D-Lib Magazine
April 2000

Volume 6 Number 4

ISSN 1082-9873

Suzanne Cohen, John Fereira, Angela Horne, Bob Kibbee, Holly Mistlebauer, and Adam Smith
Cornell University Library
Cornell University
sac29, jaf30, akh8, rk14, hlm7, [email protected]

Abstract

What is MyLibrary?

MyYahoo!, MyCNN, MyBookmarks, MyThis and MyThat. Internet users have demanded a personal face to the World Wide Web, and Web portals and information providers have responded. Why not MyLibrary? The Library and Information Technology Association (LITA) has defined MyLibrary-like services as the number one trend "worth keeping an eye on"[1]. "Library users who are Web users, a growing group," the experts agree, "expect customization, interactivity, and customer support. Approaches that are library-focused instead of user-focused will be increasingly irrelevant."

In response to the needs of web-savvy patrons, the Cornell University Library (CUL) implemented a MyLibrary service this year, making finding and using library resources easier than ever. MyLibrary [2] is an "umbrella" service for two new products: MyLinks and MyUpdates. Other products are in development.

My Library at Cornell University

MyLibrary�s MyLinks is a tool for collecting and organizing resources for private use by a patron. These resources may or may not be "official" Cornell University Library resources. Our patrons best understand this service as a "traveling set of bookmarks". Most patrons of the library use a variety of machines to access Internet resources. For example, you may have a computer at home and one at work. Why should you create your bookmarks twice, or carry around a diskette containing your bookmarks? Students who rely on lab computers never know which machine they will use next. With MyLinks, a patron�s favorite sites are just a click away from any machine.

Partial View of MyLinks Screen

MyLibrary�s MyUpdates helps scholars stay informed of new resources provided by the library. In the past, Cornell University, with its 19 major library units, has addressed this problem by creating accession lists for individual departments. But such lists are cumbersome for library staff to produce, and they can only serve a small number of patrons.

By contrast, MyUpdates periodically queries the on-line catalog to determine which resources are new. If the new resources match the pre-identified needs of a patron, the system notifies the patron. Patrons no longer need to search the on-line catalog regularly to be aware of new resources. MyUpdates searches the on-line catalog for patrons and e-mails the results to them automatically.

The current release of MyLibrary is just the first step towards offering numerous personalized services to Cornell University students, faculty, and staff. The library has invested a lot of time and resources into making MyLibrary a reality. Why is the library doing this when there are commercial sites providing similar services?

Why not use a commercially available site?

What distinguishes commercial, personalized online services from MyLibrary is their inability to meet the research needs of their users. MyYahoo! is not a library-centric service, nor is it likely ever to become one. It caters to the casual Web user�s needs for general information, such as the day�s headlines and local weather, but is unable to feed a user�s knowledge-seeking desires. Moreover, it does so without finesse. It is quite simple to find the closing stock price for Amazon.com and add that figure to a web page. It is far more difficult to suggest to a user quality reference sources that discuss the reasons behind Amazon�s rise to the apex of online specialty retailing. MyLibrary has the potential to be an academic�s primary knowledge-based web space as he or she follows a research path (or, indeed, simultaneous research paths) for many months or even several years. The proposed enhancements discussed elsewhere in this paper detail many of the research-oriented "add-ins" we believe will cause library patrons to rely on MyLibrary as a primary research tool.

Why did Cornell develop MyLibrary?

The acute need for a MyLibrary model at Cornell was recognized after patrons requested solutions for information excess. Just as the number of pages on the World Wide Web is ever increasing, so too is the range of resources in the 19 Cornell libraries; we became concerned with the possibility that the scholars in our community were becoming lost in a plethora of information.

In 1998, Cornell librarians Karen Calhoun and Zsuzsa Koltay conducted a focus group study to gauge library patron use of and satisfaction with the Cornell University Library Gateway [3]. The Library Gateway [4] is Cornell University Library�s Web presence. The Gateway offers an array of information and online services, including a selection of 2,000 networked resources.

Calhoun and Koltay found that users consider the Gateway occasionally "overwhelming" and desired "a more personal space�where they can dictate how important certain resources are based on their own needs" [5]. Additionally, they noted, "Users want to be in closer communication with the library. They feel that the library provides great services for those who happen to find out about them and teach themselves how to use them." Among their post-study recommendations were the following suggestions:

1. "Explore new ways and technologies to simplify the task of navigating a complex information scene";
2. "Provide personalized and current awareness services";
3. "Add new features to the Gateway such as "What�s New", hot topics, favorite databases, and subject guides."

The study�s findings prompted the Cornell University Library to devise a technological tool that would simultaneously fulfill as many of the above recommendations as possible. In so doing, MyLibrary has developed into a hybrid of the commercial personal portal and Cornell�s own Gateway.

