1 Maintaining the momentum of OpenSearch in Earth Science data discovery Doug Newman (NASA ECHO) & Dr Chris Lynnes (GES DISC) 12/11/13 10:50am PT IN32A-03
What is OpenSearch? 2
A ‘refresher’ From ‘OpenSearch is a collection of simple formats for the sharing of search results’ The Earth Data discovery use case: 3 HTTP GET Request keyword = air temperature bounding_box = start_date = T00:00:00Z HTTP Response ID Spatial Extent Temporal Extent Metadata Link Data Link Search Link ID Spatial Extent Temporal Extent Metadata Link Data Link Search Link OpenSearch Descriptor Document
Earth Data OpenSearch today 4
Is it successful? ESA’s ‘Next Generation User Services for Earth Observation’ will be using OpenSearch as an infrastructure standard* CEOS’s CWIC effort will support OpenSearch in a future iteration that will include ESA data providers** ESIP Federation continues to champion OpenSearch for earth science data discovery*** NASA ECHO metrics: average number of queries per week – SOAP API (2011) – 10k (243k queries that year) – REST APIs (2013) – 87 of 115k (5 million queries so far this year) 5 * ** ***
Why is it successful? Lightweight and simple Standards-based RESTful Low entry cost 6 ‘Free text + spatial + temporal’ satisfies 90% of Earth Data discovery use cases* * Based on Reverb metrics for the last year (80,420 registered users, 700k queries so far this year (11/08/13)
ECHO Reverb Statistics 7 * Based on Reverb metrics for the last year (80,420 registered users, 700k queries so far this year (11/08/13) Caters to 90% of Earth Data discovery use cases*
Maintaining success 8
Converge where possible 9 CEOS / CWIC NASA / ESIP discovery cluster OGC / ESA / CNES CNES - Centre National d'Etudes Spatiales Attribution:
Converge where possible (for real) 10 CEOS / CWIC NASA / ESIP discovery cluster OGC / ESA / CNES CNES - Centre National d'Etudes Spatiales free_text bounding_box start_date end_date free_text bounding_box start_date end_date uid place_name geometry uid place_name geometry described_by via relation
Free text + spatial + temporal = success Pro: 90% !!!* Con: lack of free text precision compared with controlled vocabularies Can free text solve this?** free text = ‘MODIS’ (693 hits) != instrument = ‘MODIS’ (543 hits) free text = ‘ozone’ (348 hits) != science keyword = ‘ozone’ (81 hits) 11 *Based on Reverb metrics for the last year (80,420 registered users, 700k queries so far this year (11/08/13) ** Based on ECHO Catalog REST API queries and ‘fuzzy’ comparisons for ‘ozone’
Understanding the API OpenSearch parameter extension* is good (once we update it) – Defining expectations of ‘free text’ search. What does ‘air temperature’ mean? – Defining subset of ‘geometry’ capabilities 12 *
Additional functionality Result ordering – Described in OSDD and implemented in results Result ranking – As per OpenSearch ‘Relevance’ extension* – For free text search results – ECHO (January 2013) 13 *
How do we achieve these goals? 14 CEOS NASA / ESIP discovery cluster OGC / ESA / CNES Doug Newman Chris Lynnes Doug Newman Chris Lynnes Jérôme Gaspari* Yves Coene** Jérôme Gaspari* Yves Coene** * CNES - Centre National d'Etudes Spatiales ** ESA
Questions? 15