Part of the impetus for this reboot was identifying several cross-cluster and cross-domain issues that could use a more dedicated space for discussion during the 2015 Summer Meeting at Asilomar. On our first call, we were reminded of the history of the earlier Web Services cluster and, for me, the lesson and reminder is that we are often asking big questions that are too big to answer today. So what are some of the smaller questions that we, as a community, can ask to continue to make progress and, in asking, how can we take advantage of the domain knowledge across the ESIP clusters?

We can start very simply with the work on service and metadata validators in the Discovery and Documentation clusters (OpenSearch, CF and ACDD). When we consider these through the BCube semantics perspective (ideally we’d have self-describing web services but we are faced with longstanding documentation issues), what commonalities do we find in these validators that we can then apply to other services without? Is there some feature about a tool that leads to higher compliance or better quality documentation?

Moving from these discrete services, eg make a request and return a data blob, and move into the “big data” spaces, we have to consider streaming data services. So we can ask what services have been identified through the ESDA use case compilation or what services can satisfy those use cases? Because this type of data service comes from a different domain, how do we integrate those into some brokering or discovery system that relies on well-described services?

Talking about service discovery and data recommendation systems, we can return to the work Tom Narock did in the Semantic Web cluster and the ToolMatch team with Products & Services. Yolanda Gil, at a recent GPF training session, mentioned the notion of a notification system that could recommend some services or datasets (or tools) to you based on the data you, as a researcher, were working with. And that is the promise of projects like ToolMatch and BCube – can we characterize the services and relate them to their datasets so that a discovery platform can integrate with it and serve as a data store for the ToolMatch recommender platform? Early days, but it holds so much promise and relies so heavily on well-described objects.

I noted recently how many discussions across clusters and other communities involved stewardship of so many of the components we need to work on any this – ontologies, vocabularies, software (I would also include XML schemas and JSON contexts). We face the same challenges regardless of data type – governance, preservation, maintenance, versioning, discovery. Again, what commonalities can we identify, across the clusters asking these questions, to better pool our limited resources?

We went a little big there at the end, but it’s fascinating to think about the evolution of these systems across domains and knowledge spaces. If nothing else, in the Web Services cluster, I would like to consider some near-term practical options and to consider success stories in current work like that discussed in a recent IT&I webinar by Rich Signell (USGS) (thanks, Annie B. and Ethan D.) that other groups could apply today.

So there’s a Doodle poll for the October meeting:

If there’s no real consensus, it will remain the second Monday of the month at 4 Eastern.


Soren Scott
Chair, Web Services Cluster