Implementing MyLibrary also furthers CUL�s goal to "provide outstanding service to the university in support of Cornell�s information needs by integrating traditional and digital resources and services" [6]. The traditional services that librarians have always provided include the selection, organization, and provision of enhanced access to useful information. In the past, libraries owned all of the resources in their collections, and patrons would need to physically visit the library to use its materials. Today, however, libraries are competing with other information sources (such as the Internet), and they are also making their own collections available to the public using digital technologies. Now patrons may be predominantly virtual users of resources, and as a result, many public service discussions revolve around how to ensure that patrons will continue to patronize physical libraries. MyLibrary is one strategic way to keep people coming to the library by providing a service that supports Cornell�s information needs. The MyLibrary model accepts the reality that people are finding information on the Internet and provides a way to integrate these "other" resources with the Library resources that have been evaluated and selected by librarians. Proposed MyLibrary enhancements will encourage use of the physical Cornell Library collection (via MyUpdates, for example) while also providing easier access to our digital holdings.

Current library literature and presentations at library conferences discuss how librarians can better market themselves to convey the valuable services that they provide. MyLibrary "allows for incredibly detailed targeted marketing of [the] library�s staff, services and resources based on such factors as patron�s unique interests, the types of sources they use most, their academic major, or any other factor that seems relevant" [7]. The librarian can learn of a patron�s interests and then forward to him/her suggested targeted resources, thereby launching the library into the world of "push" technology. We need to be proactive rather than reactive if we do not want to lose our patrons to commercial competitors.

Traditionally, if librarians know of student or faculty research interests, they attempt to keep those researchers informed of potentially useful sources. Due to the size of Cornell�s research population, it is not possible to know everyone�s research topics. MyLibrary automates this function and allows the user to initiate the updating process. Researchers can save useful resources in MyLinks (resources that they have discovered themselves or ones suggested by librarians via targeted notification) and be automatically notified by MyUpdates of new resources related to their saved search topics. Researchers who travel can access their saved resources wherever they are.

As we create our own digital library products and purchase more commercially available digital information, our patrons become more overwhelmed by their options. It is difficult to organize resources into categories that are intuitive for every patron, especially when our patrons include undergraduate students, graduate students, staff and faculty, all of whom are studying hundreds of different subjects. Cornell University Library�s "Library Gateway" attempts to be a structured web conduit, but MyLibrary gives the user a way to create their own user interfaces, ones which make especial sense to them. They can create these interfaces without even having to know how to make their own web pages.

Beyond the theory of MyLibrary as an excellent information tool lies its human reality: the creation, implementation, and enhancement of MyLibrary requires a serious commitment of staff time. Without an increase in the number of staff, the MyLibrary system must take library personnel away from other duties or projects. It is hoped that the time and effort put into MyLibrary will pay off in more information-literate users who are more aware of the resources that could be useful to them. Additionally, public services staff will be able to track user perception of the usefulness of library and other electronic resources. For example, which database or web site has a patron chosen to put into a MyLinks folder? It might also be valuable to monitor if a subject librarian�s recommendation to patrons (via the MyLibrary Announcement Service described below) results in additions to their MyLinks. This information can then be analyzed according to discipline and patron group, providing vital statistics which previously we have not been able to gather easily. Imagine if every student, staff, and faculty member made visits to their MyLibrary pages a part of their daily routine and we were tracking that activity? The library system in general, and public services in particular, can only benefit from increased patron use of MyLibrary.

How was MyLibrary developed?

The MyLibrary system was developed by the Personal Electronic Services Working Group, a collection of Cornell University librarians and programmers. The members of this working group represented seven of the libraries on campus. This mix of players helped ensure that we developed a product useful to all students, faculty, and staff. The initial charge to the team was to determine which personal services Cornell users needed. We quickly identified the need for a personal set of "shortcuts" to resources (to be called MyLinks) and for notification when items of interest are added to the on-line catalog (to be called MyUpdates). MyLinks and MyUpdates, which share a developmental approach and core technologies, in fact are individual systems unto themselves. In combination, these two products form the current MyLibrary service. The next MyLibrary release will contain additional services.

During the course of the development of MyLibrary, there were several decisions made regarding its implementation and the choice of technologies used. Two of these decisions were to use a formal design process and have the development team adopt certain standards for a web-based application with a database back end.

The Rational Unified Process (RUP) was the project methodology employed. This object-oriented process provided a strong justification to employ an object-oriented language for the development of MyLibrary. As a result, Java programs dynamically create the HTML seen by the patrons on almost all of the MyLibrary pages. Specifically, the design team chose to use Java servlets for implementing access to the database and providing dynamic web content. This choice allowed the use of the JDBC standard for querying, inserting, and modifying records in the database, as well as the use of some pre-written packages for displaying dynamic content.

