Because no library or library-relatedgroup had previously attempted to compile an authoritative list of organization names within an ERM tool, NCSU Libraries’ name authority project had to be developed in-house from scratch. The continuing and electronic resources librarian, Jacquie Samples, working as part of the E-Matrix product committee, developed a model for creation of the authority through a pilot project. By taking a small subset of organizations from the ERM data and assigning them authoritative names, she determined important specifications for the larger project, including a preliminary analysis of how assigning authorities would affect the library’s ERM data, an estimate of the project’s timeline, and expectations for the type of work that would be required to complete the authority.

The name authority pilot project assigned authoritative headings to 483 vendor names extracted from the library’s integrated library system (ILS) in November 2006. Vendor names were chosen because they were deemed most useful for developing the licensing module, which had been identified as a priority around the same time. Working from a spreadsheet, the continuing and electronic resources librarian evaluated each vendor and chose a preferred name. Altogether, 444 authoritative vendor names were selected on the basis of predominant usage within the library community and assigned to the group of original names. Thirty-nine names (8 percent of the original sample) were variant names that could be linked through the use of an authoritative heading. Cross references were also established during the pilot project to represent business relationships between vendors—for example, companies that had been purchased by larger entities—as well as variant names not native to the data set. Additional roles were also noted for many of the entries, although they could not be incorporated into the ERM data at that time. The prevalence of variant names, complex business relationships, and multiple roles all provided strong evidence that the entire E-Matrix data set required authority control of organization names.

At the conclusion of the pilot project in April 2006, Samples estimated that approximately forty hours had been spent assigning names and cross references, as well as making notes about interesting or unusual circumstances. Using the estimated number of organization names in E-Matrix at that time (about seven thousand) she determined that to assign authoritative names to the entire data set would take 580 hours—about three and a half months of full-time work for one person. Given the project loads of the librarians in the technical services departments at NCSU Libraries, the project likely would take the better part of a year for one librarian to complete.

In addition to the sheer amount of time that would be required to create a fully functioning name authority, the pilot project also revealed that much of the work would be composed of manual functions, including collocating names that represented the same entity and determining authoritative headings for each group of names using library and serials industry resources. These tasks would require the expertise of a library staff member familiar with the nature of serials and common library-based information sources. To determine authoritative names, the individual performing the work would require a strong sense of the relationships that exist between libraries and organizations. The librarian also would need the ability to follow a set of guidelines that combine accepted authority practices with the needs of the university and the library. The guidelines would not be prescriptive, and the selection of names would require a strong element of judgment and a fluid and intuitive use of available tools. The ultimate goal would be to determine headings for each organization that best fit with the library’s existing practices and the overall structure of the authority. In light of the nature of the name authority work, the possibility of selecting names in an automated fashion was rejected because it was uncertain if an automated system could effectively make the judgments needed to choose correct authoritative names.

The demonstrated importance of the name authority project, along with the volume and intensity of the work it would require, convinced the E-Matrix committee and library administration that the authority tool deserved top priority. The continuing and electronic resources librarian and a programmer from NCSU Libraries’ IT department were assigned as part of their E-Matrix work the tasks of designing a system to store and manage name authorities and of using that system to assign authoritative names to each organization in the licensor, provider, publisher, and vendor roles within E-Matrix. An NCSU Libraries Fellow, Kristen Blake, was appointed to aid in the determination and assignment of authoritative names. Together, this group made up NCSU Libraries’ E-Matrix name authority team. With the support of library administration and increased resources, the name authority had moved into the realm of the possible.

Before the name authority project could be fully implemented, a module had to be designed within E-Matrix to manage preferred names and other authority data. The name authority team created a framework that could be integrated into the ERM system functions that had already been designed while remaining true to the vision of name authority established by the librarian working on the product’s conceptual development.

