Presentation on theme: "The LOM RDF binding - update 2003-06-19 Mikael Nilsson The Knowledge Management Research Group Centre for user oriented IT design Royal."— Presentation transcript:
The LOM RDF binding - update Mikael Nilsson The Knowledge Management Research Group Centre for user oriented IT design Royal Institute of Technology, Stockholm
Overview Use cases of LOM/RDF XML binding compatibility issues Status of the binding Summary
Use cases 1: UR (Swedish educational broadcasting company) Largest content producer in Sweden (state owned) ~3000 TV/Radio programs since 2001 (30,000 since 1970), plus books, web sites, etc. +50 every week Online archive of digital versions since February: (in Swedish...)
UR: Metadata 4 application profiles (program, series, texts, websites) Several vocabularies: – DC (title, description, subject, relation, etc.) – DC Qualifiers (medium, MIME type, W3CDTF...) – LOM (difficulty, location, annotation, requirements..) – UR specific (accessibility, licenses, participants) – IMS content packaging
UR: Metadata refinements Refinements for e.g.: – identifiers (4-5 types in use in parallel at UR) – annotations (internal and external) – descriptions (different kinds for different purposes) – language (spoken/subtitles) Vocabularies for – audiences, types, media carriers, including refinements of DC & LOM vocabulary.
UR features Easily modifiable metadata set, and easily configurable editor. Without disturbing, we can: – add fields – refine fields and values – add vocabularies Supports cross-search with DC/DCQ repositories. Demonstrated yesterday at UR: – UR, OAI, Swed. Agency of Education, digital portfolios
Use case 2: Edutella P2P network for metadata exchange – Supports any RDF metadata Scenarios: – Search for nuclear physics material (DC, DCQ) – Search for learning material for primary school on nuclear physics (LOM) – Search for courses including a certain piece of content (IMS CP) – Search for annotations/opinions on this course.
Use case 3: RSS feeds Rich Site Summary 1.0 based on RDF and DC – Stephen Downes' RSS/LOM module: – – The RSS-LOM Module provides translation from IEEE-LOM to RSS 1.0 to allow learning object repositories to syndicate listings and descriptions of learning objects. – An RSS feed provided by a learning object repository is harvested by a metadata repository and aggregated with feeds from other learning object repositories. – These aggregated feeds are then made available as a searchable resource, the links provided ultimately pointing to the learning objects provided by the original learning object repositories.
RSS/LOM example (extract) Sample Learning Object en A sample learning object metadata file in RSS psychology Atomic 13 beta final National Research Council Rod Savoie Stephen Downes text/html Multi-OS NetscapeCommunicator 4.7 PT1H20M Active Exercise Low High Manager School 7-12 Easy PT1H20M This is intended to... en SomeCost SomeRestriction RightsBroker:RightsModel
RDF XML issues (for the BRG) Metadata term URIs – LOM/XML has them, LOM/RDF has them – How to sync LOM/XML – LOM/RDF wrt CORES? Vocabulary term URIs – Ability to give URI for a term. Needed in RDF – For LOM vocab: URIs for LOM vocabulary values – For local vocab: vocabulary XML format? LOM extensions are format specific – Need to make this clear in the LOM/XML spec
RDF binding status Most details resolved in the direction of simplification Ballot-ready in principle – Binding is not perfect, but given the contraints we have, it's as good as it gets Goal: Yes, you can use RDF and be LOM- compatible. Here's how to do it: This goal is achievable with the current binding See
Future Next-generation LOM visions: – allows for semantic extensions – plays nicer with other vocabularies – fits better into RDF world – in short: better knowledge representation capabilities The RDF binding work shows many of the problems and possibilities – will help inform LOM 2.0.
Issues for LOM 2.0 The issue of semantic (not structural) extension: – Subclasses, Subproperties, Element encodings This needs a meta-model for metadata. – DC has one: element, element refinement, element encoding). – LOM has one: element sub-element, etc RDF is a meta-model. DC is compatible, LOM is not. LOM's metamodel is insufficient for dealing with semantic extension As LOM is structured, not flat: Needs conceptual modeling more than DC. What are objects, what are relations?