Oracle is the database technology used for storing information about users of the MyLibrary system. It is a robust database with the capacity to handle the large number of transactions and vast amount of data an application such as MyLibrary might generate.

One of the most important aspects of the MyLibrary system is the authentication of its users. The authentication system ensures that the information users choose to put into the system remains private. The Kerberos Authentication mechanism provides user id and password authentication. Once Kerberos authenticates the user, we use the user's Cornell Network ID to show only the folders and resources associated with that ID. The system associates dynamic database content with the user�s Web browser through the use of the Java servlet sessions mechanism created when the user logs into MyLibrary. When a user is finished using MyLibrary, they merely need to click on a "Logout" link to close the session and prevent subsequent users on the same computer from viewing their personal information.

MyLinks, in both appearance and use, is a very intuitive, user-friendly product. The basic premise of the simple to understand design is that a patron has folders containing links to personally chosen Web sites. Patrons are able to name the resources and folders whatever they choose, order the folders in the manner most useful to them, add new resources or remove old ones from folders, and edit resources if the URLs change. Patrons have the option of selecting from either the librarian-gathered resources in the Gateway or from any Web resource. When patrons enter MyLinks for the first time, they will find two default folders already set up for them. One folder contains links to library services, and the other contains a wide variety of links to Web search engines. The default folders serve as examples to assist patrons in understanding the concept of MyLinks.

The development of MyUpdates was vastly different from the development of MyLinks. MyUpdates allows patrons to identify their areas of interest; the system then notifies the patrons when resources matching those interests are added to the catalog. The concept sounds simple, but the implementation was not.

The first step was to determine how the data would be accessed from NOTIS, the on-line catalog used at Cornell. NOTIS is a mainframe-based system ill suited for the kind of user interaction and search processing that MyUpdates requires. In addition to the technical limitations of NOTIS, the acquisition and cataloguing workflow at Cornell made it difficult to know when a bibliographic record in NOTIS should be considered "new." When a resource is first acquired, a corresponding record in NOTIS is created. Because we want knowledge about the future availability of the resource to be available to the public as quickly as possible, the acquisition record in NOTIS is made available to the public while the item is being catalogued. Widespread notification via MyUpdates of a new resource that was not yet fully catalogued and accessible would, however, be undesirable.

To begin to solve this dilemma, we made the decision to create a local database to temporarily store NOTIS bibliographic and holdings data for new materials. As with MyLinks, we selected Oracle for this purpose. Oracle can support large amounts of information and large numbers of users, and offers powerful techniques for searching through the information it stores. All of these features would be necessary for MyUpdates to succeed.

A cataloguer on the project team worked closely with one of our NOTIS system administrators to create specifications that would help determine which NOTIS records should be considered "new." Multiple criteria were identified to test against each record to determine their readiness for inclusion in MyUpdates, and although these criteria were not fool-proof, tests indicated that most of the records that met the criteria were in fact new; few resources that should be included were being overlooked by the process.

Once this process was complete, the two team members then created specifications for retrieving a subset of bibliographic and holdings information from each NOTIS record while preserving the diacritics each record contained. This information is then stored in our database.

A Web interface allows patrons to create multiple search profiles. Their profiles are stored in the MyUpdates database and are used to search periodically against the database of new records extracted from NOTIS. The Web interface resembles an OCLC-style Boolean keyword-searchable interface. But a concern immediately arose: because Cornell�s libraries serve many diverse disciplines, how could we provide a method of narrowing these keyword searches by subject area so that the results of each search would be more relevant?

The MyLibrary team decided that, particularly for the first release of MyUpdates, using a system like the Library of Congress Subject Headings -- or even an abbreviated version of such a system -- to define subjects would be too difficult to implement. Instead, as a way to roughly limit searches by subject, we chose to add the ability for patrons to limit their searches to materials held at one or more libraries at Cornell which themselves each have a special topical focus.

Finally, before patrons save their search profiles, they can test their searches against the previous year�s worth of new resources extracted from NOTIS. The resources that meet their criteria are displayed in the browser. Patrons can then judge by both the number and appropriateness of the resources retrieved, just how broad or narrow, and also how relevant, are their search profiles. The system calculates an average of the number of resources patrons are likely to receive each time they receive notification of the availability of new resources. By basing this calculation on an entire year of new materials data, we can more accurately reflect yearly acquisition trends and the effect of fiscal cycles on purchasing.

Once a patron is satisfied with a particular search profile, he or she saves the profile (though it can be edited again at any time). The MyUpdates system automatically runs all patrons� search profiles against the most recently-extracted NOTIS records every two weeks. The system then e-mails to each patron a report of the new records that meet the criteria in his or her saved profile.

By automatically e-mailing the MyUpdates reports, we provide a very simple way to approximate the goals of "push technology." Push technology fell from favor because traditional push techniques involved installing a client on an individual machine that waited for information. Often, people received much more information than they expected, and rather than try to limit what they received, people simply stopped using these services.