The original plan for the organization name authority within E-Matrix envisioned a record-based structure to link together related organization names and store descriptive data about each organization. Each authorized name would be stored on a record and connected to the work-level records of all resources featuring one of its variant names. The authorized name record would have the ability to store data relevant to the authorized heading as well as its variants. Variant names would remain in place on the work-level records to which they originally belonged, and those records would be used to store data unique to each variant. Additionally, the variant names on each work-level record would be assigned one or more of the roles available in E-Matrix: licensor, provider, publisher, and vendor. License records would also be linked to an authoritative name record through the work-level record and the nonauthoritative names associated with that work. All of these data—titles of works, unauthorized names, and organizational roles—would loop back and display on the authoritative record. Figure 2 shows the relationships between records that link name authority headings to work-level records in E-Matrix.

In addition to establishing associations between resources and organizations, the initial vision of the name authority tool also included the ability to store useful data on organization records. Authoritative name records would serve as the primary storage location for business history notes, internal remarks, vendor contact information, product trial details, and other information as needed. Nonauthoritative name records could also be used to store notes specific to a single variant or imprint. By incorporating a detailed record structure into the name authority, the library hoped to mimic the structure of established name authorities that serve as repositories of historical and local information, as well as guides for consistent and accurate data creation.

The realities of implementing E-Matrix forced the library’s programmers and planners to apply a phased approach to the design of the name authority application. In the interest of getting the project started as quickly as possible, some of the features in the original proposal were assigned to later phases, and a name authority system was designed that included essential functions but did not yet incorporate more robust design features. The system links organization name data through unique identifiers in a relational database. A member of the name authority team can assign a name authoritative status, which places a property in the database record for that name, indicating that status. Once a nonauthoritative name is associated with an authoritative heading, database records for nonauthoritative names and their roles link to that name algorithmically. When an authorized heading has been assigned to a certain organization name, all future instances of that name will be automatically subsumed into the existing hierarchy. This essential feature will decrease the amount of maintenance necessary once the authority project has been completed.

The inclusion of a formal record structure for storing historical and internal data was also transferred to a later phase in the design of the authority tool. In its current incarnation, instead of using formal records like those a cataloger might find familiar, the names within the authority function more like related nodes without a formal record structure. E-Matrix searches the authority database on the fly and, when data intersects, the relationships between names and roles are displayed within the E-Matrix user interface or on a report.

Within this design scheme, the system of roles assigned to each organization acts as a temporary substitute for the hierarchy of business relationships envisioned for the authority. While the authority team cannot currently assign, for example, Elsevier as the current owner of Pergamon Press, the roles assigned to each organization can be manipulated to produce reports that would reflect that same relationship. Because Pergamon may be assigned to a resource in the publisher role, and Elsevier assigned to that same resource in the provider role, a report listing all titles for which Elsevier is the provider along with their additional roles will indirectly display the relationship between Elsevier and publishers, like Pergamon, that it has acquired.

The current system allows E-Matrix to accomplish the primary task of collocating all resources associated with a certain organization, though the lack of formal records delays the library’s goals of using its ERM system to capture the syndetic structure of business relationships and use the ERM as a primary location for storing these data. The eventual addition of that functionality to E-Matrix will allow the name authority project to move forward and more strongly fulfill the local needs identified early in the development of E-Matrix.

Once this structure was in place, the next step in the implementation process was the design of a convenient and intuitive way to assign and manage authoritative names. Since the start of the project, the name authority team has worked to design an interface within E-Matrix that would allow the authority to be easily accessed and manipulated. For this phase of the project, the technical services librarians contributed ideas for the design and functionality of the interface, while the programmer translated this vision into a series of interfaces, each enhanced and refined with feedback based on actual use.

The first authoritative organization names were initially stored in a local database because the user interface for the E-Matrix authority had not been completed at the start of the project. Instead, the librarians assigning authoritative names used a Microsoft Access database for storage as they began working through a list of 7,858 publisher names culled from the library’s holdings in the SFX (Ex Libris’ link resolver) knowledgebase. These publisher names were chosen not only because they provided sufficiently complex test data for the developing interface, but also because they represented the names most desperately in need of authority control. (Publisher data from the library catalog were not initially included because much of it was duplicated in the SFX data, and the remaining publisher names could be controlled later by building on the headings already determined during the SFX phase.) The librarians evaluated each name and collocated it with others representing the same publisher, recorded their decisions in the database, and made notes specifying the justification for each decision and any problems that might need to be investigated when the E-Matrix interface became functional. This evaluative period resulted in the determination of authoritative headings that could be used later, as well as suggestions for the design of the integrated interface planned for E-Matrix.

