Data Curation and
Biodiversity Research --
Lessons from BiSciCol and
a look at the “Triplifier
Simplifier”


John Deck, University of California, Berkeley
Brian Stucky, University of Colorado, Boulder
Lukasz Ziemba, University of Florida, Gaineseville
Nico Cellinese, University of Florida, Gainesville
Rob Guralnick, University of Colorado, Boulder
BiSciCol Team
Reed Beaman, Nico Cellinese, Jonathan Coddington, Neil Davies, John
   Deck, Rob
Guralnick, Bryan P. Heidorn, Chris Meyer, Tom Orrell, Rich Pyle, Kate
   Rachwal, Brian
Stucky, Rob Whitton, Lukasz Ziemba
•   BiSciCol is National Science Foundation funded 2010 – 2014
•   Infrastructure to tag & track specimens & derivates in cyberspace
•   Relies on globally unique identifiers (GUIDs) to track objects
•   Implements a Linked Data approach
•   Provides support for the Global Names Architecture
A Biological Relationship Graph …




                          Taxonomic Type Filter




                          Class Filter
                           X  Specimens
                               Tissues
                           X   Sequences
Why Linked Data? Why BiSciCol?
Here is Gustav’s Problem


                   Generates Lots of Data…




 (Prefers to collect stuff)
Biodiversity Data Challenges



   Data is Distributed


   Rapidly Changing
   Technologies

   Covers Multiple
   Domains
Solving Biodiversity Data Challenges with
BiSciCol and Linked Data

                                             Is a dwc:Event
  Group data into classes.
                                             Is a dwc:Event
   Assign identifiers.

   Link identifiers.

                         [ ] Ocean Sampling Day
   Publish.              [X] Moorea Biocode
                         [X] SI MSNGR System
                         [+] Add My Data
The Triplifier
                  (Advanced Interface)

Loading Data

Naming and Identifying Objects

Linking Objects

Publishing


                                 Powered by:
Advanced Interface: Loading Data
                             Darwin Core
                              Archive      Darwin
                                           Core
                                           Archive




           Spreadsheets



             Mysql




                          KEMU      MySQL
Advanced Interface: Entities


                                                                                                                                         78
                                                                                  Tissue




Result is identifiers assigned to Entities:
78 a door .
427 a cat .
<http://biocode.berkeley.edu/collectorspecimens/BMOO_2665> a <dwc:Occurrence> .
<http://biocode.berkeley.edu/collectorevents/MIB_25> a <dwc:Event> .




                                                                                   From Gary Larsen and adapted by Barry Smith in Referent Tracking
                                  Ceusters W, Smith B. Strategies                 for Referent Tracking in Electronic Health R Biomed
                                                                                      presentation at the Semantics of Biodiversity Workshop, 2012.
                                                                                                           Inform. 2006 Jun;39(3):362-78.
Advanced Interface: Entity
                                                  Relations




Relations as Triples:
<http://biocode.berkeley.edu/collectorevents/MIB_25> <ma:isSourceOf> <http://biocode.berkeley.edu/collectorspecimens/BMOO_2665> .
<http://biocode.berkeley.edu/collectorevents/MIB_37> <ma:isSourceOf> <http://biocode.berkeley.edu/collectorspecimens/BMOO_2667> .
<http://biocode.berkeley.edu/collectorspecimens/BMOO_2665> <ma:isSourceOf> <http://biocode.berkeley.edu/plate_well/Plate_M037F10> .
<http://biocode.berkeley.edu/collectorspecimens/BMOO_2667> <ma:isSourceOf> <http://biocode.berkeley.edu/plate_well/Plate_M028G5> .
Triplify!: View graph based data
            Response




    Query
The Triplifier (Simple Interface)




          Publish
What challenges are we facing now?
(for BiSciCol, Linked Data, and data integration
                   In general)
Identifier Issues
                                 Persistence
                                 Solutions:
                                 • DOIs (http://doi.org/)
                                 • EZIDs (http://ezid.net/)




                                 Assignment at the source is difficult
                                 Solutions:
                                 • Calculated namespaces (e.g. geo:lat,lng) via PDAs
                                 • UUIDs (randomly unique)

           The digestible RFID tag


                                     Semantic web requires URIs but many standards (including
scheme : string                      Darwin Core) do not require URIs for identifiers
                                     Solution:
                                     • Promote use of URIs for identifiers in all Standards.
     URI
“Occurrence”      Classification Issues
                                              Inadequate representational units
                                              Confusion between representational
                                              units



“Sample, Specimen, Individual, Aggregation”

                                               Solutions:
                                               • Continue working on clarity in term
                                                  definitions
                                               • Work from upper level ontologies (e.g.
                                                  Basic Formal Ontology) to derive
                                                  definitions.
Relation Issues
        Non-sensical conclusions are possible!




