September 9, 2013

Tech Review meeting 9/9/13

Present: William Ulate, Trish Rose-Sandler, Mike Lichtenberg

Art of Life- Running of algorithm on cluster. Martin and Bob Corrigan talked last Friday. Not sure of results other than an in-person meeting was set up for Sept 23rd& 24th for Joel, John, Joe and William. Martin will be there too. Alternative ideas - Have Kyle try to run a subset of records on an IMA server in the meantime so we have something rather than nothing. If we get code from Anthony for PHP maybe we or Kyle can review it. Mike says Still dependent on knowing file paths on cluster. Talk with other places that have robust processors – IA, Egypt

JSON export data from algorithm –
Mike wants to know if we want to store all the data present in the JSON files in BHL? *We decided that no, we don’t need to pull everything out – just what we need to send to Joel. We can hold onto the JSON files and retrieve other data in the future if needed.
Sum of block coverage will probably be more useful than page resolution. Resolution from IA is often not accurate. Determine if we want Compression, Bounding and Color to be stored in BHL or left in JSON files.

TDWG – abstracts due Sept 25th. We need to have a call to plan our talks this week or next.

Exec Council meeting – link outs and citations that go nowhere. If going out of system we need a page that makes that clear. Link outs should open a new page. Put everything in beta so folks can do some beta testing. They are OK with citations that will not resolve to a record. These will not be indexed and show up in browse or searches but you can still get to them if you have the identifier number. This allows us to create persistent ids for all biodiversity resources, even those not digitized and online yet.

Outstanding grants - No word on IMLS NLG for Libraries yet. William found out we will not hear from Digging Into Data until January of next year!
Trish and William will meet with Mike Ryan on Wed to discuss future CBI grants and Library grants

Mike expects to get portal code to github this week. Lets not publically announce yet.

William says IT should have the virtual machines set up by end of Sept (not solr and refBank necessarily)
Types of cluster relationships- For anything which the machine has clustered (exact match of “1”) will get as sameAs related type. For other scores between .5-.9 there might relationships and we will set aside for futher review in the future. We can define the other relationships such as partOf, relatedTo (e.g. via subject)
Can a record be in more than 1 cluster? We could allow for that if needed.