Clustering over the cultural heritage linked open dataset xlendi shipwreck
- 1. Clustering Over the Cultural Heritage Linked Open Dataset: Xlendi Shipwreck Ben Ellefi Mohamed - Nawaf Mohamad Motasem - Sourisseau Jean Christophe – Gambin Timmy - Castro Filipe - Drap Pierre SW4CH 2018
- 2. The Context: Xlendi Shipwreck The Arpenteur Ontology The Xlendi LOD dataset GUI Tool for Clustering Data Linking Discussion Conclusion 1 2 3 4 5 6
- 3. • This work started on July 2014 • The support of the Superintendence of Cultural Heritage of Malta • Financed by the National Research Agency • The framework of the GROPLAN project (http://www.lsis.org/groplan). • This shipwreck is datable to approximately 7th century BC at a depth of 110 m. • As one of the best-preserved archaeological sites in Malta datable to the early Phoenician period. Archaeology: Timmy Gambin (University of Malta), Jean-Christophe Sourisseau, (CCJ, CNRS) Marine operation: Bertrand Chemisky, (COMEX) Photogrammetry: Pierre Drap, (LIS, CNRS) 1 The Context: Xlendi Shipwreck The bulk of the cargo still remains buried in the sediments.
- 4. High Resolution orthophoto From Xlendi
- 6. The archaeological site survey is carried out by photogrammetry. Photograph 3D Reconstruction Frederic Bassemayousse
- 7. 3D / 2D Shape Recognition.
- 8. It is not just about 3D … Analysing the distribution of amphorae can help for understanding this shipwreck. ➢ The nature of the ship ? ➢ The ship destination ? ➢ The origin ? ➢ The content of the cargo ? (…) Clustering artifacts by typologies, by orientations, etc. GROPLAN project (http://www.lsis.org/groplan)
- 9. Xlendi Amphorae Labeling Frederic Bassemayousse Program (JAVA) LOD dataset Photographs Amphorae labeling allows binding: Arpenteur ontology GROPLAN project (http://www.lsis.org/groplan)
- 10. In our model we identify three dimensions: • Typological (height, maximum diameter, volume, ...) • Photogrammetrical process (bundle model, camera, photographs, ...) • Spatial (position, convex envelope, ...) Mohamed BEN ELLEFI, et al. Cultural Heritage Resources Profiling: Ontology-based Approach. In The 2018 Web Conference Companion, 2018, Lyon, France 2 The Arpenteur Ontology
- 11. In our model we identify three dimensions: • Typological (height, maximum diameter, volume, ...) • Photogrammetrical process (bundle model, camera, photographs, ...) • Spatial (position, convex envelope, ...) 2 The Arpenteur Ontology
- 12. Typological Description The set of morphological features that characterizes a given CH typology. ➢ Based on this set of features, we are able to identify the typology of a given CH resource. Xlendi Amphorae Typology Modeling GROPLAN project (http://www.lsis.org/groplan)
- 13. Morphological model (dimension, volume, etc..) Measures are associated to annotations on specific zones Xlendi Amphorae Typology Modeling GROPLAN project (http://www.lsis.org/groplan)
- 16. In our model we identify three dimensions: • Typological (height, maximum diameter, volume, ...) • Photogrammetrical process (bundle model, camera, photographs, ...) • Spatial (position, convex envelope, ...) The Arpenteur Ontology2
- 17. Dives Survey Reports: • Dives 1: The survey begun at 11:00, 8 passes E-W and 6 passes N-S and the record finished at 12:20. ➢ The camera has been disposed at the vertical of the site. • Dives 2: The second survey has been done the 18/07/2014, the survey begun at 11:29 and finished at 12:25. 10 transects E-W and 8 transects N-S ➢ The camera has been disposed at +22.5° (…) Photogrammetrical Process Modeling Photogrammetry Description The model that describes all the involved components in the photogrammetrical process. ➢ Modeling the relation between CH resources and the corresponding cameras settings, i.e. resolution, position, orientation, etc.
