Leveraging student course enrollment data to infuse personalization in a library website
Ian Chan
Library Systems, California State University San Marcos, San Marcos,
California, USA
Abstract
Purpose – The purpose of this paper is to describe the benefits of integrating personalization within
a library web site and presents methodology for achieving this goal within an academic setting.
Design/methodology/approach – The project documented in this study explores the use of student
course enrollment data as the basis for creating a personalized library web site. Off-the-shelf, open
source applications are used in conjunction with existing university data to deliver a final product that
offers an enhanced user experience for the university community.
Findings – Adaptive personalization is increasingly commonplace on the web. Academic libraries
have a unique source of existing data that offers the potential of adding personalization to the
library web site. At present, the personalization of library online services remains largely unexplored.
This project illustrates one relatively low-cost method to help libraries interested in creating personalized
web sites.
Practical implications – This paper provides a guide for libraries interested in the implementation
of personalization within their web sites.
Originality/value – The project described in this case study is highly unique within libraries.
The paper outlines the feasibility and technical requirements associated with using course enrollment
data to add personalized content to a library web site.
Keywords User interfaces, Personalization, Human computer interaction, User-centered design,
Student-centered library, User experience design
Paper type Case study
Introduction
While academic libraries have previously offered customizable web sites, adaptive web site
personalization through the integration of student data remains unexplored. This case study describes the
implementation of a system that utilizes student course enrollment data to create adaptive personalization
within the context of an academic library web site. The study presents the design, development,
implementation, and assessment of such a system.
The challenge of building a user-friendly academic library web site
Academic libraries continue to develop their web sites as gateways to information resources, research
help, and services. This investment has led to a great deal of research focusing on the assessment and
improvement of academic library web sites (Brantley et al., 2006; Kim, 2011). A number of these studies
show that users continue to find academic library web sites complex and difficult to use. In focus groups
composed of students and faculty from two universities, many “participants commented that library
websites are overly complex and hard to navigate, and that a simplified portal designed to meet individual
needs would be welcome” (Munro and McLure, 2010).
Participants in a 2011 study of a university library web site “perceived the usability of the [library’s]
website design to be challenging” (Kim, 2011).
Recommendations aimed at improving library web sites have increasingly called for libraries to
take a user-centric approach in designing their sites. A 2008 article reviewing 111 Association of
Research Libraries web sites found “the universe of information presented on academic library
homepages still focuses on library functions, requires numerous pathways for access, has overwhelming
options, and takes a ‘one-design-for-all’ approach that fails to recognize users as individuals.”
The author recommends libraries offer each user a “personal library space,” based on that user’s profile,
to reduce information overload and present “library resources in a targeted and customized manner” (Liu,
2008). Somerville and Brar (2009) suggest that academic libraries should re-design their web sites and
online services from a user-centric rather than library-centric perspective.
Studies of factors impacting the use of academic library web sites also point to a positive
correlation between the perceived-ease-of-use (PEOU) of a site and the future intentions to use the site.
Libraries that improve the PEOU of their sites are likely to increase the desire of students to use the
library’s web site (Heinrichs et al., 2007; Kim, 2010). A study of user perceptions toward university
library web sites recommends practitioners “design user-focused library websites that enhance the
usability of [the university library website] and provide customized services to different user groups in
order to increase usage” (Kim, 2011).
Using personalization to create a user-centric library web site
Personalization can be defined as “the ways in which information and services can be tailored to match
the unique and specific needs of an individual or a community. This is achieved by adapting presentation,
content, and/or services based on a person’s task, background, history, device, information needs,
location, etc. essentially the user’s context” (Smeaton and Callan, 2005, pp. 299-300). Frias-Martinez et
al., 2009 describe two major approaches to personalization: adaptability and adaptivity.
Research on the potential benefits of personalization systems continues to grow (Park et al., 2012;
Sunikka and Bragge, 2012). Improving web site ease-of-use and reducing information overload are two
oft-cited benefits of personalization (Ketchell, 2000; Kumar and Benbasat, 2006; Liu, 2008; Wang and
Yen, 2010; Lee and Cranage, 2011). Research conducted by Liang et al. (2007) indicates that “reducing
information overload is the most important concern for users in seeking information and that
personalized recommendations can perform well when users use the media to seek specific information.”
A study by Porter (2011) on undergraduate research strategies suggests that personalized
recommendations would help students locate relevant databases.
In the early 2000s, a number of academic libraries implemented web site personalization systems.
A review of web site personalization initiatives from that time period found most favored the use of
adaptable or user-customized systems (Jeevan and Padhi, 2006). A number of academic libraries used
MyLibrary, a web-based content management system (CMS) that gave users the ability to customize the
layout and content of a library web site (Morgan, 1999; Cohen et al., 2000; Ghaphery, 2002;
Gibbons, 2003).
