Serials rationalization at Oxford University Library Services using Ulrich ’ s Serials Analysis System

Oxford University Library Services (OULS) brings together collections of 36 centrally funded libraries in Oxford. The integration of collections is a lengthy and very complex process. OULS has chosen Bowker's Ulrich's Serials Analysis System (USAS) as the most suitable tool to analyse and rationalize its serials collections. This article describes the background to the project and the implementation of USAS at Oxford University. Gathering and uploading serials information from 36 different university libraries into one system proved to be a challenging task. The internal issues universities may face as they start to gather consistent serials information are discussed, and the problems arising from uploading serials data into USAS are considered. Despite these drawbacks OULS is convinced that the new analysis system from Bowker is a potentially very powerful tool for local, regional and national collection management.


Background
The University of Oxford's many libraries contain the largest and most diverse collections for the support of teaching and research in any institution of higher education in the United Kingdom.Its library collections as a whole are world-class.Because its principal library, the Bodleian, has been a legal deposit library for almost 400 years, members of the University and scholars from far and wide expect to satisfy a very high proportion of their library needs somewhere within Oxford's libraries.
The libraries that make up Oxford University Library Services (OULS) contain more than 11 million volumes; and, if periodical parts are included, OULS adds to stock an average of well over 1,000 items per day throughout the year.It is an integral part of OULS' mission to develop and exploit these collections for the benefit of scholarship generally, and to retain and preserve the vast majority of them for posterity.
At present, OULS consists of 36 libraries and was formally established, with the approval of the University, in February 2000.This date marked the start of the integration process of centrally funded libraries.Librarians will, however, recognize that bringing libraries together on paper does neither automatically lead to the harmonization of policies and practices, nor to the integration of collections.Integration and rationalization of collections is a lengthy process which requires major efforts from both library management and staff.
The review and co-ordination of current serials holdings was seen as a major step towards rationalizing acquisitions and achieving better budgetary and management control over collections across OULS.In August 2004 the project SCORPIO (Serial Co-ordination Project in Oxford) was launched in order to: Despite these drawbacks OULS is convinced that the new analysis system from Bowker is a potentially very powerful tool for local, regional and national collection management.

ALICE KELLER DAG MARTINSEN
■ explore ways of integrating print and electronic resources in the most cost-effective manner ■ ensure that legal deposit material is used to best advantage.
It is recognized that integration of collections is a three-phase process.In the first phase comprehensive and consistent collection management information must be gathered and analysed (see Figure 1).This data will form the basis for management decisions and policy formulations during the second phase.Changes of policy can be implemented during the third phase of the integration process.If, as in Oxford, one of the aims of an integration process is the rationalization of serials holdings, the first phase would include identification of the exact level of duplication.During the second phase library managers are challenged to draw up policies on how to deal with planned and unplanned duplication, and on the integration of print and electronic holdings.It is only when these two steps are completed that selective cancellations can be made and holdings transferred from one library to another.
The quality, transparency and flexibility of data gathered during the first phase are crucial to the successful integration of collections as a whole.This article focuses on the method and system OULS has chosen to complete this first phase.

Implementation and gathering information
OULS selected Ulrich's Serials Analysis System (USAS) as the most suitable electronic tool for the SCORPIO project.This system is an add-on to the online version of Ulrich's Periodicals Directory, a database already available online at Oxford University 1 .USAS is the only known collection analysis tool to have been developed specifically for analysing serials holdings.It is relatively new to the UK, but a number of academic libraries in the United States have already used the system for various purposes including rationalization projects.The A presentation on USAS from Bowker is available at: http://www.ulrichsweb.com/ulrichsweb/analysis/help/findoutmore.asp.
Virtual Library of Virginia (VIVA) has used it for its Serials Audit Project 2 , and it is in use at Northern Illinois University Libraries.
USAS works on the basis of holdings information uploaded in the form of lists.As the project got under way an Excel spreadsheet containing all current serials subscriptions, purchases, regular donations and exchanges was requested from every library.The list would ideally include ISSN, title, vendor, price, and holdings information (see Figure 2).
In most cases the project assistant made a personal visit to each library in order to discuss the best way of producing the relevant data.What can often be a problem in large, diverse academic library systems such as OULS is that some libraries will have in-house or manual systems for registering serials.If a library already has a comprehensive list of their current holdings in an electronic format -a database or a spreadsheet -the work can be fairly easy.In some cases it will be possible to extract the required data from the automated library catalogue.In other cases much or all of the required information exists only in manual registers or files.Transferring this into spreadsheets can be an extremely time-consuming task.The success of the project depends partly on the quality of the information provided by libraries.The spreadsheets also remain important after the lists have been uploaded into USAS, as libraries may choose to subscribe to USAS for a limited time only.
In addition to the 36 lists from individual OULS libraries, a list was compiled containing all e-journals available across the university.Fortunately all e-journals are registered centrally in TDNet, which made this task quite straightforward.It was decided that serials received by legal deposit would be dealt with separately, as these titles are not registered in the same way as regular subscriptions.As no list of legal deposit titles was available, OULS asked Bowker to supply a customized data extract of periodicals with UK or Ireland as country of publication.Spot-checks between this extract and serials actually received by legal deposit were then carried out.Initially the correspondence between the list from Bowker and legal deposit holdings seemed unsatisfactorily low.However, restricting spot-checks to academic titles significantly improved this correspondence.Eventually, lacking any more accurate data, OULS decided to use the UK and Ireland list provided by Bowker to evaluate the level of duplication of academic serials published in the UK and Ireland.

