Tutorial at OAI5 (cern.ch/oai5). Abstract: This tutorial will provide a practical overview of current practices in modelling complex or compound digital objects. It will examine some of the key scenarios around creating complex objects and will explore a number of approaches to packaging and transport. Taking research papers, or scholarly works, as an example, the tutorial will explore the different ways in which these, and their descriptive metadata, can be treated as complex objects. Relevant application profiles and metadata formats will be introduced and compared, such as Dublin Core, in particular the DCMI Abstract Model, and MODS, alongside content packaging standards, such as METS MPEG 21 DIDL and IMS CP. Finally, we will consider some future issues and activities that are seeking to address these. The tutorial will be of interest to librarians and technical staff with an interest in metadata or complex objects, their creation, management and re-use.
1 of 48
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
More Related Content
Object models and object representation
1. CERN workshop on Innovations in Scholarly Communication (OAI5) 18-20 April 2007 Tutorial 4 - Object models and object representation Julie Allinson and Mahendra Mahey Repositories Research Officers, UKOLN, University of Bath with Herbert van de Sompel Los Alamos National Laboratory A centre of expertise in digital information management www.ukoln.ac.uk
2. Order of play 09.00: Overview and introduction to the scenario (Julie Allinson) 09.10: Exploring options for metadata modelling (JA) 09.45: Content packaging for complex objects (Mahendra Mahey and Herbert van de Sompel) 10.30: Break 10.50 Content packaging for complex objects continued (MM) 11.10: Concluding remarks and questions 11.30: Finish
4. Unpacking the tutorial title Object models and object representation We inherited this title! Focus is on metadata , simple, compound or complex digital objects and content packaging for interoperability across scholarly communications, with specific emphasis on ( institutional ) repositories
5. Looking at … What are we talking about? Dublin Core, MODS, DIDL, IMS CP, METS and more Why do we need to know about it? Where and when is it used? Who needs to know about it? How do we use it?
6. What is a digital object? Digital objects are anything that might be stored by a digital repository … can be any media or semantic type, e.g. an image, an article, XML metadata record, PDF etc. have (unique) identifier(s) to be considered Resources as per the W3C web architecture they must be identified by a URI convey information, digitally, i.e. they are not abstract concepts, or physical objects these are things that metadata is also used to describe or representations of these could also be called information objects, or information resources (in W3C speak) To be useful for scholarly communication they should have associated metadata metadata can be a digital object in its own right
7. Compound and complex digital objects Aggregations of related digital objects gathered together to form a logical whole. The relationship may be purely structural (e.g. a book and its chapters) Complexity is added when we begin to think beyond the structural, to a richer set of relationships between digital objects and relationships with other kinds of resources (people, organisations, concepts, events etc.) Metadata and/or content packaging help us to express structure and relationships
8. Examples – digital objects A PDF scholarly paper A JPEG image A scientific dataset Each uniquely identified by a URI and accessed from a repository or web server Simple digital objects?
9. Examples – compound objects A book and its chapters XML-encoded chapters and table of contents metadata describing each chapter content packaging wrapper enclosing all of the above An image in different resolutions RAW image file JPEG print size thumbnail
10. Towards a repository ecology – compound objects complex-ified A scholarly paper, with different versions, metadata, metadata describing the agents (author, publisher etc.), references to supplementary materials etc. An issue of an overlay journal built from distributed papers An eScience publication combining text and primary research data, simulations, statistical analysis etc. Examples can become more and more complex, if we want them to be The repository ecology – a way of examining how systems and services interact to support scholarly communication digital objects are flowing around this ecology
11. Scenario for the tutorial A single scenario to capture some of the issues raised here: A conference paper with different versions calling on additional, external, resources re-used in other compound resources See handout
12. Some areas for consideration Boundaries what is and isn’t part of a particular compound digital object what are the relationships within the object and beyond its boundary Context and use This might dictate particular requirements e.g. a preservation service needs access to the full datastream and a specific set of preservation metadata a repository may want only the descriptive metadata and a reference Expertise and local requirements This might dictate the choice of standard used
14. Overview basic metadata semantics a metadata framework for interoperability syntax vocabularies application profiles and application models metadata for compound / complex objects the scenario
15. Metadata : what? Data about data? … this isn’t very helpful “ Metadata consists of statements we make about resources to help us find, identify, use, manage, evaluate, and preserve them”. (Marty Kurth, tutorial on DC Semantics, 2006 http://dc2006. ucol . mx /program. htm )
16. Why? Without metadata, data is useless ‘ orange’ – a fruit, a company, a colour, a password, an identifier, arbitrary text string? ‘ 01234567890’ – a telephone number, an identifier? non-text resources e.g. an image Functions, a selection resource discovery (oai-pmh, rss, z39.50) identifying and differentiating resources contextual information authenticating and evaluating sharing information geographic locations metadata: Joan Miro ‘Chicago’, detail describes
17. When? Where? Who? Metadata is everywhere and is used all the time in business, in education, in libraries, on the Internet … for local purposes and for wider interoperability there are many different ‘types’ of metadata descriptive metadata rights metadata administrative metadata etc.