- 19. In our model we identify three dimensions: • Typological (height, maximum diameter, volume, ...) • Photogrammetrical process (bundle model, camera, photographs, ...) • Spatial (position, convex envelope, ...) The Arpenteur Ontology2
- 20. Spatial Description The set of features that provide information about the orientation and the location of a given CH resource in a specific geographical area. Spatial Distribution Modeling ➢ The shape➢ The localization
- 22. CIDOC-CRM: http://erlangen-crm.org/ GEOSPARQL: http://www.opengeospatial.org/standards/geosparql External Ontology Linking Arp:SpatialObject skos:closeMatch Geo:SpatialObject Arp:IPoint skos:closeMatch Geo:Point Arp:Artifact rdfs:subClassOf CIDOC- CRM:E24_Physical_M an-Made_Thing Arp:SpatialObject rdfs:subClassOf CIDOC- CRM:E18_Physical_Thi ng
- 24. The Xlendi LOD dataset3
- 25. <owl:NamedIndividual rdf:about="#Amphorae2004628751"> <rdf:type rdf:resource="#Amphorae"/> <hasBoundingBox rdf:resource=“#BoundingBox2030458766"/> <hasTransformation3D rdf:resource=“#Transfo-1953670366/> <hasBellyDiameter rdf:datatype="double">0.3512</> <hasDiameterNeck rdf:datatype="double">0.0785</> <hasName rdf:datatype="string">Amphore_A15</> … </owl:NamedIndividual> Amphora_A15 description: Xlendi.owl
- 26. <owl:NamedIndividual rdf:about="#Amphorae2004628751"> <rdf:type rdf:resource="#Amphorae"/> <hasBoundingBox rdf:resource=“#BoundingBox2030458766"/> <hasTransformation3D rdf:resource=“#Transfo-1953670366/> <hasBellyDiameter rdf:datatype="double">0.3512</> <hasDiameterNeck rdf:datatype="double">0.0785</> <hasName rdf:datatype="string">Amphore_A15</hasName> … </owl:NamedIndividual> <owl:NamedIndividual rdf:about="#BoundingBox1562249660"> <rdf:type rdf:resource="#BoundingBox"/> <hasXMax rdf:datatype="xsd:double">2.84742</> <hasXMin rdf:datatype="xsd:double">2.48428</> <hasYMax rdf:datatype="xsd:double">8.6437</> <hasYMin rdf:datatype="xsd:double">8.04083</> <hasZMax rdf:datatype="xsd:double">-100.398</> <hasZMin rdf:datatype="xsd:double">-100.504</> </owl:NamedIndividual> Amphora_A15 description: Xlendi.owl
- 27. <owl:NamedIndividual rdf:about="#Amphorae2004628751"> <rdf:type rdf:resource="#Amphorae"/> <hasBoundingBox rdf:resource=“#BoundingBox2030458766"/> <hasTransformation3D rdf:resource=“#Transfo-1953670366/> <hasBellyDiameter rdf:datatype="double">0.3512</> <hasDiameterNeck rdf:datatype="double">0.0785</> <hasName rdf:datatype="string">Amphore_A15</> … </owl:NamedIndividual> <owl:NamedIndividual rdf:about="Transfo-1953670366"> <rdf:type rdf:resource="Transformation3D"/> <hasRotationMatrix rdf:resource="Mat1220813917"/> <hasTranslation rdf:resource="IPoint3D1039759545"/> <hasScale rdf:datatype="xsd:double">0.9488028787930947</> </owl:NamedIndividual> Amphora_A15 description: Xlendi.owl
- 28. <owl:NamedIndividual rdf:about="#Amphorae2004628751"> <rdf:type rdf:resource="#Amphorae"/> <hasBoundingBox rdf:resource=“#BoundingBox2030458766"/> <hasTransformation3D rdf:resource=“#Transfo-1953670366/> <hasBellyDiameter rdf:datatype="double">0.3512</hasBellyDiameter> <hasDiameterNeck rdf:datatype="double">0.0785</hasDiameterNeck> <hasName rdf:datatype="string">Amphore_A15</hasName> … </owl:NamedIndividual> <owl:NamedIndividual rdf:about="Transfo-1953670366"> <rdf:type rdf:resource="Transformation3D"/> <hasRotationMatrix rdf:resource="Mat1220813917"/> <hasTranslation rdf:resource="IPoint3D1039759545"/> <hasScale rdf:datatype="xsd:double">0.9488028787930947</> ... </owl:NamedIndividual> Amphora_A15 description: Xlendi.owl <owl:NamedIndividual rdf:about="Mat1220813917"> <rdf:type rdf:resource="RotationMatrix"/> <has_m00 rdf:datatype="xsd:double">0.026584480435393985</> … </owl:NamedIndividual>
