A Practical Approach to Metadata Management Mark Jessop Prof. Jim Austin University of York.

A Practical Approach to Metadata Management

Mark JessopProf. Jim AustinUniversity of York

Introduction

• Follow on from DAME/Broaden (York)• Neuroscience Domain• Aims to promote collaboration and enable sharing of

resources: Data, Analysis• Online Portal to store Data, Analysis Code, Experimental

protocols.• Current deployment: Upload and annotate experimental

data and apply security, search across data, analyse with services, share and download.

• Working on workflows and a common data format.

e - Neuroscience

• Neuroscience at start of e-science journey.

• Proprietary data formats.• Private Local data.• Primitive data annotation

(Metadata).• Desktop tools.

x

x

x

xx

A

B

A: Neuro-Scientists

B: Computer-Scientists

Metadata

• Metadata is essential to allow resource discovery, interpretation, evaluation and re-use.

• Resources (Data, Services, Workflows) must be adequately annotated.

• Metadata needs to searchable.• Metadata needs to be viewable!

Managing Metadata

• Comprehensive Metadata definition provided for each resource type.

• Electrophysiology data described by the MINI.– Based on FuGE object model (Materials, Data,

Protocols).• Services described by MIAOWS.

Metadata Systems

• Symba data archive tool integrated into CARMEN.– Implements FuGE-OM.– Early in development program.

• User trials revealed problems.– Form complexity.– Response times.

Metadata Systems Take 2

• Revisit Requirements– Capture Mini Schema.– Must be responsive.– Must be simple to use.– Must be able to re-use metadata.

• System built based on forms.• Generic metadata storage.• Search across schemas.

Implementation

• Form/Schema designer generates XML description of forms and schema.

• XML schema stored in a database.• Portal requests XML which is used to generate input

forms or to display existing metadata.

Were we Successful?

• Some users complete the full document.• Some users complete enough to facilitate searching.• Templates are widely used.• Many users re-use complete templates resulting in

repeated metadata.

Metadata for Services

• Combine WSDL and user metadata to describe services.• Generate invocation forms from metadata.

– Default parameters.– Validation.

• Search on function.• Metadata applied to Service output.

– Provenance.– Re-use.

Future Work

• Ontologies.• Organisation of metadata – prevent repeats.• Smarter templates.• Ways to encourage completion of full metadata.

CARMEN Consortium

www.carmen.org.uk

Demos on White Rose Grid Stand 15:00.