1. Fourteenth International Conference on Grey Literature - Rome, 2012 Alessia Bardi, Sandro La Bruzzo, Paolo Manghi name.surname@isti.cnr.it http://nemis.isti.cnr.it/groups/infrascience CUSTOMIZED OAI-ORE AND OAI-PMH EXPORTS OF COMPOUND OBJECTS FOR FEDORA REPOSITORIES

2. SCENARIO AND MOTIVATIONS Digital Library Systems (DLSs) need to export content Export of compound objects: packages of information objects with an identity OAI protocols (typical solution) DL management systems (DLMSs) need to support export protocols DLMS issues: Absence of OAI-exports: some DLMS do not provide support for OAI protocols (Relational Databases) Pre-defined OAI exports of compound objects: shape of objects as exported by OAI protocols is pre-defined by the DLMS (Fedora) Fourteenth International Conference on Grey Literature - Rome, 2012

3. OBJECT EXCHANGE PROTOCOLS OAI-PMHOAI-ORE Exchange subjectMetadataResource Graph-based data model NoYes Data ModelSet Item MetadataPrefix Aggregation Aggregated Resource Proxy Protocol6 OAI Verbs on HTTP HTTP Data SerializationXMLRDF (XML, ATOM) Deleted records management YesNo Fourteenth International Conference on Grey Literature - Rome, 2012

4. FEDORA: FLEXIBLE EXTENSIBLE DIGITAL OBJECT REPOSITORY ARCHITECTURE A Fedora instance manages graphs of Data Objects Data Objects conform to Content Models Content Models define: Type and name of mandatory datastreams (files) Allowed relationships to other Data Objects Fourteenth International Conference on Grey Literature - Rome, 2012 Graphs of Content Models Graphs of Data Objects Compliant to Digital Library document model implementation FEDORA

5. FEDORA OBJECTS AND COMPOUND OBJECTS Fourteenth International Conference on Grey Literature - Rome, 2012 CM_article ds: ART, PDF ds: DC, XML CM_data ds: DATA, binary ds: DC, XML ds: DDI, XML USED_BY GENERATED_BY GENERATES USES hasModel Document model: implementation with Fedora Content Models Fedora instance: graphs of Fedora Data Objects Art2 ds:DC ds:ART Data 1 ds:DC ds:DATA USED_BY Art1 ds:ART ds:DC GENERATES GENERATED_BY USES ds:DDI hasModel Compound Object 1 Compound Object 2

6. FEDORA OBJECTS AND COMPOUND OBJECTS Fourteenth International Conference on Grey Literature - Rome, 2012 CM_article ds: ART, PDF ds: DC, XML CM_data ds: DATA, binary ds: DC, XML ds: DDI, XML USED_BY GENERATED_BY GENERATES USES Document model: implementation with Fedora Content Models hasModel Fedora instance: graphs of Fedora Data Objects Art2 ds:DC ds:ART Data 1 ds:DC ds:DATA USED_BY Art1 ds:ART ds:DC GENERATES GENERATED_BY USES ds:DDI hasModel Compound Object 1 Compound Object 2

7. Basic OAI- PMH Provider OAI Provider PMH-Set Each Fedora Data Object is annotated with relationships to the Sets it belongs to PMH-ItemFedora Data Object PMH Metadata Records Datastream PMH Metadata FormatDublin CoreAny format in existing datastreams Compound Object Boundaries Fedora Data Object EXISTING OAI-PMH SOLUTIONS Fourteenth International Conference on Grey Literature - Rome, 2012

8. EXISTING OAI-ORE SOLUTIONS Fourteenth International Conference on Grey Literature - Rome, 2012 OREProviderFedora2ORE ORE AggregationDefined by annotationSub graph visited starting from a given Fedora Data Object ORE Aggregated Resources Datastreams with a given name (e.g., DC) Fedora Data Objects in the sub graph ORE Proxynot supported Relationships between Aggregated Resources NO Compound Object Boundaries Fedora Data Objects annotated with the same tag Navigation depth

9. EXISTING OAI-PMH SOLUTIONS OAI-PMH Set: defined by annotations in each Fedora Data Object OAI-PMH Item granularity: Fedora Data Object OAI-PMH Metadata records: XML Datastreams Fourteenth International Conference on Grey Literature - Rome, 2012 USES Art 2 ds:DC ds:ART Data 1 ds:DC ds:DATA USED_BY Ar t1 ds:ART ds:DC GENERATES GENERATED_BY ds:DDI OAI-PMH Set Item Art1 DC Item Art2 DC Item Data1 DC DDI

10. EXISTING OAI-ORE SOLUTIONS: OREPROVIDER Fourteenth International Conference on Grey Literature - Rome, 2012 USES Art 2 ds:DC ds:ART Data 1 ds:DC ds:DAT A USED_BY Art 1 ds:ART ds:DC GENERATES GENERATED_BY ds:DDI http:// / objects/Art1/datastrea ms/DC/content Aggr http:// /o bjects/Art2/datastreams /DC/content http:// /o bjects/Data1/datastrea ms/DC/content Start FDO: Art1, navigation depth: 2

