Intelligent Agent Concepts in the Modern Library

Please do not remove this page

Dent, Valeda https://scholarship.libraries.rutgers.edu/discovery/delivery/01RUT_INST:ResearchRepository/12643389390004646?l#13643533210004646

Dent, V. (2007). Intelligent Agent Concepts in the Modern Library. In Library Hi Tech (Vol. 25, Issue 1, pp. 108–125). Rutgers University. https://doi.org/10.7282/T3B56H29

This work is protected by copyright. You are free to use this resource, with proper attribution, for research and educational purposes. Other uses, such as reproduction or publication, may require the permission of the copyright holder. Downloaded On 2021/09/30 23:21:33 -0400 Intelligent Agent Concepts in the Modern Library

Valeda F. Dent

Introduction

Agent technology, a sub-field of artificial intelligence, is not a new concept. The idea of the intelligent agent was first introduced by John McCarthy in the mid-1950s, while at MIT.

Bradshaw (1997) suggests that it was not until after World War II that the forerunners of the software agent began to emerge. Norman (1997) advances that the “most relevant” predecessors of software agents were things like factory control devices, and automatons that controlled the takeoff, landing and piloting of aircraft. Lewis suggested in 1996 that a shift away from the network operating system to Internet-based network computing was taking place. If true, we are now ten years into this shift, perhaps approaching the tail-end of it. There is no doubt that the advent of the Internet has fueled discussion about the use of agent technology, and supported subsequent implementation on a much broader scale.

The popularity of agent technology has increased, decreased and increased again over the past few decades (as the wave of professional literature published during the mid 1980s and

1990s indicates), and a number of researchers proclaimed the future belonged to agents, “Agents will be the most important computing paradigm in the next ten years…By the year 2000, every significant application will have some form of agent-enablement” (Janca, 1996). While this has not transpired, many researchers still believe that the technology is here to stay. One of the earliest intelligent agents was called ELIZA, created in 1966 at MIT by Joseph Weizenbaum, to simulate a conversation with a physiotherapist. ELIZA used natural language to communicate with the user, and her software code included only 240 lines. Today’s agents are incredibly flexible software components, used in complex systems such as traffic management and route guidance (Adler, Satapathy, Manikonda, Bowles and Blue, 2005), power grid system control

(Flynn and Dodd, 2006), and supply chain management (Xue, Xiaodong, Qiping, and Wang,

2005).

One area of intense and ongoing debate is the definition of agents. There are numerous definitions of what an intelligent agent is, and just as many that focus on what an agent is not.

1 There is also rigorous debate about the concept of the truly intelligent agent, whether it exists, or can be designed. In his compilation of essays on agent technology, “Software Agents”

(Bradshaw, 1997), James Bradshaw mentions no fewer than seven possible definitions of what an agent is in the opening chapter. A review of literature by Schleiffer (2005); Hostler, Yoon and

Guimaraes (2005); Chen and Su (2003); [Jensen (2002) REMOVED] Lieberman et al. (2001);

Shang, Shi and Chen (2001); Wooldridge and Jennings (1998); Negroponte (1995) and Maes

(1997) reveal additional definitions, each slightly different from the next. Bradshaw (1997) states that coming up with a “once-and-for-all definition of agenthood” is very difficult, because researchers and developers have so many opinions on agents, their function, and their application (p. 5). He also asserts that “there has been an explosion in the use of the term without a corresponding consensus on what it means” (p. 4). According to Bradshaw, this may be the reason that many developers mistake generic software programs for agents – “some programs are called agents simply because they can be scheduled in advance to perform tasks on a remote machine” (p. 4). Along these same lines, as Foner (1993) observes in Bradshaw (1997, p.

6), there is “little justification for most of the commercial offerings that call themselves agents.

Most are barely autonomous, unless a regularly scheduled batch-job counts”. Franklin and

Graesser (1996) are quick to point out that while all agents are software programs, not all software programs are agents. Petrie (1996) in Bradshaw (1997, p. 10) goes a step further by demonstrating that most of the “current Web-based searching and filtering “agents”, though useful, are essentially one-time query answering mechanisms”.

Defining Agent Technology

With that in mind, a cursory overview of some of the practical definitions of agents is in order. The term “agent” most often refers to a software program that gathers information or performs some other service without the immediate presence of the user. While some researchers [FININ 1997 removed] express doubt that a consensus about what an agent is will ever be reached, many researchers in the area of agent technology agree that the definition by

Shoham (1997) is accurate: “A software entity which functions continuously and autonomously in a particular environment, often inhabited by other agents and processes”. Bradshaw (1997)

2 asserts that agents are able to carry out activities without constant human intervention, while

Jafari (2002) sees agents as “a set of independent software tools linked with other applications and databases running within one or several computer environments. The primary function of an intelligent agent is to help a user better use, manage and interact with a computer application” (p.

29). Wooldridge (1997) describes an agent as an encapsulated computer system, situated in a particular environment, that is capable of flexible, autonomous action in that environment in order to meet its design objectives. Autonomy in this case indicates the ability to compute something, and the preferences over how they use that capability (Birmingham, 1995).

Laurel (1997) includes librarians as human models for agents, stating “agents provide expertise, skill and labor. They must of necessity be capable of understanding our needs and goals in relation to them, translating those goals into an appropriate set of actions, performing those actions, and delivering results in a form that we can use” (p. 71). In addition to librarians,

Laurel goes on to list teachers, secretaries and accountants as a few of the examples of real-life agents who carry out these actions.

