1、 NISO RP-21-2013 Improving OpenURLs Through Analytics (IOTA): Recommendations for Link Resolver Providers A Recommended Practice of the National Information Standards Organization Approved: April 26, 2013 Improving OpenURLs Through Analytics (IOTA) ii NISO RP-21-2013 About NISO Recommended Practices
2、 A NISO Recommended Practice is a recommended “best practice” or “guideline” for methods, materials, or practices in order to give guidance to the user. Such documents usually represent a leading edge, exceptional model, or proven industry practice. All elements of Recommended Practices are discreti
3、onary and may be used as stated or modified by the user to meet specific needs. This recommended practice may be revised or withdrawn at any time. For current information on the status of this publication contact the NISO office or visit the NISO website (www.niso.org). Published by National Informa
4、tion Standards Organization (NISO) 3600 Clipper Mill Road, Suite 302 Baltimore, MD 21211 www.niso.org Copyright 2013 by the National Information Standards Organization All rights reserved under International and Pan-American Copyright Conventions. For noncommercial purposes only, this publication ma
5、y be reproduced or transmitted in any form or by any means without prior permission in writing from the publisher, provided it is reproduced accurately, the source of the material is identified, and the NISO copyright status is acknowledged. All inquiries regarding translations into other languages
6、or commercial reproduction or distribution should be addressed to: NISO, 3600 Clipper Mill Road, Suite 302, Baltimore, MD 21211. ISBN: 978-1-937522-18-6 Improving OpenURLs Through Analytics (IOTA) NISO-RP-21-2013 iii Contents Foreword . v Section 1: Introduction 1 1.1 Purpose and Scope 1 1.2 Terms a
7、nd Definitions . 1 Section 2: Determining Optimal Element Weights 5 2.1 The Stepwise Regression Approach . 5 2.2 Preparing the Link Resolver Environment . 7 2.2.1 Automated Testing Using a Batch Process . 7 2.2.2 Checking for Full Text Targets 7 2.3 Selecting OpenURLs for the Stepwise Regression 8 2
8、.4 Conducting the Tests 8 2.5 Calculating Optimal Element Weights . 10 Section 3: Generating the Completeness Index 12 3.1 Reviewing the Theory Behind the Completeness Index . 12 3.2 Selecting OpenURLs to Include in the Calculation . 12 3.2.1 Source . 12 3.2.2 Genre . 12 3.2.3 Quantity . 12 3.2.4 Ti
9、me Period 12 3.3 Preparing for the test 12 3.4 Calculate Completeness Scores . 15 3.5 Generate Completeness Index 15 Section 4: Analyzing Results and Affecting Change 17 4.1 Interpreting Completeness Index . 17 4.1.1 Meaning of the Completeness Index Value . 17 4.1.2 Using the Completeness Index . 1
10、7 4.1.3 Analyzing the Completeness Index for a Problem . 17 4.1.4 When a Content Provider is Frequently Missing Core Data Elements 18 4.2 Additional Items to Check . 18 Appendix A A Method for Checking Link Resolution for Purposes of Calculation of the IOTA Completeness Index 20 Bibliography 21 Impr
11、oving OpenURLs Through Analytics (IOTA) iv NISO RP-21-2013 Improving OpenURLs Through Analytics (IOTA) NISO-RP-21-2013 v Foreword About this Recommended Practice NISOs OpenURL Quality Committee, which later became known as IOTA (Improving OpenURL Through Analytics), was given a charge that contained
12、 the following problem statement: The OpenURL standard is widely deployed technology to facilitate linking to resources across the library supply chain. The OpenURL-formatted URL carries the data about an item to the link resolver of the library. The resolver compares the metadata embedded within th
13、e OpenURL with what is held in the librarys collection and presents the available options in a results page. For a book, there is usually a link to the librarys catalog card; for an article, ideally this is a link directly to the full text of the article. At a typical academic library, thousands of
14、OpenURL requests are initiated by patrons each week. The problem is that too often these links do not work as expected, leaving patrons frustrated by a lower than desired quality of service. Periodically, mention is made in the library literature of problems with OpenURL linking, but since the OpenU
15、RL standard was introduced a decade ago, no systematic method has been designed and carried out to benchmark it. This work is intended to fill the gap. As part of its work, the committee analyzed millions of OpenURLs and developed the notion of a Completeness Index as a means of quantifying OpenURL
16、quality. The committee found that there was a pattern to the failures in OpenURLs. OpenURLs that included certain data elements performed better than others missing one or more of those elements. The Completeness Index was developed as a method of predicting the success of OpenURLs from a given prov
17、ider by examining the data elements that provider includes in the OpenURLs from its site. The index acknowledges that certain data elements are more critical to success than others by giving these elements a higher weight. Technically speaking, a Completeness Index for a given provider is the averag
18、e Completeness Score for all OpenURLs being analyzed from that provider. The Completeness Score is the sum of the weights for each of the core elements included in the OpenURL divided by the total potential score. If all core elements were included the score would be 1. The theory behind the Complet
19、eness Score and Completeness Index was validated by separate tests where thousands of OpenURLs were tested for successfully creating a link to full text and this success was correlated against the Completeness Score. These independent tests by both EBSCO and Serials Solutions validated the concept.