11. EXISTING OAI-ORE SOLUTIONS: FEDORA2ORE OAI-ORE Aggregation: graph visited from a given start Fedora Data Object OAI-ORE Aggregated resources: XML representation of visited Fedora Data Objects OAI-ORE Proxy: not supported  no semantic relationships Fourteenth International Conference on Grey Literature - Rome, 2012 USES Art 2 ds:DC ds:ART Data 1 ds:DC ds:DAT A USED_BY Art 1 ds:ART ds:DC GENERATES GENERATED_BY ds:DDI http:// /o bjects/Data1 Aggr http:// /o bjects/Art1 Start FDO: Art1, navigation depth: 1

12. exporting compound objects in fedora – solutions and limits Current solutions for OAI-PMH exports on Fedora......export OAI-items with the granularity of a Fedora Data Object...support the export of XML contained in datastreams Current solutions for OAI-ORE exports on Fedora… …exclude relationships and their semantics... allow limited customization of the structure of the compound objects (e.g., navigation depth) …pre-define the structure of the exported objects OAIzer works on any Fedora instance with Content Models: ORE exports include relationships and their semantics Structure of exported objects is fully customizable OAI-PMH metadata formats can be generated on request (requires the configuration of an appropriate transformation rule) Fourteenth International Conference on Grey Literature - Rome, 2012

13. EXPORTING COMPOUND OBJECTS: THE OAIZER SOLUTION OAIzer is a generic software for the customization of compound objects in Digital Libraries Software layer over the Digital Library Management System ORE exports include relationships and their semantics PMH metadata formats can be generated on request DB-OAIzer: works on RDBMS; prototype in the OpenAIRE project, to be integrated with OpenAIRE+ (Open Access Infrastructure for Research in Europe) Fedora-OAIzer: for Digital Library Systems based on Fedora frameworks with Content Models Fourteenth International Conference on Grey Literature - Rome, 2012

14. FEDORA-OAIZER: HIGH-LEVEL ARCHITECTURE Fourteenth International Conference on Grey Literature - Rome, 2012 Graphs of Content Models Graphs of Data Objects Compliant to DLS document model implementation FEDORA OAIzer Entity Graph OAI view represents View Interpreter OAI-ORE exporter OAI-PMH publisher

15. PHASE 1: ENTITY GRAPH Entity Graph: Graph representation of the document model Created exploiting information about Content Models Graph nodes represent Content Models Properties of a node represent the datastreams defined by the Content Model Edges between nodes represent the relationships defined by the Content Models Fourteenth International Conference on Grey Literature - Rome, 2012 ds:DC ds:ART name: CM_article ds:DC ds:DATA GENERATES USED_BY GENERATED_B Y name: CM_data USES ds:DDI CM_article ds: ART, PDF ds: DC, XML CM_data ds: DATA, bin ds: DC, XML ds: DDI, XML USED_BY GENERATED_BY GENERATES USES

16. PHASE 2: DEFINITION OF THE OAI VIEW Fourteenth International Conference on Grey Literature - Rome, 2012 OAI View: Tree representation of the structure of the compound objects to export Sub graph of the entity graph with entities and relationships to include in the compound object Ex. The DATA datastream of CM_data is excluded The root node is the entry point of the OAI view ds:DC name: CM_article ds:DC USES name: CM_data ds:DDI name: CM_article ds:DC GENERATED_BY

17. Fourteenth International Conference on Grey Literature - Rome, 2012 PHASE 3.1: INTERPRETATION OF THE OAI VIEW FOR OAI-PMH ds:DC ID: Art2 ds:DC USES ID: Data1 ds:DDI Set View Data Transformation Rule (XSLT): OaizerXML -> MF1 Item Instance_View_2.... OAIZER-XML Xml Serialization of the Instance View Item Instance_View_1 DC OAIZER-XML MF1 OAI-PMH View Interpreter Instance View: created by navigating the graph of Data Objects according to the paths in the view DC OAIZER-XML MF1

18. PHASE 3.2: INTERPRETATION OF THE OAI VIEW FOR OAI-ORE Fourteenth International Conference on Grey Literature - Rome, 2012 ds:DC Art2 Data 1 ds:DC USES Art1 GENERATED_BY ds:DDIds:DC GENERATED_BY http:// /oaizer/ Art1/OAIZER-XML USES proxyFor proxyIn proxyFor proxyIn Aggr proxyFor proxyIn Legend: Aggregation Aggregated resource Proxy ore:aggregates relation OAI-ORE View Interpreter http:// /oaizer/ Data1/OAIZER-XML http:// /oaizer/ Art2/OAIZER-XML

19. FEDORA-OAIZER Fourteenth International Conference on Grey Literature - Rome, 2012 PMH ItemPMH Metadata Records PMH Metadata Format OAIzerOAI View InstanceGenerated from OAI View Instance DC, OAIzerXML, any other format require appropriate XSLT ORE AggregationORE Aggregated Resources ORE Proxy OAI View InstanceFedora Data Objects in the OAI view instance Relationships between Aggregated Resources Compound Object Boundaries YESOAI View (property and navigational criteria)

20. CONCLUSIONS Don’t adapt your document model for the exports, adapt the exports to your document model! Customized, domain-dependent exports of compound objects Future work: Better graphical support for the definition of views in case of entity graphs with cycles Implementations for different back-ends Integration of DB-OAIzer with Fourteenth International Conference on Grey Literature - Rome, 2012