Uploading lists into the analysis system
Based on the ISSN, USAS 'matches' the contents of the uploaded lists with the corresponding information in Ulrich's Periodicals Directory.Once uploaded, library lists are presented on the screen in a format quite similar to the format of the directory (see Figure 3).By clicking on one of the titles, the information in the database associated with that title becomes readily available.In OULS, the most useful of these external data are subject category, Bowker list price and publisher.Other useful data elements show whether the title is refereed and if it is available electronically.Clearly, direct access to this information from an online list of a library's own titles is one of the advantages of using USAS.An additional bonus is that the system will report on any ISSNs or titles on a library list that have changed or ceased, based on the title history information recorded in the directory.These details are not automatically changed by USAS when the list is uploaded; updates need to be done manually.Bowker argues that this is an advantage; it keeps libraries in control of the information that is updated.Automatic updates could create a conflict between the USAS record and the library's other in-house records, without the library noticing.In fact, libraries may find that the most straightforward way of carrying out updates to the list in USAS is to make the changes on the original spreadsheet before uploading it again.Although it may seem cumbersome, this at least ensures that the change is recorded both online and on the original spreadsheet.
The whole process of uploading depends entirely on USAS matching up ISSNs on the library lists with ISSNs in the directory.When so much

Serials -18(1), March 2005
Alice Keller and Dag Martinsen Serials rationalization at OULS Figure 3.One library's list of titles has been uploaded into USAS.In addition to the three 'free' fields 'Usage Stats','Notes 1', and 'Notes 2', this view indicates whether the titles are refereed, abstracted and indexed, available electronically, etc. Clicking on a title will provide additional information.Note that USAS has automatically supplied the full title of the journal 'Brain' (compare with Figure 2).
depends on the accuracy of one piece of information, some disadvantages of the system soon become apparent.Academic librarians will be aware that many serials do not have ISSNs, and also that a large database such as the directory can never claim to be entirely comprehensive.For instance, the directory does not, in principle, contain local government publications.So what happens when an ISSN is uploaded that does not find a matching ISSN in the directory?If a title does not have an ISSN, or has an ISSN that is not recorded in the directory, the title will be classified as a 'nonmatch'.Ceased titles are also classified as nonmatches.It goes without saying that in some collections with many highly specialized or ephemeral titles the percentage of non-matches can be substantial.Even though non-matches will be included in the online lists in USAS, the system cannot generate statistics on these non-matches.A few OULS libraries have around 50% non-matches.
Overall, the average level of non-matches in OULS appears to be approximately 22%.Of course, not much can be done about titles having ceased, or failing to have an ISSN.A more unsatisfactory situation is the relatively high number of titles in the directory for which no ISSN has been supplied.As all matching is done by ISSN a library's list cannot match onto these titles even when it includes this reference.It is understood that the inclusion of these titles' ISSNs will depend on increased research on the part of Bowker.Bowker has promised that if libraries can present lists of directory titles found to be lacking ISSNs, an effort will be made by their editorial team to prioritize research on these.