20、The tests also confirmed a suspicion that creating a single universal set of element weights is not practical. Differences in linking environments and link resolver technologies affect the importance of certain elements. For example, a link resolver that does no enhancements to the data provided in
21、the OpenURL will have a high failure rate if no ISSN is provided; however, the typical commercial OpenURL link resolver can use the journal title to look-up the ISSN and thus the absence of the ISSN is less critical. And to use another example, a link resolver that is able to use volume, issue, auth
22、or, and article title to look-up an article in a service like CrossRef will be more forgiving to a missing Start Page in the OpenURL than one that does not offer such article metadata enhancement. Coming out of the work of the IOTA committee was the recommendation that link resolver providers (or ot
23、hers interested in an OpenURL linking environment) introduce the notion of the Completeness Index with its constituent Completeness Scores to introduce a quantitative mechanism for evaluating link quality from different providers. Because the element weights are environment-dependent (as illustrated
24、 in the ISSN and Start Page examples above), the committee has created this Recommended Practice to serve as a guide for: calculating the element weights, generating a Completeness Index, and analyzing the results and affecting change. Improving OpenURLs Through Analytics (IOTA) vi NISO RP-21-2013 N
25、ISO Discovery to Delivery Topic Committee Members The Discovery to Delivery (D2D) Topic Committee had the following members at the time it approved this Recommended Practice. Tim Babbitt Cambridge Information Group Tim Shearer University of North Carolina Chapel Hill Libraries Pascal Calarco, Co-cha
26、ir University of Waterloo Library Lucy Harrison, Co-chair Florida Virtual Campus Peter Murray Lyrasis Chris Shillum Reed Elsevier Jeff Penka OCLC Online Computer Library Center NISO IOTA Working Group Members The following individuals served on the NISO Improving OpenURLs Through Analytics (IOTA) Wo
27、rking Group, which developed and approved this Recommended Practice: Adam Chandler, Chair Cornell University Library Clara Ruttenberg University of Maryland Libraries Rafal Kasprowski Rice University Fondren Library Elizabeth Winter Georgia Tech Library Susan Marcin Columbia University Libraries Jim
28、 Wismer Thomson Reuters Oliver Pesch EBSCO Information Services Aron Wolf Serials Solutions Acknowledgements The IOTA Working Group wishes to thank Phil Davis for the valuable assistance he provided, guiding us in our stepwise regression method. Trademarks, Services Marks Wherever used in this recom
29、mended practice, all terms that are trademarks or service marks are and remain the property of their respective owners. Improving OpenURLs Through Analytics (IOTA) NISO-RP-21-2013 1 Section 1: Introduction 1.1 Purpose and Scope These recommended practices are intended to assist information professio
30、nals involved in managing OpenURL linking environments in improving OpenURL linking by providing them with the tools necessary to measure OpenURL quality. The focus of the IOTA (Improving OpenURL Through Analytics) Working Group is the quality of the metadata that is passed to the link resolver from
31、 the OpenURL source. The quality of the data in the link resolver knowledge base itself is outside the scope of IOTA; this is being addressed through the NISO KBART initiative. Also outside of the scope is the quality of the linkages between link resolvers and full text content providers. (To our kn
32、owledge, this part of the OpenURL quality problem is not being addressed by any organized initiative.) This Recommended Practice is focusing on a specific genre of OpenURLsthose intended to provide access to journal articles. The information provided in this Recommended Practice could easily be adap
33、ted for other genres such as book and book chapters. 1.2 Terms and Definitions The following terms, as used in this recommended practice, have the meanings indicated. Terms in boldface in a definition indicate the term is also defined in this section. Term Definition A the higher the value the more
34、important the element. Enhanced OpenURL The set of OpenURL elements available after the link resolver enhancers have been run on the incoming OpenURL. Typically an Enhanced OpenURL will have more data elements than the original OpenURL. enhancer An automated process that enhances an OpenURL by using