Hostler, Yoon and Guimaraes (2005) outline several classifications of agents, a typology first presented by Nwana and Ndumu: deliberative, reactive, Internet, Interface, mobile and stationary. Deliberative agents have a pre-set outline of actions to take once an event is activated by the user or system, whereas reactive agents do not store this type of script, and must meet their goals by using a “stimulus/response type of behavior” (Nwana and Ndumu, 1998, p. 30).

Internet agents are those that work to reduce information overload of the user, by using spiders and crawlers to efficiently search the Web. The interface agent helps users navigate their online environment by observing actions and improving future outcomes. Agents that have the ability to

“roam the network such as the World Wide Web, interacting with foreign hosts, performing the various duties assigned at a remote site, and come back to its user upon completion of the task” are known as mobile agents (Nwana and Ndumu, 1998, p. 36). An agent that does not roam is known as a stationary agent.

Dent, Harris, Martinez and Hall (2001) describe how agent terminology helps users to better understand agent systems, stating, “Many of the systems where agents are at work are

3 highly complex, and understanding what tasks are being undertaken by the system can be difficult…One of the benefits of agent technology is that it provides a helpful, abstract way of talking about complicated distributed systems using basic terminology” (p.56). For example, mediator agents, communication agents and query-planning agents (Birmingham, 1995) are easily recognizable terms applied to a complex set of processes that might occur within a given environment. Sundsted (1998) describes certain characteristics of agents that are relatively exclusive: they are autonomous, have the ability to learn, mobile, persistent, goal oriented, communicative, and flexible. He further advances that although agents are small programs and limited in what they can accomplish on their own, when working in concert with other groups of agents, they can perform any number of tasks.

Researchers have also tried to further specify the work done by agents. Miller’s model describes an agent architecture which includes the four functions of observation, recognition, planning and/or inference, and action or execution (1997), while Maes (1997) states that agents can help users in a number of different ways:

1. They hide the complexity of difficult tasks

2. They perform tasks on the user’s behalf

3. They can train or teach each other

4. They help different users collaborate

5. They monitor events and procedures

Wooldridge and Jennings (1998) address the usefulness of agents, and state that there are two conditions that make new technology useful: the ability to solve problems that have been unsolvable with previously existing technology, and the ability to solve current problems more efficiently (p. 5). It is obvious that there are numerous ways to describe agents, from the practical to the fantastic; the challenge for those in libraries and other information service areas is to separate fact from fiction, and utilize agent technology for support and problem-solving in a way most helpful to users.

A Brief Literature Review of Intelligent Agents in Practical Environments

4 A great deal of progress has been documented since the early study of the intelligent agent, and the professional literature is largely reflective of project-based exploration and research. The focus has gone from a rudimentary implementation of agents for the most basic tasks, to implementation of agent architecture to support information processing in complex environments like large digital libraries. Nwana (1996) sheds more light on the developmental history of agents by suggesting that there were two major timelines for agent research. In 1977, the focus of the research was on the “macro” issues related to agent programming, such as communication between agents, work distribution, and agent cooperation (p. 4). Later in 1990, the focus shifted to include a broader exploration of varying types of agents. There is a wealth of professional literature on the application on agent technology in highly specialized fields such as virtual automotive design (Park, Lee and Shin, 2005), cost projections for state power distribution

(Nordman and Lehtonen, 2005), and even weather forecasting (Lee and Liu, 2004). For those in the information and library fields, however, the most helpful literature tends to be based on case study of projects in relevant environments. Nwana (1996) suggests that overuse of the word

“agent” obscures the fact that agent research is tremendously diverse. He also cautions that much of this research is aspirational in nature rather than reflective of current practice. This brief review will cover the current use of agent technology in Web-based environments with the most relevance to libraries.

Perez (2002) discusses intelligent desktop (client) and Web-hosted search agents, both designed to assist users in searching the Web. Desktop meta-crawler agent software such as

Copernic and BullsEye Pro, and Web-based search agents like Dogpile and ProFusion use agent technology to categorize, sort, filter and report search results on behalf of the user. Perez (2002) notes several differences between the results retrieved using normal search engines versus a search using intelligent agent software. First, the agent searches are “conducted specific to the searching logic and syntax of the various target search engines” (p. 21). Results are sorted and presented in ways that add value to the information, rather than just lists of hyperlinks and other

Web sites. Roesler and Hawkins (1994) suggest that "this technology combines artificial intelligence (reasoning, planning, natural language processing, etc.) with system development

5 techniques (object-oriented programming, scripting languages, human-machine interface, distributed processing, etc.) to produce a new generation of software that can, based on the user’s preference, perform tasks for that user" (p.18).

The online shopping market is another area where a great deal of research is being done on the use of agent technology. In their study on the impact of Internet agents on the online shopper, Hostler, Yoon, and Guimaraes (2005) propose four different hypotheses. They assert that the use of an Internet agent will:

1. Reduce the amount of time end users spend searching for and selecting a

product to purchase online

2. Improve the decision quality of online shoppers’ purchasing decisions

3. Increase the user’s confidence in their purchase decisions for online

purchases

4. Decrease the amount of cognitive effort required during product search and

selection in online shopping environments

The researchers concluded that Internet agents can be helpful for users by reducing the time they spend for product search and selection, and improving decision quality. They also suggest that the findings have implications for online retailers and others in the online shopping sector.

Agent technology is also being used to support teaching and learning. Jafari (2002) presents a model for using agents within the learning environment, in the form of a digital teaching assistant. Such an assistant might communicate with the user via their computer, PDA, or cell phone, in conjunction with course management software such as Blackboard and WebCT.