We recognized that MyUpdates could suffer a similar fate if we did not carefully consider our patrons� needs. By managing search profiles through a Web interface, we hope to avoid the need to install specialized client software on the patron�s machine.

In addition, the Web interface allows patrons to test their search profiles and get a sense of how much information they will receive before they commit to receiving a particular report. Further, patrons can edit and re-test the searches at any time. Taken together, these features may help avoid the previous problems with push technologies of information overload.

Finally, the delivery of reports by e-mail means that patrons receive the reports via software they already use and are familiar with on their computers. Receiving the reports should not be intrusive. For those without e-mail, or for those who are away from the workstation where their e-mail is downloaded, the most recent reports generated for each search profile the patron has created can be viewed through the MyUpdates Web interface. There is also a link to MyUpdates from within the Library Services folder of MyLinks, making the service we call MyLibrary truly integrated.

What is next for MyLibrary?

MyLibrary was conceived not merely as a clever utility, or even a smorgasbord of utilities, but as a vehicle for increased interaction with library users in an otherwise depersonalized digital environment. Future enhancements will expand the interactive characteristics of MyLinks and MyUpdates in several directions.

One certain direction is toward building an area for delivering library-related messages within the user's personal space. A message space or personalized bulletin board could be an extremely effective way of communicating with patrons. The personalized message would avoid common pitfalls of library-patron communications, in particular such just-in-case and dubiously effective methods as mass-mailings and broadcast e-mails on the one hand, and passive media such as web pages on the other. In the MyLibrary model, patrons would choose the kind of message from the library they would want to see and would choose the delivery mechanism. For example, a patron could choose to be reminded of library workshops or instructional programs at a particular point in the semester. They might want to know when important new bibliographic databases are added to the Gateway, or when new web sites relevant to their areas of interest appear. Cross-disciplinary researchers might wish to receive notices from different subject specialty libraries.

With careful management, the library should be able to push information to patrons within parameters the patrons themselves have chosen. Patrons will be able to choose whether to receive messages through e-mail or to be notified when they log into MyLibrary that they have messages waiting on their personal page. Such a page might also prove a useful alternative to delivering MyUpdates reports via e-mail.

At the same time, a MyLibrary message page should provide patrons with easier methods of access to librarians and library services. Since the patron�s area of interest has been profiled, the patron can have e-mail links directly from his or her MyLibrary page to relevant subject specialists in the library. Possibly a link from a page like this will prove the vehicle that finally moves "chat reference" off the starting block.

MyLibrary can make services available to patrons with less fuss. We imagine calling up interlibrary loan forms, renewal requests, requests for purchase and other forms from MyLibrary with personal information already filled in. Since patrons are authenticated when they log into MyLibrary, it should also be possible to pass ids and passwords into these standard forms from a stored profile.

Another interesting direction for MyLibrary to develop is as the basis for providing customized services for groups as well as persons. One possible implementation is to customize services for academic departments. In this model, reference librarians and bibliographers can work together with department representatives to select appropriate collections of links and update profiles for broader academic subject areas. Again, paralleling the personal model, the library can push information to this broader group, while the department has access to clearly identified and easily accessible library specialists through e-mail or chat.

MyLibrary and similar personalized services will provide new ways for libraries to interact with their various patron bases. They can empower individual patrons with a strong suite of utilities and tailored communications. The same service forms a convenient way to embrace user groups such as alumni who otherwise have difficulty identifying and using their alma mater's digital resources. Departments and other academic groupings gain similar benefits, while the library assumes a higher profile in a web-flattening information environment. The MyLibrary concept is simple, but powerful: personalization of services is the key to institutional engagement for the digital library.

Notes and References

[1] "Technology and library users: LITA experts identify trends to watch." 1999. Available at <http://www.lita.org/committe/toptech/trendsmw99.htm>.

[2] MyLibrary may been seen and explored via the guest login account at <http://mylibrary.cornell.edu/servlet/GuestLogin>.

[3] Library Gateway Focus Groups Report. 1999. Available at
< http://www.library.cornell.edu/staffweb/GateEval/contents.html>.

[4] The Cornell Library Gateway is available at <http://www.library.cornell.edu>.

Cornell University Library Annual Report 1998-1999, Objectives 1999-2000. Available at <http://www.library.cornell.edu/ulib/ar98-99/report.html>.

[7] Ken Winter. "MyLibrary Can Help Your Library." American Libraries. August 1999: 65-67.

Copyright � Suzanne Cohen, John Fereira, Angela Horne, Bob Kibbee, Holly Mistlebauer, and Adam Smith

D-Lib Magazine Access Terms and Conditions

DOI: 10.1045/april2000-mistlebauer