        Solution:
        • apply directional links only where
           appropriate.
Adoption Issues
    Critical mass required for effective utilization
    Solutions:
    • Work with aggregators (GBIF, VertNet, NCBI).
    • View Triples as a publishable unit




    Reality is complicated

    Solutions:
    • Work collaboratively (e.g.
       BioPortal, hackathons, interdisciplinary
       workshops)
The BiSciCol Mission

• BiSciCol tackles biodiversity data challenges:
    •    Tracking and integration of objects across disciplines
    •    Linking derivatives back to their source
• BiSciCol is about community, collaborative practice
    •    Commitment to standards, ontologies
    •    Agreement on permanent, resolvable identifiers
    •    Triplification of data sources to enhance linked data



        http://biscicol.blogspot.com/       http://biscicol.org

Triplifier talk

  • 1.
    Data Curation and BiodiversityResearch -- Lessons from BiSciCol and a look at the “Triplifier Simplifier” John Deck, University of California, Berkeley Brian Stucky, University of Colorado, Boulder Lukasz Ziemba, University of Florida, Gaineseville Nico Cellinese, University of Florida, Gainesville Rob Guralnick, University of Colorado, Boulder BiSciCol Team Reed Beaman, Nico Cellinese, Jonathan Coddington, Neil Davies, John Deck, Rob Guralnick, Bryan P. Heidorn, Chris Meyer, Tom Orrell, Rich Pyle, Kate Rachwal, Brian Stucky, Rob Whitton, Lukasz Ziemba
  • 2.
    BiSciCol is National Science Foundation funded 2010 – 2014 • Infrastructure to tag & track specimens & derivates in cyberspace • Relies on globally unique identifiers (GUIDs) to track objects • Implements a Linked Data approach • Provides support for the Global Names Architecture
  • 3.
    A Biological RelationshipGraph … Taxonomic Type Filter Class Filter X Specimens Tissues X Sequences
  • 4.
    Why Linked Data?Why BiSciCol? Here is Gustav’s Problem Generates Lots of Data… (Prefers to collect stuff)
  • 5.
    Biodiversity Data Challenges Data is Distributed Rapidly Changing Technologies Covers Multiple Domains
  • 6.
    Solving Biodiversity DataChallenges with BiSciCol and Linked Data Is a dwc:Event Group data into classes. Is a dwc:Event Assign identifiers. Link identifiers. [ ] Ocean Sampling Day Publish. [X] Moorea Biocode [X] SI MSNGR System [+] Add My Data
  • 7.
    The Triplifier (Advanced Interface) Loading Data Naming and Identifying Objects Linking Objects Publishing Powered by:
  • 8.
    Advanced Interface: LoadingData Darwin Core Archive Darwin Core Archive Spreadsheets Mysql KEMU MySQL
  • 9.
    Advanced Interface: Entities 78 Tissue Result is identifiers assigned to Entities: 78 a door . 427 a cat . <http://biocode.berkeley.edu/collectorspecimens/BMOO_2665> a <dwc:Occurrence> . <http://biocode.berkeley.edu/collectorevents/MIB_25> a <dwc:Event> . From Gary Larsen and adapted by Barry Smith in Referent Tracking Ceusters W, Smith B. Strategies for Referent Tracking in Electronic Health R Biomed presentation at the Semantics of Biodiversity Workshop, 2012. Inform. 2006 Jun;39(3):362-78.
  • 10.
    Advanced Interface: Entity Relations Relations as Triples: <http://biocode.berkeley.edu/collectorevents/MIB_25> <ma:isSourceOf> <http://biocode.berkeley.edu/collectorspecimens/BMOO_2665> . <http://biocode.berkeley.edu/collectorevents/MIB_37> <ma:isSourceOf> <http://biocode.berkeley.edu/collectorspecimens/BMOO_2667> . <http://biocode.berkeley.edu/collectorspecimens/BMOO_2665> <ma:isSourceOf> <http://biocode.berkeley.edu/plate_well/Plate_M037F10> . <http://biocode.berkeley.edu/collectorspecimens/BMOO_2667> <ma:isSourceOf> <http://biocode.berkeley.edu/plate_well/Plate_M028G5> .
  • 11.
    Triplify!: View graphbased data Response Query
  • 12.
    The Triplifier (SimpleInterface) Publish
  • 13.
    What challenges arewe facing now? (for BiSciCol, Linked Data, and data integration In general)
  • 14.
    Identifier Issues Persistence Solutions: • DOIs (http://doi.org/) • EZIDs (http://ezid.net/) Assignment at the source is difficult Solutions: • Calculated namespaces (e.g. geo:lat,lng) via PDAs • UUIDs (randomly unique) The digestible RFID tag Semantic web requires URIs but many standards (including scheme : string Darwin Core) do not require URIs for identifiers Solution: • Promote use of URIs for identifiers in all Standards. URI
  • 15.
    “Occurrence” Classification Issues Inadequate representational units Confusion between representational units “Sample, Specimen, Individual, Aggregation” Solutions: • Continue working on clarity in term definitions • Work from upper level ontologies (e.g. Basic Formal Ontology) to derive definitions.
  • 16.
    Relation Issues Non-sensical conclusions are possible! Solution: • apply directional links only where appropriate.
  • 17.
    Adoption Issues Critical mass required for effective utilization Solutions: • Work with aggregators (GBIF, VertNet, NCBI). • View Triples as a publishable unit Reality is complicated Solutions: • Work collaboratively (e.g. BioPortal, hackathons, interdisciplinary workshops)
  • 18.
    The BiSciCol Mission •BiSciCol tackles biodiversity data challenges: • Tracking and integration of objects across disciplines • Linking derivatives back to their source • BiSciCol is about community, collaborative practice • Commitment to standards, ontologies • Agreement on permanent, resolvable identifiers • Triplification of data sources to enhance linked data http://biscicol.blogspot.com/ http://biscicol.org