In the online environment, instructors must constantly monitor their class sites to gauge student participation, retrieve assignments and homework from their dropbox, and send and receive course-related emails. These operational tasks are time consuming, and using intelligent agents for some of these tasks would lessen the workload of the instructor. Baylor (1999) suggests that agents can be used as learning tools in three specific ways: to manage information overload, to serve as a pedagogical expert, and to create programming environments for the learner (p. 36).

6 The author suggests that the goal is to make online course environments “smart”, thus more engaging and stimulating for the user, and gives the example of Writing Partner, a pedagogical application of agent software created by G. Salomon in the early 1990s (Zellermayer et al., 1991).

The software engaged students during the writing process by asking them a series of questions as they wrote their essays, each question geared towards helping the student improve their skills.

Baylor (1999) also discusses the educational value of having students learn by actually building their own agents (p. 39).

In their article, Shang, Shi and Chen (2001) also address the application of agents in teaching and learning. They suggest that the asynchronous learning commonly found in online learning environments is ideal for the use of agents (p. 4). The researchers present IDEAL, an agent-based interactive learning system designed to stimulate online community building, group learning and skill building for students enrolled in virtual courses. In their model, they describe community interaction, student modeling, curriculum sequencing, course material organization and delivery as components of a multi-agent system to support an interactive learning environment.

The very concept of agent technology – automating certain tasks and decisions previously made by humans - suggests a certain level of automation that might appear to limit the ability of the user to control their information environment. Researchers have addressed ways to make sure the user’s needs are being met. Malone, Lai and Grant (1997) suggest that when designing agents for use in learning environments, developers ensure that the agent systems are semiformal in nature (“semiformal systems”), and also easily modified (“tailorability”). Such a system is semiformal because it “involves blurring the boundary” between what humans do, and what agents can do (p. 110). The researchers also state that agent-based applications allow the user to see and modify “the same information and reasoning processes their agents are using” (p.

110). This “radical tailorability” ensures that the user can make programmatic changes to the agent without leaving the application that they are currently working with. Each of these properties ensure the user the ability to correct false assumptions made by the agent that may at some point impact user outcomes. Baylor (1999) states that “these two principles of semiformal

7 systems and radical tailorability enhance user control over managing and coordinating information” (p. 39).

Librarians and others interested in agent technology must first decide if it is appropriate for their environment. Bond and Gasser (1988) specify three rationales to be considered when thinking about agent-based solutions. First, the domain must be distributed in nature. This readily applies to many academic and public libraries, where there may be multiple branches, networked information management systems, disparate resources, and differing levels of technical expertise.

Second, the system must be composed of independent cooperating components, for instance, technical services (cataloging and acquisitions), circulation, and virtual reference. Finally, the system must contain pre-existing or legacy applications that need to interact with new software components. In a library setting, this might include an old OPAC, or an old circulation database.

Agent Technology in the Modern Library

The concept of agent technology has practical application within engineering and computing science, but it also has great potential to impact the way work is done in libraries, as the goal of improving the research experience and search results for users is one shared by every information professional. Agents can negotiate and manage the transfer of information, and this ability makes them useful for certain computer-based, library-related functions and tasks, as well as end-user support. The National Library of Medicine uses COSMO to provide automated answers to users’ questions (Moore, 2004). Librarians at the NLM continually review question and answer logs, and provide new scripts for COSMO. The agent allows users to get more specific answers to questions when searching the Internet reveals too many results, and also allows users to ask sensitive questions that they might not otherwise ask a librarian.

Zick (2000) indicates several areas where intelligent agents might be used in a library setting, including mediation between the user and information, virtual reference, automated serials processing, automated interlibrary loan processing, acquisitions, circulation and digital libraries. In addition, patron information management, cataloging, and online interactive tutorials are also areas where agents might be beneficial and reduce workload. Nardi and O’Day (1996) further suggest that information gathering agents might contribute to knowledge-bases to be used

8 to field user queries. Since the use of agent technology is not yet widespread in libraries, the authors also advance that well-written agent code be cataloged and distributed to other libraries.

A number of researchers have investigated the use of agents for user search support.

Glance (2001) states that “there are three main problems for users of Internet search engines: properly specifying their information need in the form of a query; finding items relevant to their information need, as expressed by the query; and judging the quality of relevant items returned by the search engine” (p. 91). Nordlie (1999) suggests that close to fifty percent of failed Internet searches are due to semantic errors that might be avoided if users were able to benefit from effective online searches on the same topic, done by others before them. This concept, dubbed

“collective knowledge” by Glance (2001), involves a software agent that observes and collects searches from users, and intervenes to recommend more efficient and accurate searches.

Jansen (2005) poses three questions to guide development of an agent that might assist users with their searches by intervening with meaningful feedback and assistance during the search:

1. How often do users seek and implement automated assistance in the search process?

2. Where in the search process do users seek automated assistance?

3. Where do users implement automated assistance in the search process?

The idea of helping the user navigate the electronic landscape effectively has many implications for library staff and users, especially as many students make the use of Google a standard part of their research repertoire. Detlor and Arsenault (2002) present an agent-based model for the modern library which uses interface, information and server agents to form the architecture. The interface agent is responsible for providing more meaningful interactions with the user, primarily by monitoring user behavior such as key word formulation and searching techniques. The information agent searches the library’s collections, and also keeps track of new resources that might be of interest to the user. Finally, the server agent is responsible for retrieving resources for the user from internal and external sources. Detlor and Arsenault state that “for libraries, a major benefit of an agent-based environment would be its relief of reference librarians from trying to service personal information requests for a large number of clients” (p.

