Albertoni ldq workshop ESWC 2015
Download as PPTX, PDF0 likes878 views
The document discusses a quality metric for linksets in SKOS thesauri to assess multilingual gain, emphasizing the importance of linkset quality alongside dataset quality. It introduces a scoring function for linkset importing, which evaluates how effectively links can enhance subject datasets with information from objects. The work calls attention to linkset quality's critical role in improving thesauri completeness and proposes future developments in scoring functions to further characterize linksets.
![PROPOSAL: LINKSET IMPORTING
A quality scoring function on Linkset to answer
LkImp4p, given a link, evaluates the percentage of
values not present in the subject, but “gainable”
from the Object, through the link
Importing on Links
LkImp4p(ObjEntity, Property, Link, Language) --> [0 %, 100%]
The Overall Linkset Importing is defined as the average
contribution of importing on single links of the Linkset
How good is the Linkset to import object dataset’s
INFO into the subject SKOS Thesaurus?](https://image.slidesharecdn.com/albertonildqworkshop1toshare-150601093743-lva1-app6892/85/Albertoni-ldq-workshop-ESWC-2015-5-320.jpg)
![LkImp4pL(x3, skos:prefLabel, l2, ’en’) = 0
No Multilingual gain as “Dog@en” is already in the subject
LINK IMPORTING: Examples
LkImp4p(ObjEntity,Property,Link, Language) --> [0 %, 100%]
✗](https://image.slidesharecdn.com/albertonildqworkshop1toshare-150601093743-lva1-app6892/85/Albertoni-ldq-workshop-ESWC-2015-7-320.jpg)
![LkImp4p(ObjEntity,Property,Link,Language) --> [0 %, 100%]
LkImp4pL(x3, skos:altLabel, l2, ’it’) =100%
As Cagnolino@it is gained and it is the only altLabel in the
complemeneted Subject
✓
LINK IMPORTING: Examples](https://image.slidesharecdn.com/albertonildqworkshop1toshare-150601093743-lva1-app6892/85/Albertoni-ldq-workshop-ESWC-2015-8-320.jpg)
![LkImp4p(ObjEntity,Property,L,Language) --> [0 %, 100%]
LkImp4pL(x3,skos:altLabel, l2, _ )
Disregarding the language, “_ “ means Unspecified Language
= 50% ,
x3 has another altLabel (Puppy@en)
Then we gain one out two of the altlabels.
✗
✓
LINK IMPORTING: Examples](https://image.slidesharecdn.com/albertonildqworkshop1toshare-150601093743-lva1-app6892/85/Albertoni-ldq-workshop-ESWC-2015-9-320.jpg)
Albertoni ldq workshop ESWC 2015
- 1. A Linkset Quality Metric Measuring Multilingual Gain In SKOS Thesauri name.surname@ge.imati.cnr.it IMATI - Istituto di Matematica Applicata e Tecnologie Informatiche "Enrico Magenes” – GENOVA Section, Consiglio Nazionale delle Ricerche, Italy Riccardo Albertoni, Monica De Martino, Paola PodestàRiccardo Albertoni, Monica De Martino, Paola Podestà
- 2. MOTIVATIONS So many bubbles there, THAT’S SO COOL!! BUT …. Can I exploit that third party data for my OWN ANALYSES? Linking Open Data cloud diagram 2014, by Max Schmachtenberg, Christian Bizer, Anja Jentzsch and Richard Cyganiak. http://lod-cloud.net/
- 3. MOTIVATION “QUALITY IS THE ISSUE WHEN REUSE” Well-founded works on quality for DATASETS, … but … What arrows in LOD are good for ?!? NO GROUND CONCEPTs About what makes a LINKSET suitable for A TARGET APPLICATION Considering LOD’s promise: to make the web Evolving into a Global Data Space LINKSET QUALITY should be AS IMPORTANT AS DATASET QUALITY Riccardo Albertoni
- 4. WHAT IS a LINKSET ? (see VOID) Riccardo Albertoni, Asunción Gómez-Pérez: Assessing linkset quality for complementing third-party datasets. EDBT/ICDT Workshops 2013: 52-59 owl:sameAs linksets skos:exactMatch linksets THIS WORK!!! Linkset LSubject X Object Y Inspired by LusTRE, a framework of interlinked Environmental Thesauri in the EU project
- 5. PROPOSAL: LINKSET IMPORTING A quality scoring function on Linkset to answer LkImp4p, given a link, evaluates the percentage of values not present in the subject, but “gainable” from the Object, through the link Importing on Links LkImp4p(ObjEntity, Property, Link, Language) --> [0 %, 100%] The Overall Linkset Importing is defined as the average contribution of importing on single links of the Linkset How good is the Linkset to import object dataset’s INFO into the subject SKOS Thesaurus?
