-
Notifications
You must be signed in to change notification settings - Fork 4
NotesOnTaxonomicNameIdentifiers
This is the record of an email thread about taxonomic name identifiers
From: Steve Baskauf [mailto:[email protected]]
Sent: Friday, October 16, 2015 8:09 AM
To: Dmitry Mozzherin
Cc: Chuck Miller; [email protected]; Jonathan A Rees; Shorthouse, David
Subject: Re: [tdwg-content] status of uBio
Thanks all for the information and comments about the status of uBio. I'm glad to hear that the server will probably come back up. If not, then I hope the data will be made available to those who said they would be willing to host it.
I have been interested in using the uBio identifiers for several reasons:
1. They have managed to stick around for a long time and are stable in their format (as LSIDs and HTTP proxied LSIDs).
2. The coverage of names is really good for plants, animals, different geographic locations, etc. I also use ITIS identifiers but it's fairly common for me to not be able to find one for the name I need, which almost never happens with uBio.
3. It's somewhat clear what uBio identifiers refer to: names vs. something more nebulous involving taxa or ... something. (Not trying to push your button, Rich Pyle).
4. You can actually get RDF associated with the LSID version of the uBio identifiers. I was wanting to download some to play with in our new triplestore (http://rdf.library.vanderbilt.edu) when I discovered that the server was down. The RDF is somewhat ad hoc, but hey, it's there.
There isn't really any other source that has all of these characteristics. So please keep uBio going indefinitely, if at all possible.
Steve
Dmitry Mozzherin wrote:
I had been administering uBio for the last year, but now I am moving from MBL. uBio machine is in a bad shape, and it crashes after a few hours of work. My plan is to create Docker containers for database, code and data, which should make whole system much more stable, and much more manageable. Good news I will definitely try my best to do it, the bad news I am spread thinner than usual with move, transferring hardware and grant, GN things, EOL things, and figuring out what to do with the house etc. uBio 'code' part is about 35 Gb, which makes the task more complicated, but I am quite optimistic that I will be able to make containers and put them either on an MBL machine, run it from University of Illinois, or give it to Naturalis -- depending on what will make more sense for Dave Remsen, MBL and all interested in the project.
Subject: RE: [tdwg-content] status of uBio Date: Fri, 16 Oct 2015 14:04:56 +0000 From: Chuck Miller [email protected] To: Steve Baskauf [email protected]
Hi Steve, Have you tried the GNI? http://resolver.globalnames.org/ I think GNI ingested all of the uBio names when it first started. One reason the plants are more complete in uBio was that all of the Tropicos names were given to them.
Chuck
Subject: RE: [tdwg-content] status of uBio
Date: Fri, 16 Oct 2015 14:07:46 +0000
From: Nicky Nicolson <[email protected]>
Hello Steve,
Thanks for triggering an interesting thread.
Just waving the IPNI flag for a minute:
1. They have managed to stick around for a long time and are stable in their format (as LSIDs and HTTP proxied LSIDs).
At ipni.org we support our own HTTP proxy for LSIDs:
http://ipni.org/urn:lsid:ipni.org:names:12345-1
… and it’s been an age since I tried to resolve an LSID using the formal LSID spec but a quick run-through today shows that all the steps appear to be in working order.
2. The coverage of names is really good for plants, animals, different geographic locations, etc. I also use ITIS identifiers but it's fairly common for me to not be able to find one for the name I need, which almost never happens with uBio.
IPNI is comprehensive for vascular plants (at species level). We’ll be addressing the data gaps at infra-specific level – but it’s very useful for us to be armed with reasons (from users like yourself) as to why we should spend time doing this.
3. It's somewhat clear what uBio identifiers refer to: names vs. something more nebulous involving taxa or ... something. (Not trying to push your button, Rich Pyle).
IPNI only serves data about names, no taxa here.
We are pushing IPNI IDs into our taxonomic resources so that a user can flexibly match a name and get an IPNI identifier (the nomenclatural part), and then as a later step query a taxonomic resource for their current view as to the taxonomic status of that name. We are aiming for a clean separation of names matching from the (multiple, potentially different) uses of those names to form taxonomies.