9 407). The authors go on to say that agent environments in libraries must be able to support a wide variety of user actions, including browsing and conducting specific searches.

Jansen and Pooch (2004) discuss the agent paradigm called AI2RS (Agent to Improve

Information Retrieval Systems), a model which would be useful in a library setting. The model includes monitoring user actions, saving those actions and saving the information about the object receiving the action. These action-object pairs (Jansen and Pooch, 2004) accumulate as the user continues to use the system, and eventually form a fairly complex model of the user’s information needs (2004, p. 22). This profile allows the agent to determine what type of support, if any, is needed in the following Web-searching situations: structuring queries, spelling, query refinement, managing results and relevance feedback (Jansen and Pooch, 2004, p. 23). Jansen and McNeese (2005) conclude “automated assistance searching systems may improve the Web searching experience by aiding in locating a greater number of relevant documents” (p. 1499).

Each of these areas has great implications for the library user – librarians spend a great deal of time trying to teach users which Boolean operators to use and when, encouraging users to verify their spelling before submitting a search, and to refine their searches. Though the AI2RS will never take the place of library instruction, the applicability in search environments is easily recognized.

Although agent technology is a powerful tool for today’s library, it has many limitations.

Nardi and O’day (1996) define seven things that librarians can do that agents cannot. Agents cannot speak and understand, they cannot read and understand content, make connections across diverse sources, access paper sources not available online, evaluate the quality of information sources, or offer the “human touch” (p. 82). Laurel (1997) expresses objections to agents, stating “If I were looking for a specific book for a research project, I’d probably use a reference librarian; if I were browsing with an opportunistic eye, I’d want to go into the stacks” (p.

69). Wooldridge and Jennings (1998) warn developers to be cautious in their application of the technology. Developers, for instance, should not use agents to solve isolated problems that might be solved in other ways. Norman (1997) suggests that exaggerated expectations must also be tamed, so that users have realistic expectations of what agents can and cannot do. Putting agent

10 technology to work in libraries means collaboration on the part of librarians, systems staff and to some extent, the user. For the use of agent technology within libraries to be fully realized, collaboration and partnerships are essential.

There are a number of examples where agent architecture has been used successfully in library settings, including the University of Michigan Digital Libraries (UMDL) initiative, the United

Kingdom-based Joint Information Systems Committee (JISC)/Electronic Libraries Programme

(eLib) project MALIBU, and the DAFFODIL digital library system. Each of these projects applied agent technology in different ways, but are prime examples of practical uses of agent technology within the framework of today’s electronic library.

The University of Michigan Digital Libraries

Today’s digital library seeks to take advantage of both the well-organized and service- based library, and combine that with the ability to access information from anywhere at any point in time. According to Birmingham et al. (1994), the digital library has the potential to provide access to collections of multimedia information in a variety of formats including text, audio, images, graphics, and video; to support customization for information searching and retrieval; and to reduce the barriers of distance and time with regard to research and learning. In 1994, the

National Science Foundation selected the University of Michigan libraries to participate in one of the then largest digital library creation projects in the nation, the University of Michigan Digital

Libraries (UMDL) Initiative. The goal of the project, which was co-sponsored by the Defense

Advanced Research Projects Agency (DARPA), National Aeronautics and Space Administration

(NASA) and the National Science Foundation (NSF) was to provide access, both intellectual and physical, to the large body of information in various multimedia formats. The project sought to use agents as the underlying architecture to provide this access, and to perform a wide variety of administrative and organizational tasks. One of the justifications for using agent architecture was that by moving administrative tasks previously accomplished by humans into the system, librarians were freed to do what they do best: organize, evaluate and catalog material

(Birmingham, 1995).

11 The UMDL had a difficult goal to accomplish – to facilitate the user in terms of the search and retrieval of information, to allow for the acquisition of new resources in a timely and efficient manner, and to integrate these tasks in such a way as to be scalable to accommodate future growth. This goal demanded a stable environment that would be flexible and change as content, publishers, user need and user habits changed, and the UMDL set out to meet this goal with an architecture based on distributed agents (Wellman et al., 1996). The agents at work in the UMDL process user searches and display the results, filter large quantities of information, monitor usage patterns, and pass information on to other agents for further processing (Birmingham et al.,

1994). The system uses a number of different agent types - the user interface agent, the mediator agent and the collection interface agent, each customized for the digital library environment. The user interface agent is responsible for making sure information about the user is gathered and sent to the appropriate guide agent, ensuring that any searches performed by the user are efficient. The mediator agent facilitates all aspects of the user’s query, from referring the initial request to the proper collection to transmitting the results of a search. Collection interface agents make sure that all the details related to the digital library’s collections are grouped and published in order to be used by other agents.

The UMDL remains one of the most ambitious initial digital library projects, and the project outcomes provided the foundation for the UMDL of today.

Agent Technology and the Hybrid Library: Project MALIBU

In 1998 the United Kingdom-based Joint Information Systems Committee

(JISC)/Electronic Libraries Programme (eLib) funded a project known as MALIBU – Managing the

Hybrid Library for the Benefit of Users. The hybrid library is defined by Rusbridge (1998) as providing access to both digital and non-digital resources within a common information framework, and using different technologies from the digital library world to integrate access to all four different kinds of resources (legacy, transition, new, and future) across different media.

