain And Seman0c Web Applica0ons ESWC 2015 Workshop hCp://cs.unibo.it/ledaswan2015
Background
• Licences are important in the web of (linked) data• We can describe licences/policies (RDF+ODRL)• Exploiting these descriptions for license selection
is hard due to the high number of different features
2
Question
How can we reduce the effort for licence identification and selection?
3
Hypothesis
An ontology defining relevant classes of licences, formed on the basis of the key features of the instances, should facilitate the selection and identification of a suitable licence.
4
Methodology
A bottom-up approach to ontology construction • RDF License Database• Ontology production with Contento• Based on Formal Concept Analysis• User can annotate and prune concepts• Produces an OWL Ontology
• License Picker Ontology (LiPiO)• Evaluation: License Picker Webapp
5
Formal Concept Analysis (FCA)
• A data analysis method• Studies how objects can be hierarchically grouped together
according to their common attributes• Input is a Formal Context
– a binary matrix of objects/attributes• Basic unit is a Concept:
– (O,A) => (Extension,Intension)– Operator ’ … (O,A) is a concept when O’=A and A’=O
• Classifies concepts hierarchically in a concept lattice– Top: all objects, no attr, bottom: all attributes, no obj
6
7
Contentohttp://bit.ly/contento-tool
Support the user in the generation and curation of concept lattices from formal contexts and use them as semantic web ontologies.
Formal Context
Concept LaNce
Modeling (Naming & Pruning)
Export as OWL
Ontology
8
Contentohttp://bit.ly/contento-tool
Support the user in the generation and curation of concept lattices from formal contexts and use them as semantic web ontologies.
Formal Context
Concept LaNce
Modeling (Naming & Pruning)
RDF Licence Database
Licence Picker
Ontology
Export as OWL
Ontology
Fo Co MR LicEx
RDF License Databasehttp://datahub.io/dataset/rdflicense *
~140 licences expressed in RDF/ODRL
9
<hCp://purl.org/NET/rdflicense/cc-‐by-‐nc4.0> a odrl:Policy ; rdfs:label "Crea0ve Commons CC-‐BY-‐NC" ; rdfs:seeAlso <hCp://crea0vecommons.org/licenses/by-‐nc/4.0/legalcode> ; cc:legalcode <hCp://crea0vecommons.org/licenses/by-‐nc/4.0/> ; dct:alterna0ve "CC BY-‐NC" ; dct:hasVersion "4.0" ; dct:language <hCp://www.lexvo.org/page/iso639-‐3/eng> ; dct:publisher "Crea0ve Commons" ; dct:0tle "Crea0ve Commons -‐ ACribu0on-‐NonCommercial 4.0 Interna0onal -‐ CC BY-‐NC 4.0"@en ; odrl:permission [ odrl:ac0on cc:Distribu8on , ldr:extrac8on , ldr:reu8liza8on , cc:Deriva8veWorks , cc:Reproduc8on ; odrl:duty [ odrl:ac0on cc:ABribu8on , cc:No8ce] ] ; odrl:prohibi8on [ odrl:ac0on cc:CommercialUse ] ; foaf:logo <hCp://i.crea0vecommons.org/l/by-‐nc/2.0/88x31.png> .
Fo Co MR LicEx
(*) Victor Rodrıguez-Doncel, Serena Villata, and Asuncion Gomez-P ́erez. A dataset of RDF licenses. In Rinke Hoekstra, editor, Legal Knowledge and Information Systems. JURIX 2014: The Twenty-Seventh Annual Conference. IOS Press, 2014.
Formal ContextObjects: Apache 2.0, CC-A-SA 3.0, …Attributes: duty attribution, permission Copy, prohibition derive, …
10
RDF Licence Database
Licence,Policy,Bool Apache,Duty No0ce,True
… (as CSV)
Fo Co MR LicEx
Concept Lattice• Applying FCA, a concept lattice is produced• The top concept representing All the licences, while the bottom
concept, None, includes all the attributes, and no licence. • However, this lattice need to be interpreted…
11Fo Co MR LicEx
Modeling: Naming & Pruning
• Inspect each concept• If it is meaningful annotate it
– eg: “should others be allowed to distribute the work?”
• … otherwise delete it– and the hierarchy is adjusted
accordingly– objects are present in direct
upper concepts, attributes in direct lower
12Fo Co MR LicEx
• Approach: a concept is meaningful if it’s intent introduces new features (policies)
• If not:– it is a union of the intents of the
respective upper concepts,– the respective licences will be
present in (at least one of) the upper concepts,
– no new question need to be asked to identify them.
13
Modeling: Naming & Pruning
Fo Co MR LicEx
Modeling: Naming & Pruning
• With this process the lattice has been reduced significantly.• Almost 20% of the initial lattice • Labels being synthetic names referring to policies/attributes that have
been introduced in that point of the hierarchy • Each concept has a Question
14Fo Co MR LicEx
Export as OWL Ontology
15
rdfs:subClassOfrdf:type
ex:has-‐aBribute . . .
give semantics!
Fo Co MR LicEx
contento_ns:7582 rdf:type contento:Concept ;
rdfs:label "Copyleft" ;
rdfs:comment "Should the licence require to share any derived or combined content under the same terms?" ;
rdfs:subClassOf contento_ns:7856 ,
contento_ns:7678 ,
contento_ns:7956 .
contento_ns:7586 rdf:type contento:Concept ;
rdfs:label "Read" ;
rdfs:comment "Should the licence permit to read (access) the data?" ;
rdfs:subClassOf contento_ns:7648 .
contento_ns:7632 rdf:type contento:Concept ;
rdfs:label "ND" ;
rdfs:comment "Should the licence prohibit derivative works?" ;
The Licence Picker Ontologyhttp://bit.ly/licence-picker-ontology
Can be exploited as a decision tree!
16
Focus is on Key Questions
Fo Co MR LicEx
Evaluation The Licence Picker Webapphttp://bit.ly/licence-picker-webapp
17
The user is engaged in answering questions regarding her requirementsBy applying the Licence Picker Ontology (LiPiO), this task is reduced to answering an average of 3/5 questions and assessing the best licence from a small set of choices.
Conclusions and Future Work• A bottom-up approach on ontology building in this scenario opens
new interesting challenges • The RDF description of licences is an ongoing work • Data has been curated in advance (harmonised)• Data has impact on the whole process!
– explore methods to automate some of the curation tasks (especially pruning)
– to integrate changes in the formal context incrementally – to support the ontology designer in the adaptation of the ontology
• New features: type of asset, jurisdiction …• Contento: a generic tool, need user-based evaluation
18