18. How? metadata semantics … metadata describes resources these resources can be digital , physical or abstract things as a general principle metadata describes one, and only one, resource (the 1:1 rule) metadata descriptions contain statements about the resource a statement consists of a metadata property (aka, an element ) and a value ( a property/value pair )
19. Metadata semantics diagram this is very basic! … over-simplified from the Dublin Core Abstract Model http:// dublincore .org/documents/abstract-model/ value property resource description 1 0..n 1..n 1 describes statement statement statement statement
20. Example, based on our scenario A conference paper [the resource] with the title [property] ‘Signed metadata : method and application’ [value] and the resource type [property] ‘Text’ or ‘ConferencePaper’ [value] MODS metadata: <titleInfo> <title>Signed metadata</title> <subTitle>method and application</subTitle> </titleInfo> <typeOfResource>Text</typeOfResource> Dublin Core in XHTML: <meta name="DC.title" content=“Signed metadata : method and application" /> <meta name="DC.type" content=“ConferencePaper" />
21. Metadata example diagram metadata value vocabularies provide standard vocabularies for expressing specific values metadata property vocabularies provide standard vocabularies of metadata properties value property 1 0..n 1..n a conference paper resource description 1 describes statement statement statement statement type or title ConferencePaper Text Signed metadata : method and application
22. Hang on, there’s more … Metadata vocabularies are only one piece of the jigsaw For exchange, we also need machine-readable Metadata Formats Application Profiles draw together properties from one or more namespaces, for a particular purpose, Abstract Models provide a model, or a set of rules for how descriptions are constructed (this may be embedded in the metadata standard itself) an abstract model can act as a mechanism for mapping between syntaxes together these give us the foundation for a metadata framework
23. A metadata framework? Abstract Model Metadata Vocabularies Application profiles Metadata formats properties values Application models XML, RDF/XML, XHTML UK LOM Core, CanCore, RSLP Schema, AGRIS, Scholarly Works AP, NDLTD (theses), DDF-MXD, DiVA DCMI Abstract Model Embedded data models conforms to DCMES, MODS, LOM, ISO 19115, e-GMS, PREMIS, FOAF DCMI Type, DDC, LCSH, TGN, MARC Relator Codes
24. Taking the next metadata step – from ‘flat’ to ‘modelled’ traditionally metadata has been seen a largely ‘flat’ set of metadata elements describing a single resource (e.g. a scholarly paper) but often the metadata is implicitly describing other resources (e.g. an author) but without explicitly recognising these as resources ‘ flat’ metadata cannot adequately describe multiple resources and the relationships between them nor can it group together descriptions of resources that are closely related
25. Some of the possible resources and relationships in our scenario publisher related materials version 1.0 electronic book image version 2.0 published proceedings pdf pdf 0..n is published by author author 0..n is created by 0..n see also 0..n has version 0..n is part of 0..n has part 0..n has format
26. Metadata is not flat! in Dublin Core the Dublin Core Abstract Model introduces support for ‘description sets’ it is for application profile developers to define the relationships they want to support in MODS the < modsCollection > wrapper element can be used to group < mods > descriptions < relatedItem > facilitates the capture of hasPart, isPartOf and seeAlso type relationships
27. Dublin Core as a case study The Dublin Core Abstract Model attempts to make explicit the model that underpins Dublin Core the DCAM starts from the central notion of a ‘description set’ a set of ‘descriptions’ about a group of related ‘resources’ where each description is about a single ‘resource’ (the 1:1 rule) and where each ‘description’ comprises property/value pair ‘statements’ ‘ description sets’ are instantiated as ‘records’ (e.g. using XML, RDF/XML or XHTML) for the purpose of exchanging information between networked systems Slide courtesy of Andy Powell, Eduserv Foundation http://www.slideshare.net/eduservfoundation/the-dublin-core-abstract-model-a-packaging-standard
28. Dublin Core Abstract Model summary record (encoded as HTML, XML or RDF/XML) description set description (about a resource (URI)) statement property (URI) value (URI) vocabulary encoding scheme (URI) value string language (e.g. en-GB) Slide courtesy of Andy Powell, Eduserv Foundation http://www.slideshare.net/eduservfoundation/the-dublin-core-abstract-model-a-packaging-standard syntax encoding scheme (URI)
29. Dublin Core Abstract Model contd. the DCAM is open about the relationships between resources described in a description set whole / part (book, chapter, section, page) physical / digital (painting / digitised image) object / human (document / author) conceptual / physical (work / item) the relationships between things must be articulated in an ‘application model’ and captured using the properties specified in an ‘application profile’ Slide courtesy of Andy Powell, Eduserv Foundation http://www.slideshare.net/eduservfoundation/the-dublin-core-abstract-model-a-packaging-standard
30. Scholarly Works Application Profile Also known as the Eprints Application Profile the application model says what things are being described the set of entities that we want to describe and the key relationships between those entities each entity and its relationships are described using an agreed set of properties the application profile describes these properties model vs. Model - the application model and the DCMI Abstract Model are completely separate the DCMI Abstract Model says what the descriptions ‘look’ like
31. FRBR for eprints FRBR (Functional Requirements for Bibliographic Records) provides the basis for our model it’s a model for the entities that bibliographic records describe but we’ve applied it to scholarly works FRBR is a useful model for scholarly works because it allows us to answer questions like: what is the URL of the most appropriate copy (a FRBR item) of the PDF format (a manifestation) of the pre-print version (a expression) for this eprint (the work)? are these two copies related? if so, how?