However, research on how users perceived the adaptable personalization systems
offered by libraries has shown those systems encountered limited success. Gibbons (2003)
Student course enrollment data and Ciccone (2005) analyzed usage of the library web sites at their
academic institutions and discovered that only a small percentage of users choose to use the
customization functions. Responses to a survey conducted by librarians at Oregon State University,
indicated that “supposedly tech-savvy students were unenthusiastic about a customizable [library] portal”
and that “if the burden of customization [was] on them, only a small percentage would take the time to set
up and use those features” (Nichols and Mellinger, 2007). In a 2011 study of academic library users,
participants reported that it was difficult to customize the library web site and many did not utilize
the customization functions (Kim, 2011).
The limited acceptance of customizable library web sites may be due to their use of the adaptable
approach to personalization. In contrast, the adaptive approach is generally perceived to be more
favorable among users. Results of a five-year study at the Galter Health Sciences Library “indicated that
users were receptive to personalized resource selection and that the automated application of
specialty-based, personalized HSLs was more frequently adopted than manual customization by users”
(Shedlock et al., 2010). This supports the results of earlier studies where users preferred adaptive over
adaptable personalization (Frias-Martinez et al., 2006, 2009).
The adaptive or system-driven approach to personalization utilizes automation to
produce user profiles that are based on the analysis of user interests and behavior
(Frias-Martinez et al., 2006). In some cases, content relevance is dependent on data provided by the user
as part of an account registration process. The personalization system correlates each user’s data with
relevant products, services, and information in order to generate recommendations. The adaptive
approach might also incorporate analysis of a user’s browsing and purchasing patterns (Forsati and
Meybodi, 2010). Some adaptive systems include functions that make adjustments based on implicit user
feedback (Pu et al., 2012).
Major challenges in the implementation of adaptive personalization include the
development of an appropriate user model and accurate analysis of user browsing data. “There is
considerable difficulty in getting real and correct user interests and mapping them effectively into the
products and services offered by the library. Also, the interests of users keep on changing continuously”
(Sirisha et al., 2009). Unless a user logs-in with the system or has previous browsing history on a site, the
system will not have a basis from which to generate content relevance. Even if a user authenticates or
allows tracking of their browsing history, their goals and interests may differ between or within visits.
Research on the use of adaptive personalization in the context of academic library
web sites and digital libraries is heavily focused on deriving user interests through the analysis of user
searches and click-throughs (Sunikka and Bragge, 2012). The integration of user data held within the
student information systems (SIS) of academic institutions is a heretofore unexplored area within the
literature.
Applying an adaptive personalization solution utilizing student course enrollment data
As with other academic libraries, a primary goal of the California State University San Marcos (CSUSM)
Library is to design efficient pathways to information resources. To achieve this goal, the Library
focusses on reducing barriers to resources and ensuring consistent, user-friendly information-finding
tools. The Library’s web site plays an essential role in this effort as it provides the primary means by
which users locate and retrieve information. To support this mission, the Library invests significant
resources toward creating a positive and fulfilling online research experience for its users.
The CSUSM Library also focusses heavily on providing course-integrated information literacy
instruction and personalized research assistance. This instructional strategy places an emphasis on
meeting the specific research needs of individual students. Library web site personalization was perceived
as a strategy that might improve the learning experience as it would focus on each student’s
area of study. With the goal of providing a better, targeted research experience for students, the Library
initiated a project to investigate the use of personalization on its web site.
As described in the preceding sections, research on the use of personalization in libraries clearly
indicated that an adaptive approach was more likely to achieve success. However, the development of a
data-mining system for automated personalization was beyond the scope of the Library’s personnel
resources.
In addition, an approach based purely on the data-mining of web site usage would have been difficult and
unlikely to succeed for two reasons: students often enroll in multiple classes of different subjects and the
development of an algorithm to accurately analyze the vast amount of usage data would have been very
complex. This need to consider alternative methods of adaptive personalization led to the realization that
academic libraries have access to a tremendous amount of information on their users: course enrollment
data. Using this data as the basis for personalization would allow us to generate user profiles without
requiring implicit input from our students. Content on the library web site could then be associated with
each user’s profile based on relevance to the student’s courses. With this data model in mind, we choose
to explore the use of course enrollment information as the foundation of our personalization strategy.
Project outcomes and requirements
The overall goal of the personalization project was to enhance the research experience and create a more
user-centric library web site. The primary outcome was to develop a system that automated the process of
connecting users with the library resources most relevant to their research needs. It would offer simpler
pathways for accessing online resources and enrich the overall user experience.