35、 internal and external sources to supplement the OpenURL data elements. An example would be an enhancer that looks up article-level metadata from CrossRef using the DOI presented on an OpenURL. fail Describes the state of an OpenURL that does not generate any item-level links to full text items. See
36、 also success. NOTE: Within the context of this recommended practice the following are not considered links to full text items: links to a journal homepage, the table of contents for an issue, or a search page at the vendor site. full text target link A link to the complete textincluding all referen
37、ces, figures, and tablesof an article on the target site. If the referenced item is a journal article, the full text article link would direct the user directly to that article at the content providers site without requiring further navigation or searching. Improving OpenURLs Through Analytics (IOTA
38、) NISO-RP-21-2013 3 Term Definition knowledge base A database used with OpenURLs that contains information about what targets are available to users of the link resolver. Data within the knowledge base includes but may not be limited to link syntaxes and holdings with coverage details. The link reso
39、lver environment for a given institution can be customized to reflect that institutions collection and to only provide links to targets to which the library subscribes. link resolver Technology that controls the linking between sources and targets. The link resolver accepts and deconstructs the Open
40、URL describing a content item from a source and uses its knowledge base and associated programs to determine full text and other targets appropriate for the user and create predictable links to these. The role of the link resolver is context sensitive linking to the appropriate copy of a content ite
41、m. OpenURL As defined in ANSI/NISO Z39.88-2004, The OpenURL Framework for Context-Sensitive Services, a URL designed to transport metadata and thus enable linking from information resources such as abstracting and indexing databases (sources) to library services (targets), such as academic journals,
42、 whether online or in printed or other formats. The linking is mediated by link resolvers, or link-servers, which parse the elements of an OpenURL and provide links to appropriate targets available through a library by the use of a knowledge base. Source: Wikipedia boldface notation added PubMed ID
43、A unique identifier assigned to a record in the PubMed databases (www.pubmed.org). A PubMed record typically describes an article or other content item. Using a PubMed ID it is possible to look-up descriptive metadata of the content item through a free online service provided by the National Library
44、 of Medicine. referrer The identity of a website or discovery platform that is the source of the OpenURL. In an OpenURL this is represented as the SID (Source ID) or the Referrer ID. In some contexts, referrer is used interchangeably with source. required data elements The data elements that must be
45、 available to create a successful link to an OpenURL target. For example, if a publishers link to a full text article is constructed from ISSN, Volume, Issue and Start Page values, then these fields are considered required data elements since the link will fail to access the full text if any one of
46、them is missing. Improving OpenURLs Through Analytics (IOTA) 4 NISO RP-21-2013 Term Definition source The website where the user discovered an item of interest and from which the OpenURL was initiated. The source is identified by the Referrer ID. In some contexts, source is used interchangeably with
47、 referrer. success Describes the state of an OpenURL that is able to generate an item-level link to one or more full text items. The link generated is intended to populate the link resolver menu; however, the designation of “success” does not guarantee a user will get to the full text item if the li
48、nk is followed all the way to the full text content providers site. See also fail. Success Score A value attributed to an OpenURL based on its success in generating a link to one or more full text items. If the OpenURL generates a link to a full text item, it is given a Success Score of 1; if not, t
49、he Success Score is 0. target The website where the full text resides and where the user will be linked to view the item via the link resolver. Example targets could include content in publisher platforms, institutional catalogs or repositories, and content gateways. Improving OpenURLs Through Analytics (IOTA) NISO-RP-21-2013 5 Section 2: Determining Optimal Element Weights This section of the Recommended Practice describes the steps to follow to create Completeness Index Element Weights that have been optimized for a particular linking environment. It starts with the theory