- 29. <owl:NamedIndividual rdf:about="#Amphorae2004628751"> <rdf:type rdf:resource="#Amphorae"/> <hasBoundingBox rdf:resource=“#BoundingBox2030458766"/> <hasTransformation3D rdf:resource=“#Transfo-1953670366/> <hasBellyDiameter rdf:datatype="double">0.3512</hasBellyDiameter> <hasDiameterNeck rdf:datatype="double">0.0785</hasDiameterNeck> <hasName rdf:datatype="string">Amphore_A15</hasName> … </owl:NamedIndividual> <owl:NamedIndividual rdf:about="Transfo-1953670366"> <rdf:type rdf:resource="Transformation3D"/> <hasRotationMatrix rdf:resource="Mat1220813917"/> <hasTranslation rdf:resource="IPoint3D1039759545"/> <hasScale rdf:datatype="xsd:double">0.9488028787930947</hasScale> </owl:NamedIndividual> Amphora_A15 description: Xlendi.owl <owl:NamedIndividual rdf:about="IPoint3D1039759545"> <rdf:type rdf:resource="IPoint3D"/> <hasName rdf:datatype="xsd:string">pt</> <hasX rdf:datatype="xsd:double">2.48428</> … </owl:NamedIndividual>
- 30. Arpenteur GUIGUI Tool for Clustering4
- 31. Arpenteur GUIGUI Tool for Clustering4
- 32. GUI Tool for Clustering4
- 33. cluster num 0 nb item4 Data analysis on 4.0 elements Average 0.5110184066310204 Median 0.5109524164523762 MAD 0.00148657127091667 RMS 0.5900761571720867 Max 0.5130573506564815 Min 0.5091114429628476 Index of Max elem 1.0 Index of Min elem 3.0 Data analysis on 7.0 elements Average 0.39443207822179266 Median 0.39451410787725383 MAD 0.0012772255320951942 RMS 0.4260424415650642 Max 0.39701713467045296 Min 0.390012183094278 Index of Max elem 0.0 Index of Min elem 6.0 nb Amphorae cluster 11 cluster 0 4 cluster 1 7 cluster 2 4 cluster 3 4 cluster 4 3 cluster 5 4 cluster 6 3 cluster 7 5 cluster 8 3 cluster 9 3 cluster 10 4 GUI Tool for Clustering4
- 37. Each setting is so unique that when you dismount the camera and plug it back, it is no longer the same ! <owl:NamedIndividual rdf:about="DigitalCamera1962398162"> <rdf:type rdf:resource="DigitalCamera"/> <hasDistortion rdf:resource="Distortion1263391254"/> <hasFocalLength>14.4146</> <hasFrameHeigthInPixel>2000</l> <hasFrameWidthInPixel>3008</l> <hasName>D100_LIS_14mm_H2O</> … </owl:NamedIndividual> SameAs ? SeeAlso ? (..) Data Linking Discussion5
- 38. ADS triplestore does not contain amphorae data, but rather a SKOS thesaurus of amphorae data types. http://archaeologydataservice.ac.uk http://data.archaeologydataservice.ac.uk/ Data Linking Discussion5
- 44. ▪ CH resources Model within Arpenteur ontology: ➢ typology, photogrammetry, spatial ➢ LOV: http://lov.okfn.org/dataset/lov/vocabs/arp ➢ Alignement: CRM, GEOSPARQL ▪ Xlendi LOD dataset: ➢ Datahub: https://old.datahub.io/dataset/xlendiamphorae Conclusion5 Gathering Segmentation Profiling Query Clustering ▪ Clustering ➢ GUI Tool ➢ SQWRLTab + SWRLTab
- 45. Future Directions and Open Issues ▪ Easy query GUI for archaeologists ▪ Xlendi Dataset need to be linked: ➢ Data source selection … ➢ Automatic linking tools … ➢ Type of links ? (seeAlso, exact, broad, narrow, … ) Conclusion5
- 46. Thank you for your attention ! ➢ Pierre Drap, et al. Underwater photogrammetry and object modeling: a case study of Xlendi Wreck in Malta. In Sensors (2015). ➢ Pierre Drap, et al. Ontology-Based Photogrammetric Survey in Underwater Archaeology. In ESWC 2017 ➢ Mohamed BEN ELLEFI, et al. Cultural Heritage Resources Profiling: Ontology- based Approach. In WWW ’18 Companion ➢ Jérôme Pasquet, et al. Amphora Detection Based on a Gradient Weighted Error in a Convolution Neuronal Network. In IMEKO (2017). • The ontology: http://www.arpenteur.org/ • The Imareculture project: http://imareculture.weebly.com/