Throughout this early stage, designing a name authority interface shared top priority status with assigning names. Based on input from librarians who had tested the interface, the E-Matrix programmers produced a rudimentary beta interface, which was then adopted on a trial basis as part of the process of assigning names. The interface allowed the name authority team to select, on one screen, an authoritative organization name, as well as a group of names that should fall under its domain, and record that relationship within E-Matrix. This interface allowed the authority team to transition from recording their decisions only in the local database to actually entering them into E-Matrix, where they could be used experimentally by library staff. The beta interface had many limitations, however, and the local database was maintained as a backup and a place to record comments and problems.

The authority team continued working within E-Matrix, and the team’s observations helped the programmers develop an improved name authority interface, which was released with E-Matrix 1.0 in December 2007. This interface allows names to be assigned using a simple four-step process. In step 1, the browse or search feature is used to identify and select names that will be collocated under a common authorized heading. Figure 3 shows the results of a search for “Duke University” and lists the names that need to collocated.

In step 2, an authoritative name can be chosen from among those selected or a new name entered manually. In step 3, the user can perform a final review of selections and assignments and submit them. The fourth and final step confirms the assignments made and offers the user a link to begin the process again. Steps 2–4 are shown in figure 4.

The new interface made the team’s work easier because it included expanded display options that clarified whether an organization already had an authoritative heading assigned to it and what that heading was. The authoritative relationships were also displayed in more places within E-Matrix, making the product useful as a source of organizational data for the authority team and all library staff using the ERM. In comparison with the beta version, the first production interface was more intuitive, incorporating browse functionality, a cleaner design, and the ability to correct errors by reassigning authoritative headings. With the addition of a notes feature in a future release, the team will transition their workflow completely into the ERM system.

At the same time that an interface was being designed, the continuing and electronic resources librarian was working to create the practical guidelines and specifications needed to choose authoritative names and assign them to organizations within E-Matrix. The most important deliverables were guidelines that ensured accuracy and consistency in the selection of authoritative names and the identification of organizations that will fall under them and a set of tools to help apply those guidelines.

Creating clear, accurate authoritative headings that were useful for the NCSU Libraries staff has been the primary consideration in deciding how names should be assigned within E-Matrix. In the interest of local policies, the first step taken in establishing name selection guidelines was to consult the library’s collection managers. Because they were the people who would be making use of the authority data most frequently for collection evaluation, the authority needed to reflect their preferences and standards. The collection managers indicated that their chief goal was to preserve the organization name that most directly reflected the intellectual content of a work. Most commonly, this directive affected the assigning of authoritative publisher names, which often are more directly tied to specific content areas than a vendor or provider name. Thus, to maintain that connection, the original publisher of a title is almost always chosen as the authoritative heading, even if that publisher has since merged with or been acquired by another entity. In other words, the “statement of responsibility” takes precedence over any current business arrangements. Journals that have been published by Academic Press, for example, are kept under the Academic Press heading, even though Academic Press has long been an imprint of Elsevier.

In many cases, publisher statements that have been imported from the library catalog are long and convoluted. Large societies often issue publications on behalf of smaller societies and, in these cases, collection managers presume that the larger society is acting as a kind of benefactor, while the smaller society represents the creator of the journal’s content and, therefore, the better choice for an authoritative heading. For example, a publisher statement reading, “Published for the International Union of Biochemistry by Elsevier” would be assigned the authoritative name “International Union of Biochemistry” because that group entity is presumed to be responsible for the biochemistry-related content of the titles associated with the organization. Elsevier’s role in the creation of the material is not lost because it can be assigned to those resources as a provider and licensor.

