Biodiversity and Ecosystem Informatics: Research, Technology - PowerPoint PPT Presentation

Biodiversity and Ecosystem Informatics: Research, Technology Transfer, or Application Development? Jessie Kennedy http://www.soc.napier.ac.uk/jessie 02/12/2002 VLDB 2002 1

An ecological question… ➤ What is the effect of change in (ozone) on the distribution of ( Bellis perennis ) in (temperate grasslands) in (Europe)? ➤ Answer requires (amongst other problems) integration of many BIG assumption………. different databases Content area responsibilities of GBIF Sequence Data (GenBank, RNA, protein, etc.) Biological Catalog of Catalog of CHM Specimen Names of Names of Data Known Known Geospatial Organisms Organisms Data Climate Data Search Ecosystems Engines Data Access/Inter- operability Ecological Data Courtesy of Global Biodiversity Information Facility - http://www.gbif.org

Biological taxonomy ➤ How do we catalogue all of the species of plants on Earth? Conversion to electronic ➤ Plant Taxonomy (classification) media isn’t really a DB problem ➤ Data -> Real specimens - is it? ➤ Stored in herbaria, museums…. ➤ Recorded in notebooks, books, journals ➤ Masses of information inaccessible NHM London

Specimen - Description ➤ Taxonomic characters Fruit of Torilis japonica ➤ annual, leaves hairy, lanceolate, mostly white flowers….. ➤ No agreed terminology ➤ plant structure ➤ Attributes ➤ Values ➤ Ontology problem ➤ DB research? ➤ DB support ➤ definitions ➤ exemplars

Classifying and naming plants ➤ Specimens classified into taxa then named by set of rules ➤ Revisions are common There is over 250 years Retrieve info on legacy data ➤ Taxa or specimens may appear in many classifications simultaneously G albiflora ? contributing to this problem... genus Globba Globba Are these two taxa (concepts) the same? What about this specimen – what Ceratanthera Marantella Globba section Cerantera species is it? G pendula G calophylla G siamensis G albiflora G siamensis species G pendula

Multiple Overlapping Classifications

What are the DB issues? ➤ DBMSs don’t provide sufficient semantic mechanisms to support required application functionality……. ➤ Orthogonality of classification and data ➤ Objects are not designed to be classified ➤ Support for trees/graphs ➤ Multiple overlapping classifications -> a directed acyclic graph ➤ Nodes (taxa or specimens) are complex objects ➤ Levels (ranks) contain information ➤ Each classification is independent from all the others ➤ Support for traceability ➤ Rationale for classification is important ➤ Support domain specific rules ➤ data derivation ➤ constraints

What have I learned…. ➤ Need to understand the domain problems to provide good solutions. ➤ Accurately representing data from observations or experiments is vital if data is of use in future ➤ Database technology plays an important role in ensuring this ➤ Data modelling research and incorporation into DB systems ➤ Lots of semantic models – but still not in DBs in usable/efficient form ➤ Query languages suitable for end-users with complex queries ➤ Data visualisation tools ➤ query as fast as I would brushing over a visualisation ➤ Ontology problem ➤ how can we get the meaning in to the DB in a manageable way. ➤ Data Provenance ➤ annotations, workflow ➤ Core DB ➤ all these extra semantics means even better performance needed.

SEEK Science Environment for Ecological Knowledge (SEEK) Analysis and Modeling Algorithm Execution Layer PM 1 A 1 PM n A n Output System Domain Mediation Semantic Map Integrated Analysis View Semantic Mediation Layer Engine Mediated Data View Metacat Species Analyst SRB U Information Access D Layer CM 1 CM 2 CM n D S 1 M 1 S 2 M 2 S n I M n This incorporates the taxonomy 5 year NSF funded project problem

Questions ➤ Is there original DB research in biodiversity / ecological informatics? ➤ Yes - Many of the same general problems but with challenging difficulties ➤ Do they need off the shelf applications ➤ Yes - but they’re not there in any usable form ➤ Organizational infrastructure for supporting data re-use ➤ Yes - vital they re-use concepts accurately ➤ Training for ecologists in using systems? ➤ Yes - but not inappropriate ones that waste their time... ➤ More ecologists doing domain research? ➤ Yes - but with tools to help them do the job more efficiently and accurately…..