Generating reports and statistics
USAS can be used to analyse and compare serials holdings of individual libraries, of groups of libraries, or of consortia.It can also be used to compare reported holdings with the entire content of the directory (approximately 175,000 active titles), or with a 'core' set of academic titles (approximately 50,000 active titles) defined by Ulrich's3 .All reports are downloadable in spreadsheet format and can then be utilized and manipulated for a range of purposes.
The analysis system enables an academic library system to better understand subject coverage or serials expenditure, highlight strengths and weaknesses in the current collection, and establish areas and level of duplication.A 'comparison report' facility makes it easy to identify duplicates across a selection of libraries by counting the number of occurrences of every title (see Figure 4).This is one of the major advantages of USAS.Users can choose which lists to include in each comparison report.In addition, it is possible to choose to include details associated with the title in the directory, such as subject category and Bowker list price.Once downloaded in a spreadsheet, the titles can be sorted according to the number of occurrences USAS has identified.
As already mentioned, statistics cannot be generated on non-matches.However, non-matches are included when USAS generates reports that count duplicates.This is a useful feature, despite the fact that additional data such as subject category and list price cannot be retrieved due to these titles' status as non-matches4 .
A more serious problem concerns USAS' ability to generate comprehensive analyses by subject category.In theory, USAS enables libraries to analyse their holdings in terms of the Ulrich's subject categories.In the directory, each title has one or more main subject categories (of which there are approximately 200) or subcategories (of which there are almost 1,000).However, when the analysis system generates statistics, it can only associate each title with one of its subject categories.Accordingly, if a title has more than one subject category, it will only be included under one of these in the statistics.This may not be convenient for certain kinds of analyses.It has been pointed out that since there are many journals which can be placed under more than one subject heading, the analyses provided by USAS must unavoidably be incomplete and even misleading 5 .Even though Bowker claims that only 5-6% of all titles in the directory are indexed under more than one subject category, this is not an insignificant number.Bowker has recognized this problem and is investigating various solutions, especially with regard to titles that are markedly interdisciplinary.A related problem identified by OULS concerning subject categories is that the specific needs of academic libraries are only partly catered for.While the categories within science and medicine seem to reflect 'academic' subjects quite well, Ulrich's humanities categories are sometimes too broad and are not readily transferable to an academic library environment.One example would be the subject heading 'Literature' whose subcategories include 'Adventure and Romance' and 'Poetry'.

Conclusion
Ulrich's Serials Analysis System is a relatively new analysis tool for libraries.The backbone of USAS is the information provided in Ulrich's Periodicals Directory, a product which enjoys a very good reputation in both academic and public libraries.USAS enables libraries to exploit this high quality, aggregated serials information and use it in their decision making processes.This combination makes USAS a potentially very powerful tool for local, regional and national collection management.
The experiences of OULS show that it is not a trivial undertaking to gather and upload serials information from 36 different university libraries into one system.On the one hand it has proved difficult to obtain automated lists of current subscriptions from all libraries.Even in an academic environment many libraries still operate with manual registers.Some of the problems pointed out here have since been rectified, while others persist.

■
gather and analyse serials management information ■ identify the level of duplication across the OULS ■ investigate ways of reducing serials expenditure ■ prepare transfer of subscriptions between OULS librariesSerials rationalization at OULS Alice Keller and Dag MartinsenSerialslibraries in Oxford.The integration of collections is a lengthy and very complex process.OULS has chosen Bowker's Ulrich's Serials Analysis System (USAS) as the most suitable tool to analyse and rationalize its serials collections.This article describes the background to the project and the implementation of USAS at Oxford University.Gathering and uploading serials information from 36 different university libraries into one system proved to be a challenging task.The internal issues universities may face as they start to gather consistent serials information are discussed, and the problems arising from uploading serials data into USAS are considered.

Figure 1 .
Figure 1.Diagram showing the steps of uploading individual libraries' lists, the matching process in USAS, and the subsequent downloading of comparison reports and statistics.

Figure 2 .
Figure 2.An example of a list in spreadsheet format provided by a library.USAS allows users to upload three 'free' fields in addition to ISSN and title.OULS chose to include name of vendor, price (or 'DO' if a donation), and holdings.The file needs to be converted to plain text format before being uploaded into USAS.To follow the title 'Brain' on its way through USAS, see also Figures 3 and 4.

Figure 4 .
Figure 4.A sample of a report that identifies duplicates within a chosen number of holdings lists, downloaded in a spreadsheet format.One can choose to include additional information such as list price.Along with the total count of occurrences of a title ('found in'), users can see the names of the libraries in which each title has been found.