Less is More Lightweight Ontologies and User Interfaces for Smart Labs J. G. Frey, G. V. Hughes, H. R. Mills, m. c. schraefel, G. M. Smith, David De Roure CombeChem Project Electronics and Computer Science / School of Chemistry University of Southampton
Combechem Aims Support end-to-end collection & sharing of data and metadata for chemistry experiments. Accessible (online) publication of results Make data and records available Accessible to humans and computers
The Chemistry Experiment Plan Perform Ponder Publish
Planner Experiment planning tool: Basic metadata (short description, etc.) Ingredients Planned quantities Processes Based on COSHH form Simple web application in PHP
COSHH
critical data entry
A digital lab book replacement that chemists were able to use, and liked.
Architecture Viewing Tools Sem. Web Apps Services Results Data Semantic Data Compute services Weights & Measures Bench Planner RDF over SOAP
getRecord()
getObservation()
Design discussion Handling RDF both ends – tricky Structure to data: good: clients can add what they want Ontology extensions: Uncertainty to measurements bad: clients can add what they want Is the structure youre given navigable in the way you expect?
Design Discussion We have an ontology (shared understanding) But... experiment structure is a higher- level entity Must be created and maintained by good programming, not simply by adherence to rules of the ontology
Current & Future Work Generating an experiment report for publication Natural language generation considered, but rejected by chemists Standard language for papers is dense, cryptic, and frequently unhelpful
Culture change Different methods of publication On-line Electronic data sent to print journals for peer review and publication Different form for procedure description
Summary System to support end-to-end capture of experimental work in chemistry Light-weight in-lab systems Desktop
Credits & More Information Jeremy Frey, David De Roure, Gareth Hughes, Hugo Mills, monica schraefel, Graham Smith.
Process record notation Analyse a real experiment What information do chemists record? What should they record? What do they want to record? How does this differ from their experiment plan? Evolved our own graph of plan and record
Information collection Process record Provenance record Measurements Processes Annotations Service invocations Secure time-stamps etc… Increasing detail
Data model Process record Provenance record Measurements Processes Annotations Service invocations Secure time-stamps etc… Increasing detail Plan Intended actions: guide to chemist, or [later] workflow