MALIBU’s goals were to develop models for management and organization of the hybrid library, one of which was an agent-based search engine that allowed for the search and retrieval of

12 information from disparate resources. The prototype was deployed at three academic institutions in the United Kingdom: King’s College London, Oxford University and Southampton University.

The MALIBU search engine prototype was designed to provide users with a way to search heterogeneous, distributed resources using one interface. A single, web-based interface gave users access to a variety of resources, including the Wilfred Owen Multimedia Archive at

Oxford University, the Refugee Studies Centre Library catalog at Oxford University, the Papers of the First Duke of Wellington at Southampton University, and local library catalogs at Kings

College London, Southampton University and Oxford University. External resources included a variety of JSTOR collections, Google, and the Arts and Humanities Citation Index. Each of these resources were referred to as “targets”. The resources themselves included a variety of formats such as images, video, sound and multiple types of digital data (including structured metadata).

The prototype allowed for a variety of search types (asynchronous, prioritized synchronous and linked) by creating agents that communicated with each target, and used the relevant protocol

(HTTP or Z39.50) to search and retrieve information for the user. The use of agents allowed the end-user the convenience of searching on-line databases, web resources, internal and external resources during a single search event. The interface was sparse in its design, and basic features included keyword, author, and title searches, profile management, (this feature allowed the user to select only those targets which might be most suited to their research needs), and exportation of results via email, HTML, plain text, RTF or RDF.

The multi-agent architecture used for the MALIBU search engine was relatively straightforward. A central communication agent enabled agents to enter into dialogue with one another, verifying user requests and information exchange, and routing messages to the correct agent facilitator, while query agents were used to communicate with each target. Like the UMDL, there were several intrinsic and highly specialized features that made multi-agent architecture a logical choice for the MALIBU search engine prototype, including ease of distributed development, ease of system maintenance, and the ability of the agents to interact with each other on behalf of the user within the system (Dent, Harris, Hall and Martinez, 2001).

13 One of the features common to many systems where agents are at work is the need for a common vocabulary allowing agents to communicate with one another. Such a vocabulary is known is an ontology. Finin, Labrou and Mayfield (1997) suggest that the use of agents requires more than just a communication language, that there must also be some “conceptualization of the world that is embodied in concepts, distinctions, and so forth, in a formal representation scheme”

(p. 296). This scheme (or ontology) allows agents to interact with each other and the user within a well-defined framework. For instance, the ontology used by MALIBU allowed one agent that received a request from a user (to access a certain digital resource, for instance) to contact another agent to obtain certain permissions for that user. The agent receiving the communication would understand the request because of the common vocabulary, and answer accordingly. [a sentence was removed here – it said “note project outcomes.”]

Project MALIBU presented an integrated approach to the use of agent technology which highlights many of the strengths that might be applied within any hybrid library setting.

DAFFODIL: User-Friendly Access to Digital Libraries

DAFFODIL (Distributed Agents for User-Friendly Access of Digital Libraries), a project initially funded in 2000 by the German Science Foundation, is an agent-based system that allows users to search groups of scientific digital library collections at once. The initial prototype, designed for computer science and currently in use at the Institute of Informatics and Interactive

Systems at the University of Duisburg-Essen in Germany, allows for the integrated searching of more than eighteen digital collections and other resources (including ACM digital library, Springer,

Google, GetInfo, HCIBib Human Computer Interaction Resources, the Collection of Computer

Science Bibliographies (Achilles), the Digital Bibliography & Library Project (Universitäte Trier in

Germany), the Directory of Open Access Journals, the Scirus scientific resources search engine, and Cornell University’s ArXIv online database). DAFFODIL’s design allows for strategic, efficient searches, facilitated by agents, across these heterogeneous sources. One of the main developmental goals of the system was to overcome the differences in the interface layout, search functionality and query languages of these individual digital libraries, each of which has the potential to confuse and frustrate novice and experienced users alike. DAFFODIL unifies not

14 only the interfaces, but also the metadata and actual resources, “DAFFODIL produces synergies while exploiting different digital libraries: information from different digital libraries about the same document in different digital libraries is merged” (Govert, Fuhr and Klas, 2000, p. 1).

The strength of the DAFFODIL model is its ability to offer the user support for both low and high-level information seeking. Low-level information seeking is mostly comprised of

“moves”, as defined by Bates (1990). These basic moves might include adding a keyword to a search, or following a link. The relevant DAFFODIL tools include a personal library and interactive tools such as a “Did you mean…” feature that checks the search terms in a query and makes suggestions/corrections.

High-level information seeking is characterized by Bates (1990) in three ways:

1. Tactics - Include strategies such as broadening or narrowing a search

2. Stratagems - Combine basic moves and tactics to further refine a query, for instance,

browsing resources within a given domain such as a journal title, or browsing resources

by a certain author

3. Strategies - Combine moves, tactics, and stratagems to execute a comprehensive

Fuhr, Klas, Schaefer and Mutschke (2002) assert that while “typical information systems only support low-level search functions (so-called moves)”, DAFFODIL also supports high-level functions (p. 2). Many of DAFFODIL’s tools are provided at the stratagem level, and include: reference and citation management, journal and conference proceedings searching, author searches, a classification tool (presents a topic and domain-based representation scheme), and a thesaurus tool. The researchers describe “tactics, stratagems and strategies as plans by which the respective target can be reached”, and the provision of these search functions enable the user to make the most of each search event. There are five layers that comprise DAFFODIL’s platform.