In another common scenario, groups of small societies publish a title jointly, and the publisher statement contains the names or two or three discrete entities with each given equal weight. E-Matrix’s current functionality allows for only one authoritative organization heading to be assigned to each work. In these instances, every effort is made to determine which organization is chiefly responsible for the work in question and to assign authority accordingly. For example, considering the publisher statement “American Society for Environmental History and Forest History Society,” the librarian determining the authoritative name must conduct a thorough investigation that includes identifying and viewing publications linked to this statement, then reading front matter and publisher information to determine which publishing group is the primary contributor to the content of the title.

Because NCSU Libraries holds a fair number of foreign language titles, the formatting of the authoritative names for these titles is also important to the library’s collection managers. They requested that all foreign language authorities using Roman script remain in the original language. Authority headings for titles using non-Roman scripts have been translated into English rather than transliterated. E-Matrix does not allow for the inclusion of non-Roman characters, and translations were determined to be clearer and easier to assign than transliterations.

Beyond these few specific requests, NCSU Libraries’ collection managers and the E-Matrix development team felt comfortable relying on traditional library resources for the determination of names. The Library of Congress Name Authority File (LCNAF) (http://authorities.loc.gov) has been used whenever possible as a source for preferred names, as long as they do not conflict with local customizations. If no heading is available from the LCNAF, trustworthy serials databases such as Ulrich’s Periodicals Directory (http://ulrichsweb.com) and the ISSN Portal (http://portal.issn.org) have been used as secondary sources. NCSU Libraries’ sources correspond almost exactly to those made by Connaway and Dickey of the OCLC’s monograph-focused publisher name server.¹⁸ That project uses the LCNAF as the chief source of authorities, followed by Books in Print (http://booksinprint.com) and the International ISBN Agency (http://isbn.org), monographic counterparts of Ulrich’s and the ISSN Portal, respectively.

Unlike OCLC, which has formalized its choice of sources for name selection, NCSU Libraries has the flexibility to put its local needs first. Rather than make a hard and fast rule about the order of sources consulted, the librarians at work on the name authority project have the freedom to make decisions on the basis of what will best fit the library’s needs. The decision to disregard name changes, mergers, and other business-related changes until richer syndetic functions can be incorporated into E-Matrix illustrates a fundamental application of the local needs principle. Another example is the decision to apply Anglo-American Cataloguing Rules, 2nd ed., rules for formatting of complex names to all headings, regardless of their source.¹⁹ This practice ensures consistency and, just as importantly, enhances the browsablity of organization data. When all departments, research centers, publishing arms, and other offshoots of a major institution are grouped together using consistent formatting, finding and grouping resources emanating from a particular group becomes easy, while still retaining more detailed information about the subgroups involved in its production. E-Matrix’s browse features, as well as the sorting capabilities of the reports module, are greatly enhanced by this practice.

Librarians assigning authoritative names are also encouraged to use their knowledge of local practices to make case-by-case exceptions when necessary. Often, these types of decisions are used to resolve small quirks that might never be addressed by a stricter set of rules. For example, the authoritative name chosen for the Institute of Electrical and Electronics Engineers is IEEE, even though the LCNAF has authorized the full version of the name. While working through the list of organizations, the continuing and electronic resources librarian recognized, for example, that the acronym IEEE was simply more recognizable to library staff than the organization’s rarely used full name. The full version of the name will not be lost from the name authority data, however, because it is stored as a searchable cross reference. On both the conceptual and practical levels, the flexibility to tailor the name authority specifically to NCSU Libraries’ interests has been essential to its success.

Equally important as the authoritative sources are other sources that provide insight into unique or problematic names not addressed in the traditional library databases. Many of the publisher names came in the form of obscure initialisms, and, in these cases, the website Acronym Finder (http://acronymfinder.com) suggested leads that eluded typical research sources. In general, the Web search engines proved vital to researching the kind of obscure names that are routinely not found in the LCNAF. Often, use of E-Matrix itself was necessary to find the name of one or more journal titles associated with a specific publisher, and then a Web search engine was used to trace those titles back to the primary source. Viewing the publisher name in a table of contents or on an authoritative website provided an extra level of confidence in the decision-making process for publishers not found in an established source.

