approaches Dave Reynolds, Epimorphics Ltd @der42 Validation - PowerPoint PPT Presentation

Validation: requirements and approaches Dave Reynolds, Epimorphics Ltd @der42

Validation requirements based on experiences with data.gov.uk Linked Data  Most current Linked Data in data.gov.uk is:  described using a range of vocabularies and documentation  validated , if at all, by publisher using internal/ad hoc tooling  Emerging requirement for shared validation approach:  to enable interoperability  so publishers know the shape of data required  publishing tools can e.g. auto-populate forms  consuming tools know what to expect  Key requirements:  declarative – easily inspectable by tools  declared – can locate the structure definition for a data set  accessible to mortals

A spread of requirements  regular data  statistics, financial, environmental measurements, ...  irregular data  organizational structure, strategic plans, ...  controlled terms  code lists, regulated entities, geographic regions, ...

Regular data  use Data Cube vocabulary  http://www.w3.org/TR/vocab-data-cube/  meets the requirements:  declarative specification of structure - Data Structure Definition (DSD)  declared: all observations link to DataSet link to DSD  fairly understandable: :complianceDsd a qb:DataStructureDefinition; rdfs:label "complianceDsd"@en; qb:component [qb:dimension :bathingWater], [qb:dimension :samplingPoint], [qb:dimension :sampleYear], [qb:measure :complianceClassification], [qb:attribute :inYearDetail]; qb:sliceKey :complianceByYearKey, :complianceBySamplingPointKey .

But how to validate a data cube?  Specification now defines “well - formed” cubes  closed world notion of compliance with DSD  integrity constraints specified by a set of SPARQL queries  Lessons:  SPARQL was sufficient to express all the required ICs  some of the queries are convoluted and non-obvious  at least one is quadratically slow unless optimizer is magic  Useful compromise  SPARQL doesn’t meet requirements of inspectable and understandable  but tools and humans can operate at the DSD level

Irregular data  typically mix-and-match range of vocabularies  declare usage via void:vocabulary  target users find OWL impenetrable  requirement for “vocabulary profiles”  closed-world constraints on properties (cardinalities, ranges)  expressivity of closed-world OWL would be sufficient  but need a presentation layer to simplify authoring and consumption – OSLC resource shapes?  discovery mechanism

Controlled terms  the other 80% of the problem  common resource shapes the easy part  interoperability means re-using terms for things in the domain  sets of controlled terms (URI sets, code lists etc)  can be very large  often managed by third parties independent of data publisher and vocabulary definer  can be dynamic  typically handled by some form of registry  governed, closed-world, lists of approved terms at point in time  implication  need ability to validate against external services such as registries

approaches Dave Reynolds, Epimorphics Ltd @der42 Validation - PowerPoint PPT Presentation

Validation: requirements and approaches Dave Reynolds, Epimorphics Ltd @der42 Validation requirements based on experiences with data.gov.uk Linked Data Most current Linked Data in data.gov.uk is: described using a range of vocabularies

God of War or God of Peace? Question Question Various approaches Question Various approaches

God of War or God of Peace? Question Question Various approaches Question Various approaches

Identifying DRR Measures through participatory approaches approaches Karina Barquet, Stockholm

700MHz 700MHz awards awards and approaches in and approaches in the r the region egion 2nd

God of Peace? Question Question Various approaches Question Various approaches Suggestions

Modeling and high-throughput approaches in developmental patterning approaches in developmental

New MTT - Approaches 20 Marine Raker Pile Installation New MTT - Approaches Excavation

Various Approaches Various Approaches acoustic classic The Prosody The Prosody measurement

Learning Approaches to Estimate Depth from RGB Lecture 5 What will we learn - Latest Approaches

Approaches in modelling tritium uptake by crops EMRAS II Approaches for Assessing Emergency

End User Development: Approaches Towards A End User Development: Approaches Towards A Flexible

CONTENTS Motivation Approaches in the Skyrme model Conventional approaches / short

Heuristic Approaches Mark Voorhies 5/5/2017 Mark Voorhies Heuristic Approaches PAM (Dayhoff)

Novel Approaches to Further Antibacterial Drug Development: New Approaches to the Clinical

Solution approaches for Solution approaches for address-selection problems address-selection

Adversarial Approaches to Bayesian Learning and Bayesian Approaches to Adversarial Robustness

AIP Plexos Market Simulation Model Validation Project Workshop 1 Proposed Model Review

Chapter 2 Primitive Data Types and Operations Basic computer skills such as using Windows,

Data types and printf Eric McCreath Learning C It is difficult to present material on C

Drug Trials Snapshots and Transparency: Opportunities and Challenges Naomi Lowy, MD Lead

Off-line Data Vali lidation for Water Network Modeling Studies M. Quiones* G, C. Verde** and

Privacy-Preserving Computation with Trusted Computing via Scramble-then-Compute Hung Dang, Anh

HOW TO SET UP A COMMUNITY-BASED ENVIRONMENTAL HEALTH SENSING PROJECT Dawn Nafus @dawnnafus

1 Automating Machine Learning and Deep Learning Workflows 2 Information Name: Mourad