1. Layer 0 - Targets or digital libraries connected to the system

2. Layer 1 – The moves layer, which includes a single interface that serves as a

bridge between the user and the digital collections in Layer 0

15 3. Layer 2 - The tactics level, where the different types of information objects found

in the digital libraries are combined

4. Layer 3 - The stratagems level, where agents facilitate specific search types

within a given domain

5. Layer 4 - The strategies level. Here the system provides support and guidance to

users as they combine the steps above and execute comprehensive information

search activities. (p. 3).

DAFFODIL is a practical model for other agent-supported digital library systems designed to provide a personalized searching environment for the user. The distributed environment of digital collections on the Web made agent architecture a natural choice for DAFFODIL, the UMDL and MALIBU. In each of these cases, agent technology provided functionality that allowed for mediation between the user and information in some form. One additional feature worth highlighting is the fact that librarians and library researchers were key players in the design, testing and implementation for each of these successful projects.

An Agent-Based Personalized Library Model for an Urban Academic Library

Librarians do not have to work hard to imagine a practical use for agent technology in their libraries. Research by Schaefer, Jordan, Klas and Fur (2005) found that “users mostly employ only simple or non-effective strategies and don’t know how to escape from critical situations, or how to start a search when the search goal is unclear or only vaguely defined. They often don’t even realize their strategic problems and missed search opportunities” (p.1). In keeping with this and other research that suggests that even savvy users need assistance with research, librarians at Hunter College Library in New York City designed a conceptual, agent- based personalized research module (Figure 1).

TAKE IN FIGURE 1

They used observation of initial library activities, tasks and reference questions of first-year students, and their first attempts to write a research paper for a basic English composition course to provide a contextual environment for the model.

16 Hunter College is an urban “vertical” campus with a large, diverse population of first year students, most uninitiated academic library users. The module, based on a review of professional literature from French and Viles (1999), Maes (1997), Jafari (2002), Baylor (2001), Malone, Lai, and Grant (1997), Glance (2001), Zick (2000), Jansen and Pooch (2004), and Dent, Harris, Hall and Martinez (2001), was developed to support student users by guiding them through the library research process, providing basic, on-demand information literacy instruction, and practical support like facilitating interlibrary loan and locating electronic course reserves. Librarians determined that a functional version of the personalized environment would need to support a number of library and research-related tasks, including:

1. Finding local resources

2. Finding non-local resources

3. Navigating online databases

4. Formulating good search queries

5. Sorting and prioritizing search results

6. Taking advantage of virtual on-demand “just in time” bibliographic instruction

7. Supporting compatibility between library technologies and tools, and portable devices

such as PDAs, cell phones

8. Personal notification of new relevant resources and services

9. Finding material related to classes, such as electronic reserves

10. Understanding the library’s physical layout

The role of agents within this system would vary. As in Detlor and Arsenault’s model

(2002), interface agents would greet the student when they log on, and personalize the information on the screen according to the student’s user profile information. What each student sees might be totally different: the college catalog and certain databases might be featured prominently for one student, whereas another student’s interface might include the latest additions to the print collection in Art History, or news about an upcoming drop-in research session. Information agents might maintain a student profile database and be responsible for retrieving and storing relevant information from a variety of internal and external sources, such as

17 circulation data, demographic information, and course registration information. Server agents might negotiate for access to various digital collections on behalf of the student, and facilitate requests for access. This type of integration would, for instance, support direct links to the appropriate electronic reserves for a student based on the classes the student is taking. Jansen and Pooch (2004) discuss an agent that provides customized research assistance. In the Hunter model, this research/instruction agent might observe students’ online searches, intervene to suggest more accurate keywords, and provide the option for on-demand instruction if a student seems to be completely lost while using a certain database. These same agents might also facilitate community learning and interaction between students by identifying good search strategies done by others, and linking those students who are researching similar topics (with the user’s permission), as detailed in research by Shang, Shi and Chen (2001), and Baylor (1999).

Hunter College Library provides a virtual tutorial and diagnostic that each incoming first year student must take, and librarians were also interested in integrating this information into the personalized research module. Agents might identify first year students as they log on, retrieve grades from the diagnostic, compare the results with the students’ current use of the system

(places they repeatedly had trouble, when they asked for help, etc.) and route this information electronically to reference librarians who might later customize in-person library instruction for that student accordingly. Agents might also use this same strategy to identify trends in certain classes/groups of students – making live instruction even more targeted. In addition, agents might suggest certain tutorials – call number tutorials, database-specific tutorials, strategic tutorials on good keyword formulation – for the student to review, based on the information from the same diagnostic.

Basic functionality of the Hunter module might mimic that used in the MALIBU model

(Dent, Harris, Hall and Martinez, 2001), and include exportation of results from searches directly to students’ handheld devices, and references to citation management software such as

EndNote. The system might further assist the user by automatically completing and submitting an interlibrary loan form if a resource is not otherwise available.

18 Hunter librarians conceptualized a long list of additional functionalities that such a system might have, the most practical and useful are those mentioned above. Further development of the model would be done in collaboration with the library’s systems team.

Conclusion

Griffiths (1998) suggested the core characteristics of the information professional – to act as a guide in the face of uncertainty, to collaborate, prioritize and maintain flexibility in the face of changing goals, empower users and understand one’s organization and colleagues - are not so far from those of the information agent. The use of agent technology in libraries does however present certain challenges, as there are both technical and social implications that are not easily addressed. Researchers and developers advance that libraries are a perfect match for implementing agents in any number of scenarios, but librarians involved with reference, systems, technical services and instruction must be able to first define their work, then conceptualize appropriate uses of agent technology that will be of the most benefit to the end user.

