Lightweight Advertising and Scalable Discovery of Services
Discovery and use of Web Services for querying, accessing, and processing Earth Science datasets is hampered by the lack of an open and web-scalable services and data collection registry that provides rich search for all available services and datasets, interesting geophysical events, and data granules relevant to studying those events.
Existing registries, like NASA's Global Change Master Directory (GCMD), EOS Clearinghouse (ECHO), or the Global Earth Observation System of Systems (GEOSS), provide limited search capabilities (only text keyword), often have difficult interfaces for registering services or datasets, each use their own metadata standard, don’t evolve their metadata fields (registered information) rapidly to suit user needs, and compete with each other for adoption, thereby fragmenting the user base. To find all available web services, the user must search all three registries, and even then the results may omit many unregistered services and datasets.
The PI has promulgated a "service casting" approach in which services are openly advertised on the web in Atom syndication feeds, which are searchable at Google and can also be aggregated to provide smart, faceted search (semantics beyond text keywords). The serv-cast v1 (metadata) standard is currently being vetted by Earth Science Information Partners (ESIP) members and a catalog of published servcasts is already being accumulated.
Servcasts can be used to search by service taxonomy, look up the service interface (e.g. Web Service Definition Language), machine auto-invoke the service, or click through to service documentation for humans. If a provider's services change or expand, the advertisement can be modified and re-cast.
The service provider is in control, and can add metadata fields, or even information intended for its own use, to the extensible servcasts at any time.
We propose to develop a lightweight service advertisement mechanism that will improve web discovery of provider's Earth Science services, while also being backward-compatible by providing search over existing metadata repositories. The objective is to provide a one-stop search box in the browser where users can search by keyword and service taxonomy. By combining service casting with meta-search, we will provide a novel and complete solution:
- Provide a lightweight, easily-authored mechanism for ES service producers to publish and promote their services and datasets;
- Enable service consumers to easily find and invoke services, both those published as service casts and those registered in the GCMD, ECHO, and GEOSS repositories ("meta-search");
- Provide a rich search interface as a browser plug-in that has modern, web 2.0 capabilities: search term suggestions for the user, term synonyms and broadening using semantics, search by user tags, and integrated social tagging so users themselves can enhance the categorization of services, rank them by usefulness or performance, tag them for use in collaborative service chains, etc.
Similarly, metadata describing interesting geophysical events (hurricane tracks) or periodic structures (El Nino), with links to related datasets and services, can be advertised in discoverable event-casts.
Under this work, we will define the event-cast standard, leveraging International Organization for Standardization metadata standards, and populate a body of Hurricane casts as examples. Users will be able to tag services and events using a pre-defined hierarchical taxonomy or their own categories, publish new events, link datasets to events, and rank services by suitability for purpose, performance, availability, etc. The social tagging and semantic search capabilities will be implemented by extending Noesis technologies developed by Co-I Ramachandran.
Service casting and meta-search will improve the 'findability' and usability of all of the service and dataset metadata currently registered in existing repositories: GCMD, ECHO, and GEOSS. Increasing the use of Web Services will increase the use of NASA's Earth science data, especially for large-scale automated workflows that support climate science.
Last Updated: Oct 19, 2018 at 2:39 PM EDT