- 6. LINK IMPORTING: Examples Accurate formal definition in the paper, here I am using a simplified version Lkimp4p(l,P) =100*(1- | Val4PropertyP_in_l's_Subject| |Val4PropertyP_in_l's_Subject È Val4PropertyP_in_l's_Object| ) if denominator = 0 then we define Lkimp4p(...)= 0 Lkimp4p(...)=100% iff there are values in the object and no values in the subject LkImp4p assumes the link is correct…
- 7. LkImp4pL(x3, skos:prefLabel, l2, ’en’) = 0 No Multilingual gain as “Dog@en” is already in the subject LINK IMPORTING: Examples LkImp4p(ObjEntity,Property,Link, Language) --> [0 %, 100%] ✗
- 8. LkImp4p(ObjEntity,Property,Link,Language) --> [0 %, 100%] LkImp4pL(x3, skos:altLabel, l2, ’it’) =100% As Cagnolino@it is gained and it is the only altLabel in the complemeneted Subject ✓ LINK IMPORTING: Examples
- 9. LkImp4p(ObjEntity,Property,L,Language) --> [0 %, 100%] LkImp4pL(x3,skos:altLabel, l2, _ ) Disregarding the language, “_ “ means Unspecified Language = 50% , x3 has another altLabel (Puppy@en) Then we gain one out two of the altlabels. ✗ ✓ LINK IMPORTING: Examples
- 10. LkImp4pL(x5, skos:broader, l3, _) It can be applied to any properties, e.g., skos:broader, In this case, we gain entities instead of RDF literals = 50% , x3 and y3 are mapped so only y6 is considered as a entity gained ✗ ✓ LINKSET IMPORTING: Examples
- 11. EXAMPLE of APPLICATION (eENVPlus) LINKSET IMPORTING can be exploited to check the complementation potential of any SKOS property.. But in the application, we focus on skos:prefLabel and skos:altLabel in order to address INCOMPLETE LANGUAGE COVERAGE INCOMPLETE LANGUAGE COVERAGE arises when skos:prefLabel and skos:altLabel are provided in all the expected languages only for a subset of the thesaurus concepts
- 12. EXAMPLE of APPLICATION (eENVPlus) Multilingual Gain for skos:prefLabel Multilingual Gain for skos:altLabel EARTh LINKSETS to • GEMET (E2GEM - 4365 links) • AGROVOC (E2AGR - 1436 links) Is E2GEM better than E2AGR ?? In term of # links sure!! (4365 links >> 1436 links) The best depends on which set of languages and properties we are focusing on… But al least our Importing quality measure enables in a deeper analysis than #link or link coverage
- 13. CONCLUSION 1 2 1 2 We draw the community attention to the critical issue of linkset quality Applicable for estimating multilingual gain Importing as estimator for the “completeness of complented thesauri” (experimental validation) Two-fold contribution We propose LINKSET IMPORTING to measure the gain when complementing Thesauri Example in the context of EU project eENVplus Future work Further scoring functions to fully characterize linkset quality space and dimensions
- 14. 2° Workshop on Linked Data Quality LDQ2015 at ESWC 2015 - June 1, 2015 – Portorož, Slovenia THANK YOU Questions ?? ALBERTONI@GE.IMATI.CNR.IT