Abstract
Troubleshooting electronic resource linking issues can seem to be an insurmountable task—so many resources, so little time. Using interlibrary loan (ILL) data on requests for materials available online, the electronic resources staff at the Samford University Library detected problems with the implementation of their new link resolver. This data also provided a window into some systemic issues within the metadata of certain sources and the link resolver knowledgebase. In addition to helping us improve linking for our users, the establishment of a workflow for communicating cancelled ILL transaction data on an ongoing basis has also improved the communication between electronic resources staff and the ILL department regarding the overall linking process.
INTRODUCTION
Librarians are well aware that users who experience difficulties accessing library resources often neglect to report their problems to the library. This is a frustrating situation, because when librarians are not made aware of an issue, they cannot take steps to address it. In July 2012, Samford University Library implemented new link resolver and knowledgebase systems; and it was known that the change was sometimes leading to errors in OpenURL linking to online resources. However, librarians were getting relatively few reports from users that would allow them to identify and fix specific problems. It was decided to look at the information in cancelled interlibrary loan (ILL) requests to help find and troubleshoot these errors. The experience has also been documented in a 2015 article published in Serial Review.Footnote1
Founded in 1841, Samford University, located in Homewood, Alabama, has almost 5,000 students, representing 44 states and 22 countries. The university has 25 undergraduate and graduate degrees, and the university library serves all of these programs except for the Cumberland School of Law, which has its own library. The university library has almost 30 employees, including an electronic resources librarian and a cataloging technical assistant who splits her time between copy cataloging and e-resource duties.
USING ILL CANCELLATION DATA TO IDENTIFY LINK RESOLVER ISSUES
At Samford, ILL requests are cancelled when a requested item is found to be available online. An e-mail is sent to the requesting patron with correct citation information for the item, along with instructions on asking for help in using library services. While in some cases, a user has not checked the library’s holdings, and is simply using an ILL request to get the library to locate the material for them, the majority of ILL cancellations represent situations where the user legitimately could not find the desired material, even though they are supposed to be able to access it.
In a 2007 presentation at the OCLC ILLiad International Meeting, Karen Janke referred to interlibrary loan cancellations as “the indicator species of the library … a species whose presence, absence, or relative well-being in a given environment is indicative of the health of its ecosystem as a whole.”Footnote2 ILL cancellations, specifically, requests for materials the library already owns, indicate a potential failure within the library system where a user has selected but is unable to successfully access an item owned by the library. The Electronic Resources department requested that the ILL department copy them on the e-mails sent out when ILL requests were cancelled because of online availability of the requested item. Access to the citations in these e-mails was tested by the Electronic Resources Librarian, using the three most common pathways employed by users:
Testing journal or book title access via the library catalog
Testing book, article, or chapter title access via the library’s discovery layer
Testing book, article, or chapter title access via Google Scholar
Additionally, at the end of the school year, all data in the library’s ILLiad system about relevant cancelled requests were reviewed by the Electronic Resources Librarian and the cataloging assistant with the aim of identifying significant patterns.
In seeking to classify the types of problem encountered, the team referred to the Online Computer Library Center, Inc. (OCLC) whitepaper, “Success strategies for electronic content discovery and access: A cross-industry white paper,”Footnote3 which describes three core data quality problems encountered in resource linking:
The data used to make the link are incomplete or inaccurate.
The bibliographic metadata used by the link resolver and the library’s holdings data are not synchronized.
Libraries receive data in multiple formats, and must spend time and resources reformatting and completing the data.
Many of the systemic issues identified at Samford were found to correlate to these three core problems, as described in the examples that follow.
PROBLEM TYPE #1: INCOMPLETE OR INACCURATE DATA
An example of the first type of problem, incomplete or inaccurate data, was a case where elements of date information were omitted when a citation was passed from the database to the link resolver. The database contained date information from a particular publisher at the year and month level, for example 2007–12, or December 2007. However, when this information was passed to the link resolver, only the year (2007) was passed. Consequently, the link resolver added date information that did not consistently match the citations. In the above example, month and day values were added, making it appear that the article was published on 1 January 2007 (2007-01-01). When the fuller, but incorrect, citation was received by the target, it would be unable to find the requested article, because no such article exists with that title or page number in the January issue of that journal for that year. This type of error also occasionally led to articles erroneously being identified as included in, or excluded from, the library’s knowledgebase because Samford’s access did not include the full year. For example, if the library had access to the journal only from July 2007 onward, and the desired item appeared in the December 2007 issue of the journal, the mistaken conversion of the date to January 1, 2007 would make it appear that the resource was not available and the user would not be offered a link to the resource.
The second issue in this category involved inconsistencies between the metadata passed by the link resolver and the metadata stored for the article in the target database. In these cases the target was an aggregated database that would conduct a search for the article in question based on the metadata passed by the link resolver. In most cases this metadata was correct enough to support a successful search that led to the article desired by the user. However, in a number of cases identified by cancelled ILL requests slight differences caused the search to fail. For example, in the case of a journal with a single level of enumeration at the issue level, a request may get passed to the link resolver for issue 2. The link resolver would then pass this on to the target as issue 2, but when the target encountered a single level of enumeration, it was interpreted as a volume number. Because volume 2 of the requested journal did not exist, the search failed. Other errors of this sort included searches for an International Standard Serial Number (ISSN) that was correct but not recorded as the ISSN for the title in the database, and issues with non-matching first page numbers due to the inclusion or exclusion of the initial letter that is sometimes used in numbering the pages of journal supplements.