34. Example properties ScholarlyWork: title subject abstract affiliated institution identifier Agent: name type of agent date of birth mailbox homepage identifier Expression: title date available status version number language genre / type copyright holder bibliographic citation identifier Manifestation: format date modified Copy: date available access rights licence identifier
36. Multiple expressions, manifestations and copies scholarly work (work) version (expression) format (manifestation) copy (item) signed metadata doc institutional repository copy published proceedings no digital copy available (metadata only) e-book copy institutional repository copy pdf conference repository copy institutional repository copy pdf xml print copy pdf2 institutional repository copy Accepted MS Version 2.0 Author’s Submitted MS Version 1.0 Translation (Spanish)
37. Boundaries of this profile created to fulfil a specific set of requirements, chiefly richer metadata set & consistent metadata unambiguous method of identifying full-text(s) versions & most appropriate copy support identification of open access materials identification of the research funder and project code limited support for part/whole relationships and for related materials but it is modular and extensible and it fits well with the semantic web implementation with throw up new requirements
38. MODS overview The Metadata Object Description Schema (MODS) XML schema is intended to be able to carry selected data from existing MARC 21 records and to enable the creation of original resource description records includes a subset of MARC fields and uses language-based tags rather than numeric ones maintained by the Network Development and MARC Standards Office the Library of Congress with input from users a range of mappings and stylesheets are available
39. Considering the scenario in MODS <mods> records for each version <relatedItem> sub-element used within a <mods> description for see also, has part, is part of relationships published proceedings e-book copy conference repository copy Accepted MS Version 2.0 Author’s Submitted MS Version 1.0 Translation (Spanish) <modsCollection> can be used to wrap together several <mods> records
40. MODS observations MODS has more built-in richness than Dublin Core it is closely aligned with the library community and MARC, yet is simpler and more user friendly than MARC it has some support for creating collections, or sets and for describing related materials (within a single metadata description) but, it doesn’t have an ‘abstract model’ so is more difficult to map to other syntaxes and doesn’t support the extensibility and flexibility of application profiles and the DCAM and is less interoperable with semantic web approaches
41. Other approaches Danish DDF-MXD format created by DEFF to support the DDF - Danish Research Database – the national CRIS and the exchange of metadata Metadata model based on CERIF Common European Research Information Format designed to describe a much richer set of information captured by Research Information Systems
42. Entities described by DDF-MXD Research database with repository Persons (researchers and their competences) Organisations (universities, institutes, labs etc.) Projects (research and development projects) Events (conferences, workshops etc.) Documents (books, articles, eprints, slide shows, software, patents, data sets, simulations, learning objects, etc. etc.) Slide courtesy of Mogens Sandfaer, Tech nical Knowledge Center of Denmark
43. Data model Slide courtesy of Mogens Sandfaer, Tech nical Knowledge Center of Denmark An article in the journal by the author presented at the conference published by from the project
44. Thoughts about metadata Metadata is an essential element of the scholarly communications chain there are a number of existing property and value vocabularies application profiles can be developed for specific communities or purposes, using properties from existing vocabularies to facilitate interoperability, in a machine-to-machine context, metadata must be expressed in an encoding format/syntax such as XML adhering to an abstract model helps achieve understanding and agreement and provides a mechanism for mapping between syntaxes
45. Towards content packaging Metadata is not limited to describing flat, single-entity items Metadata models and application profiles can be used to describe complex/compound objects and can offer some degree of content packaging ‘by reference’ (i.e. by providing a URI) Content packaging standards are another mechanism for gathering together multiple metadata records alongside the digital objects they describe either ‘by reference’ or ‘by value’ (i.e. by embedding the object within the package)
47. Final thoughts … Interoperability is achievable But communities need to work together Standard metadata formats, application profiles, abstract models and content packaging standards really can help And they can also interoperate with each other For example, the RAMLET project <ieeeltsc.org/wg11CMI/ramlet/> Particularly if we agree things between us
48. Final, final thoughts Don’t underestimate local expertise Don’t forget that our world is in a constant state of flux The future will see scholarly communication happening in an increasingly seamless and joined up way For example the OAI-ORE project <www.openarchives.org/ore> Hopefully!