In order to maintain data security and user privacy, it was critical that the system utilize a reliable
and secure user account management system. To minimize maintenance overhead for the Library and for
the CSUSM Instructional and Information Technology Support group, the personalization system would
have to leverage existing user data stores. The system would need to visually highlight resources and
research guides relevant to the researcher’s subjects of interest. It would also display the appropriate
librarian profiles within a user’s subjects of interest. Users would not be required to create accounts, self-
select subjects areas, or login to additional systems. Lastly, it was essential that the system utilize existing
open source applications and that it would not require significant investment of programming resources.
Developing a data model
The effectiveness of a personalization system is dependent on the acquisition of pertinent user data and
utilizing that information within the framework of a data model. The process can be summarized as
follows: “(a) the collection of Web data, (b) the modeling and categorization of these data (preprocessing
phase), (c) the analysis of the collected data, and (d) the determination of the actions that should be
performed” (Eirinaki and Vazirgiannis, 2003, p. 4).
The initial phase of the project focused on building a reliable transfer of course enrollment data
and developing methods of applying that data to personalization. For the project’s data requirements, it
was necessary to develop a recurring process that would make available the course numbers and
instructor names associated with each student’s course enrollments. An existing process provided an up-
to-date extract of student account information that was transferred on a regular basis from the SIS to
the integrated library system (ILS). By expanding this process to include student course enrollment data
from the SIS, we were able to meet the data requirements of our personalization system.
To associate library resources with individual students, we needed to cross-reference the course
enrollment of each student with a set of primary subject headings. All of the Library’s resources are
organized into 30 primary subjects and each subject corresponds to one of the disciplines associated with
the courses offered by CSUSM. To create the cross-references between courses and library primary
subject areas, we setup a database table to link all course numbers with their relevant library subject areas.
This table matches the departmental prefix of each course number with the associated library
subject. For example, all courses prefixed with “MRKT” are identifiable as relevant to the areas of
marketing and business.
In the personalization system, each user profile would include the library subject areas most
relevant to the student’s course enrollments. By cross-referencing subjects in the user’s profile with those
associated with the resources, guides, and librarian profiles on the Library web site, the system would be
able to generate recommendations.
Selecting Drupal as the CMS for supporting personalization
Building a homegrown system with user profiles, account management, resource metadata, and content
management was beyond the scope of the Library’s resources. It was more cost-effective and sustainable
to customize and extend an existing web-based CMS that already incorporated the requisite functionality.
Drupal (http://drupal.org) is an open-source CMS offering a combination of features that make it
an ideal candidate for building a web site personalization system. Those features include user accounts,
highly extensible content types, and a robust taxonomy system. In addition, it offers a powerful, browser-
based tool for creating and displaying relational database queries. Its extensive list of add-on modules
meant we would not have to invest resources toward developing new functionality to meet our project
requirements.
The following is a list of requirements meet by Drupal:
ability to import external data and map to internal content fields;
integration with external Lightweight Directory Access Protocol (LDAP)
authentication systems;
ability to create and manage a highly varied set of content types, each with its
own set of data fields;
ability to create and display complex queries without extensive programming
knowledge;
a flexible and secure user profile system; and
a highly configurable presentation layer.
Based on these requirements, Drupal was selected as the CMS for the personalization system. When
Drupal was implemented as the Library’s web site CMS, all of the Library’s e-resource profiles, research
guides, and librarian profiles were created as “nodes” of content within the system. In Drupal, a “node”
refers to a single entity of content that may contain any number of fields. Within the context of the node
system, it is possible to designate any number of node types, each with its own configuration and set of
fields. Each node type may have any number of fields. Creating and managing node types, node fields,
and nodes is accomplished via Drupal’s web-based administrative interface. Prior to version 7, managing
the fields of a node type was enabled via an add-on module known as the Content Construction Kit
(CCK). In Drupal 7, the node management features of the CCK module are incorporated within the core
application.
Within the CSUSM Library web site, the e-resource node type is an example of how we use
Drupal’s node-based content system. This content type is used to store information describing the
attributes of online resources such as databases, e-books, and e-journals. The fields in the e-resource
nodes hold information such as database URL, authentication method, full-text availability, dates of
coverage, and more. User profiles, research guides, course guides, and librarian profiles are also nodes of
information within Drupal. Building relevance between content and users is possible because each node is
associated with subject terms entered into the system’s central taxonomy.
An essential element of a personalized web site is the privacy and security of user data. The
Drupal user account system restricts public access to user data while allowing users to access and update
their own profiles. For added security and ease of maintenance, we do not store student ID numbers,