Problems with individual titles can be addressed as they are reported, but the systemic nature of the problem means that similar issues with other titles may still occur in the future, depending on the vagaries of the metadata involved. For particularly troublesome databases, a potential compromise solution is to turn off article-level linking and link only to the journal title.
PROBLEM TYPE #2: LACK OF SYNCHRONIZATION BETWEEN LINK RESOLVER AND LIBRARY HOLDINGS DATA
The second type of problem, resulting from the lack of synchronization between bibliographic metadata and the library’s holdings data, was exemplified by issues experienced with Google Scholar. A popular resource for faculty and students, Google Scholar can be configured to display links to a library’s holdings next to search results. It was discovered that sometimes, rather than in the expected place, these links are displayed under the “More” link below the citation, where they would be if the item was not found to be in the library’s collection and where users are unlikely to look for a link. This issue, which occurred even with items that had been listed in the library’s knowledgebase for a long time, can only be resolved by a better synchronization of Google’s service with library holdings metadata.
PROBLEM TYPE #3: DELIVERY OF DATA IN MULTIPLE FORMATS
The last example showed a problem with metadata harvested by the library’s discovery service from an Open Access database. While all the necessary information appeared to be present in the discovery service record, elements were mislabeled so that, among other problems, article titles were also passed to the target as source titles. Using this mismatched information, the link resolver was unable to find the desired article. While direct links to the Open Access content are provided in the discovery service records, they occur at the end of the full record and are difficult for users to locate. Moreover, the link may no longer lead to the desired content, making it a better option to use the link resolver to locate the appropriate copy. This kind of issue could be solved if providers used more consistent data formats, such as Knowledge Bases and Related Tools (KBART).
WORKFLOW IMPROVEMENTS
After the first year, it was clear that collecting data from cancelled ILL requests could continue to provide valuable information supporting identification and correction of electronic resource linking issues. However, it became apparent that simply having the ILL department copy the Electronic Resources department when sending an e-mail to the user with the correct citation was not a sustainable way to support this process. Sometimes, the ILL department neglected to copy Electronic Resources when sending the e-mail, and more importantly, the e-mail messages sent to users omitted useful information such as the name of the database involved in the access failure, so it was difficult to determine when a particular source was consistently passing bad information to the link resolver. A workflow was developed for the Electronic Resources department to access and query data in the ILLiad system on a daily basis, allowing them to examine more extensive information about cancelled ILL requests than was provided in the e-mails from ILL to patrons. This allowed them to see the source of the citations involved in problems, making it easier to identify where the error was occurring. In the case of particularly persistent problems, the user could be contacted directly with more information or a copy of the desired item while the problem was being addressed. Access to ILLiad also made it possible for the Electronic Resources Librarian to export data about the cancelled requests into a spreadsheet, where it can be anonymized and coded to permit tracking of troubleshooting performance and other analysis.
FURTHER BENEFITS
Beyond enabling identification and resolution of issues with the link resolver, Samford’s analysis of data from cancelled ILL requests has brought benefits in a number of other areas such as:
Instruction and outreach: Demographic groups who are particularly inclined to make ILL requests for locally available materials can be identified as targets for supplementary instruction in using the link resolver and requesting assistance when errors are encountered.
Improved collaboration between departments: The Electronic Resources department benefits from receiving valuable data from ILL, and in turn provides links to include in e-mails sent to users when requests are cancelled.
Interface design: Analysis of linking failures has suggested possible enhancements to the link resolver window, including improved placement of links and implementation of “one-click” linking to full text.
Thinking about inclusion of links to Open Access materials (via a Google Scholar search) and print holdings in the link resolver, to increase alternatives for access.
Staff training in troubleshooting the knowledgebase and link resolver.
DISCUSSION
After the presentation, an attendee asked about the option of simply eliminating sources of problematic citations, by deselecting them in the discovery layer configuration. The presenter answered that this approach had been considered at Samford, but pointed out that the choice is not easy if the resource in question provides valuable information. Another attendee asked if there was any response from users regarding the assistance provided in accessing material. The presenter estimated that while about 75% of users contacted remained silent, the other 25% responded very positively, confirming for her that this process was a service opportunity offering a good way to establish positive relationships with users by saving their time.
Additional information
Notes on contributors
Beth Ashmore
Beth Ashmore is Metadata Librarian for Serials & Electronic Resources, Samford University Library, Birmingham, Alabama.
David Macaulay
David Macaulay is Collection Development Librarian, University of Wyoming Libraries, Laramie, WY.
Notes
1. Beth Ashmore, Emily Allee, and Rebekah Wood, “Identifying and Troubleshooting Link-Resolution Issues with ILL Data,” Serials Review 41 no. 1 (2015): 23–29. doi:10.1080/00987913.2014.1001506 (accessed 29 June 29, 2015).
2. Karen L. Janke, “The $64,000 Question Answered: Why Do Patrons Place ILL Requests for Items that the Library Already Owns?” (paper presented at the OCLC ILLiad International Meeting, Virginia Beach, Virginia, March 15, 2007). https://scholarworks.iupui.edu/handle/1805/901?show=full (accessed June 29, 2015).
3. Suzanne Saskia Kemperman et al., “Success Strategies for Electronic Content Discovery and Access: A Cross-Industry White Paper” (Dublin, OH: OCLC, 2014). https://www.oclc.org/content/dam/oclc/reports/data-quality/215233-SuccessStrategies.pdf (accessed June 29, 2015).