4. You can actually get RDF associated with the LSID version of the uBio identifiers. I was wanting to download some to play with in our new triplestore (http://rdf.library.vanderbilt.edu) when I discovered that the server was down. The RDF is somewhat ad hoc, but hey, it's there.
The HTTP proxy above returns RDF metadata for the specified record.
New developments (most of the functionality outlined above is 8-10 years old), I’ve been working on exposing IPNI data through standard match services – namely the Open Refine reconciliation API, which permits much more flexible names look-ups using a lot of heuristics gathered from matching data port exercises carried out at Kew over the past few years. Currently the main hitch with this service is that it includes duplicate records. This is something we’re working on resolving right now.
The service is outlined here: http://data1.kew.org/reconciliation/ and details of the IPNI service in particular are here: http://data1.kew.org/reconciliation/about/IpniName
If you’re interested in using IPNI, I’d be happy to hear any comments on the functionality above and / or any requests as to how we can make it more useful for you.
cheers,
Nicky
Subject: Re: [tdwg-content] status of uBio Date: Fri, 16 Oct 2015 15:29:47 +0000 From: Roderic Page [email protected] To: Steve Baskauf [email protected]
Hi Steve,
if it’s animals you are after and the following characteristics (shared with IPNI):
- comprehensive
- names not taxa
- LSIDs that deference to RDF using consistent vocabulary (TDWG LSID)
- closed, or at least non-open
then ION (http://www.organismnames.com ) is for you. Each name has an LSID that actually works and returns a name plus (often) a bibliographic citation as a string.
My frustration with Ion being closed, and only serving bibliographic strings instead of DOIs, Handles, BHL BioStor etc. was what led me to cough copy ION and augment it to create BioNames http://bionames.org
For example, ION has urn:lsid:organismnames.com:name:5109777 and BioNames has http://bionames.org/names/cluster/5109777 (plus the original description with DOI http://dx.doi.org/10.11646/zootaxa.3616.4.3, see http://bionames.org/references/384cdd42133009235d59af9b41c55290
I’m working on a similar mapping of references to DOIs, etc. for IPNI...
ION exposes RDF, BioNames doesn’t (which won’t surprise you I guess, it has a JSON API). If it was of use I could add RDF to BioNames, probably JSON-LD.
Regards
Rod
Subject: RE: [tdwg-content] status of uBio Date: Fri, 16 Oct 2015 09:25:21 -1000 From: Richard Pyle [email protected] Reply-To: [email protected]
It's somewhat clear what uBio identifiers refer to: names vs. something more nebulous involving taxa or ... something. (Not trying to push your button, Rich Pyle).
:-)
Actually, saying that uBio identifiers refer to “names” is a bit like saying uBio identifiers refer to “stuff” (which is only slightly more ambiguous than “names”). I think what you mean (a bit more explicitly) is that uBio identifiers refer to “name-strings”, wherein each unique literal UTF-8-encoded string of characters purported to represent a scientific name receives an integer as an identifier.
GNI also assigns unique identifiers to such name-strings, but the difference is that the GNI identifier is a hash of the string itself. Thus, given a string, you can algorithmically derive the GNI identifier for it. There is no way to algorithmically convert the uBio name-strings into their corresponding integer uBio identifiers.
I have a question, and a suggestion.
Question: Are new uBio integer identifiers ever going to be minted? Or do we just want to maintain them for legacy purposes?
Suggestion: Regardless of the answer to the question, I suggest that Dima generate an index for uBio integer identifiers and the corresponding hash uuid used by GNI. I will import these into BioGUID.org, so going forward we will always have an index to translate the uBio integers into the GN hashed UUIDs. From there, it would be a simple step to allow GN services to process uBio identifiers natively, and do all the cool things that Steve (and others) would like uBio/GN to do.
Aloha,
Rich
Subject: Re: [tdwg-content] status of uBio Date: Fri, 16 Oct 2015 18:35:46 -0400 From: Dmitry Mozzherin [email protected]
All name strings from uBio are in GNI and they do have UUIDs v5 generated from them as it is described in this blog post - http://globalnames.org/news/2015/05/31/gn-uuid-0-5-0. And we do have uBio IDs associated with these UUIDs via GNI.