The combination of an E-Matrix interface designed with user input and a set of fluid guidelines for name selection has made the day-to-day work of assigning name authorities a smooth and intuitive process. While unexpected and unusual names can slow what has become a speedy process, these are usually resolved through discussion and attention to the established guidelines and the library’s needs. Again, the flexibility of the assigning process and the value placed on librarians’ judgment have been essential to the implementation of this original and complex project.

The tools and procedures needed to create a functional organization name authority within E-Matrix were put into place by summer 2007. As of this writing, practical implementation of the name authority project has been underway for nearly a year. During that period, the NCSU Libraries has focused on the establishment of authoritative names for organizations stored in E-Matrix and in the use of those names to enhance data integrity, licensing procedures, and reporting capabilities.

As of March 2008, the name authority team had evaluated 1,319 organization names, not including those names originally evaluated in the pilot project. After authority control was applied, this group of names was reduced to 532 authorized organization names, a 59 percent reduction. These results were significantly more dramatic than the 8 percent drop seen in the pilot project, but that discrepancy can be explained by the post–pilot focus on normalizing the library’s big deal packages. For instance, 58 Elsevier variants were reduced to only one authoritative name, a 98 percent reduction in the number of variants and errors, contributing to a more dramatic decrease overall. Similarly, the 532 assigned authoritative names relate to 21,672 titles, a substantial percentage of the more than 35,000 unique manifestations of resources currently managed through E-Matrix, further confirming the widespread effects of controlling the names of major organizations.

On the broadest level, the introduction of authority control of names into E-Matrix has demonstrated progression toward cleaner, more usable implementation of data on the basis of the concept of roles as defined in the DLF ERMI. By eliminating duplication, variation, and errors in the organization data, E-Matrix enables use of role relationships. Titles that share a common publisher, vendor, or provider can now be identified using the authoritative name that links them to a single organization. Without authority control, an individual would have to manually account for each organization variant every time he or she worked with ERM data.

The E-Matrix licensing module also benefits from the use of clean data. As the library begins its license mapping process, human data entry will be used to map the details of a license into structured data elements within E-Matrix. Each organization name entered into the license form must correspond to the correct serial resources. To ensure that a license is correctly applied to all resources whose terms it dictates, the licensing module will allow only authoritative names to be entered into the licensor field. Limiting the available licensors within E-Matrix preserves the appropriate use of roles and relationships throughout the data.

The benefits of clean role relationships can be seen even more substantially in E-Matrix’s reporting module. The name authority team, in collaboration with collection managers and programmers, has begun to test the capabilities of the module to incorporate authoritative names in ways that enhance the comprehensiveness and flexibility of reports. An authoritative publisher report displays every serial title in E-Matrix associated with a publisher. This very basic report serves mainly as a test object to illustrate how authoritative names have been incorporated into the data. Within E-Matrix, users can link from this report to detailed displays of publisher and resource information, facilitating discovery of related organizations and titles. Using the report module’s export tool to transfer the data to a spreadsheet or database, all resources published by the same entity can be easily identified, and groups of similar publishers explored. A similar test report displays each authoritative organization along with its related titles. This report expands the authoritative publisher report across all roles, allowing for a more complete picture of how organizations relate to works and highlighting implicit business relationships by illustrating the multiple organizations associated with a single resource through their respective roles.

In addition to these preliminary reports, the name authority team also has conceived more advanced reporting using authoritative names. For example, by leveraging the subject categories that have been assigned to every resource in E-Matrix, reports can be generated listing the most common authoritative publishers within any given subject area. That data could be combined with the money spent on each resource to provide a comprehensive picture of the amount spent per publisher in a certain subject area. Without the authoritative name data, such advanced reporting would be much more difficult because the proliferation of variant organization names would make the process of identifying and collocating all instances of a publisher tedious and prone to error. The name authority team hopes to see staff from other departments take advantage of the clean relationships between organizations and resources to produce analogously sophisticated custom reports.

To ensure continuing data integrity across all modules of E-Matrix, the organization name authority project will continue to be maintained and enhanced once the primary authority project has been completed. Monthly maintenance reports will list any new organization names that have come into E-Matrix since the authority was last updated. Organization names with existing authorities will automatically be subsumed under the proper heading. In this way, the library will retain control over organization data within E-Matrix.