search

ohbm / osr2020

Website for the Open Science Room at the OHBM 2020 meeting
https://ohbm.github.io/osr2020
Other
18 stars 6 forks source link

Open Data 2.0 (Lightning talk): Concept Annotations via NIDM-Terms Fosters Improved Search of OpenNeuro Datasets #27

Open jsheunis opened 4 years ago

jsheunis commented 4 years ago

Concept Annotations via NIDM-Terms Fosters Improved Search of OpenNeuro Datasets

By David Keator, University of California, Irvine

Abstract

Efficient use of existing data relies on (meta)data being FAIR: Findable, Accessible, Interoperable and Reusable. A critical barrier to FAIR data is that metadata descriptions vary widely in degree of detail and are inconsistent in terminologies used, making neuroimaging data often not reusable without significant interactions with the original authors. This lack of consistent metadata makes comparisons and integration of data across studies and sites difficult. The Brain Imaging Data Structure (BIDS) was designed to provide software developers and the neuroimaging community with file and directory naming conventions for organizing imaging data which has improved our ability to reuse existing data. In BIDS, the organization of the data is required to conform to strict naming and directory structure organizations. Beyond file naming and tightly-controlled imaging directory structures, there are very few constraints on ancillary variable naming/meaning or experiment-specific metadata. As such, searching and combining information across independent BIDS datasets can be difficult.

To facilitate search across datasets we developed the Neuroimaging Data model (NIDM), built using linked-data techniques to unambiguously describe experimental data, workflows, and results. The NIDM-Terms component (NIMH RF1 MH120021) aims to provide techniques for annotating datasets with the explicit goal of improving search across publicly-available data. Here we report on improvements in annotating BIDS datasets with important properties necessary to fully understand the variables contained within. We facilitate search across BIDS datasets via concept annotations from the Cognitive Atlas and the InterLex information resource. We apply these techniques to OpenNeuro BIDS datasets and show the benefits of concept annotations and NIDM for improved search.

Useful Links

https://github.com/NIDM-Terms/terms https://scicrunch.org/nidm-terms

Tagging @dbkeator

dbkeator commented 4 years ago

Looks good, thanks!

dbkeator commented 4 years ago

Slides link (note these don't have the CC/audio/video overlays done for the OS SIG talk but I'm happy to connect with you for questions): https://docs.google.com/presentation/d/1hz-h66wCKBwmzDU-2uTfTs_q4FcXPqRHBXjoLFPsLU0/edit?usp=sharing