Allyson Lister
Breakout 3: Author identity – Creating a new kind of reputation online: Duncan Hull, Geoffrey Bilder, Michael Habib, Reynold Guida
Geoff Bilder starts with a definition of terms. - Martin Fenner
...and with a Mini Me - Allyson Lister
The same person can have many profiles. - Martin Fenner
....and even more Mini Mes - one for each persona - Allyson Lister
identifier for a profile; multiple personas for each person, each with their own profile - Allyson Lister
Why do we want an author identifier: to have a scholarly record. - Martin Fenner
Two separate issues of identity: security and access control vs. identification. - Martin Fenner
Nice slide depicting the "scholarly record" - Daniel Mietchen
Requirement for access control systems are far higher than for identification systems. - Martin Fenner
Scope of author identifier: international, interinstitutional and interdisciplinary. - Martin Fenner
CrossRef has assigned DOIs to scholarly work as far back as the 1600s. - Martin Fenner
CrossRef deals with 200.000 DOIs per month, on average 5 authors per paper. Whatever we build has to scale. - Martin Fenner
99% accuracy is a disaster. - AJCann
hmm, I look forward to the discussion, I've heard all of the content of the presentations before, though they are being presented well here. - Ian Mulvany
I agree, the interesting part is how we move forward from here. We have the right panel for that. - Martin Fenner
Bouncing users from one site to another is a disaster - looks like phishing to most users. We're trying to train them to avoid this! - AJCann
Duncan Hull summarised the weaknesses of OpenID nicely. - Daniel MacArthur
Geoff Bilder made a strong case against OpenID and other distributed identification systems here: - Martin Fenner
Scopus Author ID. Pros: automated, validated by publications, 99% precision, 95% recall. Cons: Impersinal: 99% precision, 95% recall. - Martin Fenner
auto-generated about 7 million profiles of current researchers, pretty impressive. - Ian Mulvany
Scopus Author ID allows feedback from authors. - Martin Fenner
no discussion so far about foaf, "rel=me" or knitting the web together in a way that enables auto-aggregation of profiles. - Ian Mulvany
What we notice is that the number of multi-author papers has started to increase, while single-author papers have decreased. - Allyson Lister
ok, now this is getting interesting, we have someone from Thompson, so we have AuthorID, ResearcherID, OpenID and proponents of DOI as ID all sitting at the same table, I think they should have a jello wrestling contest to determine which system gets adopted. - Ian Mulvany
scopus's good stats are surely skewed by so many many single publication authors. my experience of scopus is not so good. - Joe Dunckley
Author identities can help with the discovery process. - Martin Fenner
Introduction of ResearcherID: open and secure. - Martin Fenner
researcherid is concerned to be secure and open, secure in that the researcher controls their profile, and open in that they provide widgets, that allows the researcher to link to their profile (I don't know if that is the same kind of open as the kind of open that I'm used to thinking about) - Ian Mulvany
have institutional tools to allow institutions to batch create profiles, could be smart. now Q & A. - Ian Mulvany
ResearcherID uses Web of Science as its publication database. Yep, really open. - Bob O'Hara
there is a lot of tip-toe talk to avoid a turf war going on here - Allison Coles
+1 Joe re: Scopus stats. The 99% figure seems a bit too high! - Euan
Well, my question to the panel did not tip-toe! ("Do all these systems talk to each other?"). Have also heard Scopus not as good as that via some other conference report. - Maxine
@Euan - I agree, from using Scopus and being very frustrated with their author profiles, there's no way they can claim 99% precision - but what do precision and recall actually mean in this case? I would understand accuracy huh. - Christina Pikas
I read through these posts and I worry that the vision that Geoff espouses is so far off as to never gain traction in the marketplace. It's similar to XRI, which requires registration through a central service. While it's fair to point out that we "lease" domain names, there are legal protections for them as well. Anyway, it's an interesting discussion — glad OpenID is part of it. - Chris Messina
Hi Chris, I am working out a response, but in the meantime, a related and more detailed presentation of Geoff's is available here: Duncan's presentation focused on OpenID more than ours. All our presentations should be up on Nature Proceedings in good time. - Michael Habib
@Chris - the reason why centralized author/contributor IDs is expected to work is that there's a big incentive for the various stakeholders (authors and publishers alike) to sort out what is a pretty poor situation in the scholarly publishing domain, with the author name problem and whatnot. There's precedence for this: DOIs for scholarly publiations and associated social/technical infrastructure. - 'Mummi' Thorisson
Yeah — I understand. I guess I think that an OpenID-style solution is what should be the foundation of such a system — and that the centralized service should act as a "metadata provider" — basically like Gravatar: input a particular author's URI/URL, get out a list of known/verified publications that they contributed to. - Chris Messina
@Chris - that's where opinions start to differ :) Some argue Contributor IDs should *be* OpenIDs, while others argue for a dedicated identifier and OpenID should be the authentication mechanism (one of perhaps several, including Shibboleth, username/pwd etc.) for contributors to sign in to manage their profile and interact with it in other ways. - 'Mummi' Thorisson
...a major point supporting the latter is that the contributor identifiers need to exist for authors who never obtain an OpenID and aren't likely to, say one-time authors who never published more than that one article in 1994, or are inactive for some reason or another, e.g. deceased. - 'Mummi' Thorisson
It's worth point out that Geoff's perspective comes from CrossRef where centralization has been essential to maintain active redirects for DOIs. There is a serious social problem for research that means we have really bad users of technology combined with really serious requirements for reliability. That said given the mess there is going to be in this space keeping registration, certification, authentication, and access control as lightly coupled as possible may well win. We can centralize registration without loosing the benefits of distribution (maybe, perhaps). Actually - that's just what Chris just said a whole lot better isn't it? ;-) - Cameron Neylon
@Cameron - yeah, that's right. You da man :P - 'Mummi' Thorisson
I'll just leave it up to you to fix shall I Mummi? - Cameron Neylon
This may be useful for the discussion: summary and meeting minutes from the IRBW2009 workshop on research identity - - 'Mummi' Thorisson