Development and implementation is another matter to be considered. Though there are quite a few off-the-shelf agent building toolkits that might be used in the library setting, librarians must still be able to work in a programming environment, or collaborate closely with IT staff to implement agents. Library computing environments must be able to support agents, allow for ease of deployment, manipulation, and modification. Librarians must also understand fundamentally how agents work, and provide some background information about their function for the user. Librarians and others interested in the actual creation of agents from scratch must have an impressive range of skills, not limited to communications technology, reasoning, and agent communication languages. Laurel (1997) also suggests that “rapid prototyping techniques” be used to support ongoing development and evaluation of agent applications (p. 76). Using intelligent agents requires that practitioners monitor their performance, always with an eye towards improvement.

In the case of the digital library, collections would need to be substantial enough to make implementation of agent technology worthwhile. What would the impact on librarians and library staff be? Meaningful uses of intelligent agents for routine tasks would ideally free up library staff

19 to pursue more meaningful work, and also, add to the library’s technical profile. More involved use of agents requires careful consideration of the challenges and benefits to the user.

References

Adler, J., Satapathy, G., Manikonda, V., Bowles, B. and Blue, V. (2005), “A multi-agent approach to cooperative traffic management and route guidance”, Transportation Research, Vol. 39 No. 4, pp. 297-318.

Bates, M. (1990), “Where should the person stop and the information search interface start?”, Information Processing and Management, Vol. 26, pp. 575-591.

Baylor, A. (1999), “Intelligent agents as cognitive tools for education”, Educational Technology, Vol. 39 No. 2, pp. 36-40.

Baylor, A. (2001), “Permutations of control: Cognitive guidelines for agent-based learning environments”, Journal of Interactive Learning Research, Vol. 12 No. 4, pp. 403-425.

Birmingham, W. (1995), “An agent-based architecture for digital libraries”, D-Lib Magazine, Vol. 1 No. 1, available at: www.dlib.org/dlib/July95/07birmingham.html (accessed 11 January 2006).

Birmingham, W., Drabenstott, K., Frost, C., Warner, A., and Willis, K. (1994), “The University of Michigan Digital Library: This is not your father’s library”, in Proceedings of the First Annual Conference on the Theory and Practice of Digital Libraries, available at: http://www.csdl.tamu.edu/DL94/paper/umdl.html (accessed 11 January 2006).

Bond, A. and Gasser, L. (1988), Readings in Distributed Artificial Intelligence, Morgan Kaufmann, San Francisco, CA.

Bradshaw, J.M. (Ed.) (1997), Software Agents, MIT Press, Menlo Park, CA.

Chen, J. and Su, S . (2003), “AgentGateway: A communication tool for multi-agent systems”, Information Sciences, Vol. 150, pp. 153-164.

Dent, V., Harris, S., Hall, W., and Martinez, K. (2001), “Agent technology concepts in a heterogeneous distributed searching environment”, VINE: The Journal of Information and Knowledge Management Systems, Vol. 31 No. 2, pp. 55-63.

Detlor, B. and Arsenault, C. (2002), “Web information seeking and retrieval in digital library context: Towards an intelligent agent solution”, Online Information Review, Vol. 26 No. 6, pp. 404- 412.

Finin, T, Labrou, Y. and Mayfield, J. (1997), “KQML as an agent communication language”, in Bradshaw, J.M. (Ed.), Software Agents, MIT Press, Menlo Park, CA.

Flynn, D. and Dodd, G. (2006), “Autonomous systems and intelligent agents in power system control and operation”, in Rehtanz, C.(Ed.), Control Engineering Practice, Springer, Berlin.

Foner, L. (1993), “What’s an agent anyway? A sociological case study”, Agents Memo Media Lab MIT, pp. 93-101.

Franklin, S. and Graesser, A. (1996), “Is it an agent, or just a program? A taxonomy for autonomous agents”, in Proceedings of the Third International Workshop on Agent Theories, Architectures and Language, Springer Verlag, Berlin, available at http:www.msci.memphis.edu/~franklin/AgentProg.html (accessed Janaury 3, 2006).

20 French, J. and Viles, C. (1999), “Personalized information environments: An architecture for customizeable access to distributed digital libraries”, D-Lib Magazine, Vol. 5 No. 6, available at: http://www.dlib.org/dlib/jun99/french/06french.html (accessed 26 January 2006).

Fuhr, N., Klas, C., Schaefer, A., and Mutschke, P. (2002.), “DAFFODIL: An integrated desktop for supporting high-level search activities in federated digital libraries”, in Research and Advanced Technology for Digital Libraries, Proceedings of the European Conference on Digital Libraries (ECDL 2002), Springer Verlag, Berlin, available at: http://www.is.informatik.uni- duisburg.de/bib/pdf/ir/Fuhr_etal:02.pdf (accessed 15 January 2006).

Glance, N. (2001), “Community Search Assistant”, Proceedings of the 6th International Conference on Intelligent User Interfaces, pp. 91-96.

Govert, N., Fuhr, N., and Klas, P. (2000), “DAFFODIL: Distributed agents for user-friendly access of digital libraries”, in Research and Advanced Technology for Digital Libraries. Proceedings of the 4th European Conference on Digital Libraries (ECDL 2000), Springer Verlag, Berlin, available at: http://www.is.informatik.uni-duisburg.de/bib/pdf/ir/Goevert_etal:00.pdf (accessed January 4, 2006).

