iDigBio will be at the annual TDWG conference again this year and we hope to see you there! TDWG 2021 will be hosted virtually by the University of Florida Biodiversity Institute, which iDigBio PI Dr. Pam Soltis is affiliated with.
See more about the conference, including information about registration, program, etc. on the TDWG website here. Events that iDigBio is co-organizing include:
SYM01 Applications of machine learning in biodiversity image analysis
Organizers: Quentin Groom, Meise Botanic Garden, Meise, Belgium; Elizabeth Ellwood, iDigBio, University of Florida, Los Angeles, CA USA
Openly licenced images from natural history collection specimens number in the tens of millions, as do images of biodiversity from research, community science and camera traps. This is only likely to increase, as is the rate of growth. In parallel, we have seen numerous experiments with automated image analysis, which have proven the feasibility and usefulness of machine learning with images of biodiversity. These techniques include species identification, recognising species interactions, image segmentation, object detection, trait detection, and trait measurement. Currently, the outlook looks bright for future research on machine learning on these images and for new ecological research fueled by data from machine learning. Yet we are only now discovering what the limitations are and what infrastructure we need to underpin such research. This symposium will showcase some of the potential applications of machine learning as well as the technical and scientific techniques used with the aim of encouraging research in this field.
SYM07 Digital Extended Specimens
Organizers: Alex Hardisty, Cardiff University, Cardiff, UK; Andrew Bentley, University of Kansas, Lawrence, KS USA
A plenary keynote talk will explore the kinds of science that can be enabled by Digital Extended Specimens as counterpart objects on the Internet coupled to physical specimens in collections; science that either can't be done today or that is very difficult to do now.
A complementary symposium session will explore Digital Extended Specimens (DES) and the outcomes of the 2021 global community consultation coordinated by the alliance for biodiversity knowledge and the exciting possibilities of digital representations of the billions of specimens currently held in the world’s natural science collections for research. Mechanisms for how our global community can collaborate will be explored together with details and explanations of technicalities, the role of openDS and other standards; approaches to implementation; and implications for institutions.
The aim will be to demonstrate how people, processes and tools can unite and align around DES to work better together to build a fully integrated next-generation digital data infrastructure for biodiversity and other natural science data in which DES act as anchoring points for connecting the world of natural sciences specimens and other data, thus making reliable knowledge and evidence about the natural world available to all.
SYM17 Assuring trust on community science biodiversity platforms: Policies and approaches
Organizers: Elizabeth Ellwood, iDigBio, Los Angeles, CA, USA; Rob Stevenson, University of Massachusetts Boston, Boston, MA, USA
Data collected by citizen scientists make up a strong and growing part of biodiversity occurrence data yet there is still much discussion around the best approaches to produce, identify and make available high quality, trustworthy data. How do citizen science managers, projects, and platforms streamline workflows such that the data produced are trusted and ready for research? What metrics are most helpful for evaluating trustworthiness in citizen science data? Is it possible to reach a point where trustworthy data are apparent and speak for themselves? In this symposium, hosted by the TDWG Citizen Science Interest Group, speakers will discuss trustworthy data, the ways in which projects and platforms may programmatically classify, identify and account for data quality, and the role of standards and protocols in robust citizen science projects.
UNCF19 The API Unconference
Organizers: Deborah Paul, Illinois Natural History Survey, Species File Group, Champaign, IL, USA; Quentin Groom, Meise Botanic Garden, Meise, Belgium; Nicky Nicolson, Kew, Richmond, London, England; Matthew Yoder, Illinois Natural History Survey, Species File Group, Champaign, IL, USA; David Shorthouse, Ag-Canada, Ottawa, Canada; Cat Chapman, iDigBio, University of Florida, Gainesville, FL, USA
This half-day workshop/hackathon will explore how connected our global community is by proxy of how our collective Application Programming Interfaces (APIs) can be theoretically or practically linked together. Our focus on existing data-serving APIs has several benefits. First, it focuses on actually built software, rather than proof of concept pilot projects; this exposes and illuminates the differences between standards and implementations so that improvements can be made on both sides. Second it explores a long-term future scenario, where every data provider exposes their own API and scientists, and users crawl across "the source", rather than inferred or interpreted versions of the source. That is, if we are unable to link the existing ecosystem of APIs at present, where can we make improvements so that this vision is met?
First, together we will quickly enumerate the existing APIs we each know about. Our goal is not to be comprehensive, but rather to collectively aggregate known application resources (this itself is a nice proxy test of accessibility). From there we will proceed to the best of our ability, in the time provided, by linking the APIs, and documenting these linkages, in three ways: 1) by quickly observing that some combination of responses from 2 or more APIs could be used to do X, i.e. "theoretically linked"; 2) by identifying sets of specific (real) API calls across two or more providers that represent a specific use case, i.e. "demonstrably linked" and 3) time permitting, by coding tiny proofs of concepts that demonstrate integration across two or more APIs, i.e. "actually linked". Each of these three ways can be run in parallel based on participants' interest. We plan to offer a small prize in each category.