Slide 1

KOS: A new paradigm for discovery and access to complex ecological data David Turner, Paul Chinnick, Andrew Graham, Matt Schneider, Craig Walker Logos used with consent. Content of this presentation except logos is released under TERN Attribution Licence Data Licence v1.0

Slide 2

Goal: To enable the preservation and appropriate re-use of rich ecological information www.nswrail.net Fragmentation: Many different ways of measuring/observing/ expressing similar ecological concepts * Rapidly evolving Dispersal: Data is stored in many storage locations and formats Diversity: Ecological Data covers a wide range of topics www.derm.qld.gov.au Alamay Source:Forestcheck: www.dec.wa.gov.au Complexity: Data usually needs explanation and context before it can be accurately used

Slide 3

Needs and expectations of users An interface to allow for: Data discovery Immediate or facilitated data access Comprehension A common information platform Easy manipulation Familiar tools and or intuitive interface Cater for individuality Each user will have different data requirements

Slide 4

The Information Landscape

Slide 5

Preservation?

Slide 6

The ecological data delivery paradox IT solutions require rigid models to process data hence tend to force things into narrow structure BUT flexible models are required to preserve the full richness of data and context Software is not intelligent and requires fixed structures and fixed rules (logic) Understanding data involves many subtleties which are complex to model Models and processing dependent on the question

Slide 7

Towards the new paradigm Accept that flexible data needs flexible information models A sparse semi-structured graph oriented knowledge model (not a data model) Model must be extensible (ontology based) Create descriptive artifacts for consumers Provide the understanding of context (express subtleties) and avoid the significant complexity of modelling By capturing the full richness, we enable informed re-use for the broadest range of questions The user can assess suitability for their purpose and can obtain complete information

Slide 8

KOS Operational Model Data Source Data Source KOS Repository KOS Repository KOS Portal Common Information Model (Ontology) Contextual Description ETL Script KOS DSL Engine Data Portal Traits (tags from controlled vocabulary) Data under Common Model Data Custodian Subject Expertise Integration via a common structure Enrichment Search Index Semantics Data Ingestion Graph visualisation Description products

Slide 9

The user experience Key features Access and licensing Build complex queries From simple blocks Site context

Slide 10

The information model Select my site Establish it, and make some observations Visit 1 Etc... Make more observations Visit 3 Make more observations Visit 2

Slide 11

The user experience Record instances e.g. visits Observation instances Observation instance details Search level controls granularity

Slide 12

Concept alignment We integrate data from many different sources, using different classification systems All datasets indexed with common set of search terms (vocabularies) Taxonomy is a good example Many other themes for search from project metadata to observation level data

Slide 13

Descriptive artefacts Context represents a communication challenge not a modelling problem Standard structure Bite sized chunks Direct linkage between data and description Authored synthesis based on information derived from multiple sources (published and unpublished) Image of descriptive doc Once available

Slide 14

Current status

Slide 15

Live system demo... Members Dining Room 3 15:30 17:10