Managing freely available e-resource collections with today's vendor provided OpenURL knowledgebases: A challenge in quality control
Mike Bloomberg, Systems Librarian, Augsburg College
Chad Hutchens, E-Resources Librarian, University of Wyoming
Background Information:
Chad and Mike were interested in comparing the problems and good things about free e-resources in the SFX and Serials Solutions KBs. Mike works with SFX; Chad works with Serials Solutions.
"Open Access" vs. "Free" - they're not the same, and the presentation isn't going to glorify one or the other.
Creative Commons - some equate CC licenses with "free" or "open access" but that's not always the case. Some CC licenses are actually very restrictive
The fun ones: sorta free, sorta open access, and other variations
The problems with free/OA resources are similar to problems with subscribed e-resources, and we contract out these services to vendors, b/c we don't have time to deal with them.
Chad points out a problem with their study - it's difficult to identify free resources in link resolver KBs, but they picked a specific list from both link resolvers.
Assumptions:
Free/OA collections in OpenURL KBs are maintained just like any subscription collections.
There are criteria for inclusion of titles in proprietary collections.
Titles in such collections offer full-text access.
Titles in Dollections such as DOAJ and Highwire Free are indeed free or OA.
Questions to answer:
Are the assumptions correct?
What are the criteria for inclusion?
Are titles' free and OA statuses being evaluated by SFX and SerSol?
Why are there so many errors in these collections?
Something else to think about:
Do we as libraries place the same emphasis on maintaining access to and understanding what's actually in free and OA collections as we do to their importance in promoting the concepts of "free" and "OA"
Free and Open Access survey
Not spending a ton of time on the survey itself, but results are available - they will send as a CSV file.
Survey was online and anonymous
Advertised on numerous email lists
Used U.Wyoming's in-house survey tool
Chose the resources they did because of the survey results.
97 respondents
89% academic libraries, 8% special, 3% public
60% have 3 or fewer people working on e-resources
52% use SerSol, 29% SFX, 19% other
45% use a MARC service, 55% do not
Themes and observations on the survey results:
DOAJ and Highwire Press and 1st and 2nd in both SerSol and SFX
Proprietary "free" are 3rd most predominant.
Majority of respondents indicate a "why not turn on?" rationale for inclusion of free/OA collections.
About half of the libraries are turning on collections entirely; the rest are doing title-by-title selection.
Many respondents are aware that such collections are problematic and unstable.
Some facts about free/OA collections:
DOAJ issues:
Not all O/A (seem MacCallum, C., 2007 - in PLoS)
Title list and error rates were very different b/t SerSol and SFX (could not download a master list from DOAJ itself)
Different definitions of OA!
Proprietary "free" collection issues:
Lack of maintenance
Lack of selection criteria
Not necessarily free/OA
Not necessarily journals
Analysis of collections
Ran links through a link-checker (used Linklint, b/c don't have to feed it an html page - can give it a list. http://www.linklint.org)
Manually checked for accuracy of listings among successful links (random sample of successful links)
DOAJ:
All three listings - SFX, SerSol, and doaj.org have different numbers of titles. Mike showed the Linklint output and described his process of dumping the output into Excel and reading the errors to decide what was a permanent vs. nonpermanent error (Linklink looks at html header codes - 404, 500, etc).
Results:
SerSol: 1.5% looks like permanent error
SFX: 3% looks like permanent error
Free Titles
Much worse error rate:
SerSol: 6.4% permanent error; 13.5% temporary error
SFX: 8.8% permanent error; 2.2% temporary error
(but Mike only ran links through the checker once; to replicate, he might do twice to check the temporary error rate)
Manual title checks (randomly-selected 50 titles from each KB provider):
No titles demanded registration
Date ranges were wrong about 40% of the time (Mike says in SFX, date threshold was less than actual available, which is bad, but not as bad as the other way).
Overall, found easy to find articles; they were usually 1 or 2 clicks from the entry URL
The obvious:
Dead links reduce patron confidence
Incorrect metadata confuses patrons
We pay subscription fees primarily because we don't have the staff to maintain a KB
We pride ourselves on collection development criteria, but don't apply that criteria to these collections
Paradox between promotion of free/OA as a concept, and not spending a lot of time vetting these collections.
Interesting questions from audience:
Did look at link resolver data to find out if the targets in question were being used? How often are people trying to get full-text through those targets? One attendee had looked at this question at the target level, and in ~6 months of activation, usage made up 17% of full-text access attempts.
Comments