Griffiths, J. (1998), “The new information professional”, Bulletin of the American Society for Information Science, Vol. 24 No. 3, pp. 8-11.

Hostler, E., Yoon, V., and Guimaraes, T. (2005), “Assessing the impact of internet agents on end users’ performance”, Decision Support Systems, Vol. 41, pp. 313-323.

Jafari, A. (2002), “Conceptualizing intelligent agents for teaching and learning”, Educause Quarterly, Vol. 3, pp. 28-34.

Janca, P. (1996), “Intelligent agents: Technology and application”, GIGA Information Group.

Jansen, B. and McNeese, M. (2005), “Evaluating the effectiveness of and patterns of interactions with automated searching assistance”, Journal of the American Society for Information Science and Technology, Vol. 56 No. 14, pp. 1480-1503.

Jansen, B. and Pooch, U. (2004), “Assisting the searcher: Utilizing software agents for web search systems”, Internet Research: Electronic Networking Applications and Policy, Vol. 14 No. 1, pp. 19-33.

Jansen, B. (2005), “Seeking and implementing automated assistance during the search process”, Information Processing and Management, Vol. 41, pp. 909-928.

Laurel, B. (1997), “Interface agents: Metaphors with character”, in Bradshaw, J.M. (Ed.), Software Agents, MIT Press, Menlo Park, CA.

Lee, R. and Liu, J. (2004), “iJADE WeatherMAN: A weather forecasting system using intelligent multiagent-based fuzzy neuro network”, IEEE Transactions on Systems, Man and Cybernetics Part 3: Applications and Reviews, Vol. 34 No. 3, pp. 369-377.

Lewis, J. (1996), “Netscape gets serious about infrastructure”, The Burton Group.

Lieberman, H., Rosenzweig, E., and Push, S. (2001), “Aria: An agent for annotating and retrieving images”, IEEE Computer, July, pp. 1- 8.

21 Maes, P. (1997), "Agents that reduce work and information overload", in Bradshaw, J.M. (Ed.), Software Agents, MIT Press, Menlo Park, CA, pp. 145-164.

Malone, T., Lai, K., and Grant, K. (1997), “Agents for information sharing and coordination: A history and some reflections”, in Bradshaw, J.M. (Ed.), Software Agents, MIT Press, Menlo Park, CA, pp. 109-144.

Miller, M. (1997), “Software Agents”, CHI Proceedings, available at: www.acm.org/sigchi/chi97/proceedings/tutorial/mm.htm (accessed 12 January 2006).

Moore, M. (2004), “Using an automated knowledge agent for reference and customer service”, Journal of the Medical Library Association, Vol. 92 No. 2, pp. 271-273.

Nardi,B. and O’Day,V. (1996), “Intelligent agents: What we learned at the library”, Libri, Vol. 46 No. 2, pp. 59-88.

Negroponte, N. (1995), Being Digital, Alfred Knopf, NY. . Nordman, M. and Lehtonen, M. (2005), “Distributed agent-based state estimation for electrical distribution networks”, IEEE Transactions on Power Systems, Vol. 20 No. 2, pp. 652-658.

Nordlie, R. (1999), “User revealment-A comparison of initial queries and ensuing question development in online searching and in human reference interaction”, in Proceedings of SIGIR99, Berkeley, CA.

Norman, D. (1997), "How might people interact with agents?", in Bradshaw, J.M. (Ed.), Software Agents, MIT Press, Menlo Park, CA, pp. 49-56.

Nwana, H., and Ndumu, D. (1998), “A brief introduction to software agent technology”, in Wooldridge, M. and Jennings, N. (Eds), Agent Technology: Foundations, Applications, and Markets, Springer Verlag, Berlin, pp. 29-48.

Nwana, H. (1996), “Software agents: An overview”, Knowledge Engineering Review, Vol. 11 No. 3, pp. 205-244.

Park, S., Lee, J., and Shin, B. (2005), “Introduction to development of e-Engineering framework for the automotive module design”, Proceedings of the 9th International Conference on Computer Supported Cooperative Work in Design, Vol. 1, pp.146-151.

Perez, E. (2002), “Web search agents: One-stop shopping for researchers”, Online, March/April, pp. 21-25.

Petrie, C. (1996), “Agent-based engineering, the Web and Intelligence”, IEEE Expert, Vol. 11 No. 6, pp 24-29.

Roesler, M. and Hawkins, D. (1994), "Intelligent agents; software servants for an electronic information world (and more!)", Online, Vol. 18 No. 4, pp. 18-29.

Rusbridge, C. (1998), “Realizing the Hybrid Library”, D-Lib Magazine, Vol. 4 No. 9, available at: http://www.dlib.org/dlib/october98/10pinfield.html (accessed 11 January 2006).

Schaefer, A., Jordan, M., Klas, C., and Fuhr, N. (2005), “Active support for query formulation in virtual digital libraries: A case study with DAFFODIL”, in Rauber, A., Christodoukalis, C., Tjoa, A. (Eds.), Research and Advanced Technology for Digital Libraries, Proceedings of the European

22 Conference on Digital Libraries (ECDL 2005), Springer Verlag, Berlin, available at: http://www.is.informatik.uni-duisburg.de/bib/pdf/ir/Schaefer_etal:05.pdf (accessed 10 January 2006).

Schlieffer, R. (2005), “An intelligent agent model”, European Journal of Operational Research, Vol. 166, pp. 666-693.

Shang, Y., Shi, H., and Chen, S. (2001), “An intelligent distributed environment for active learning”, ACM Journal of Educational Resources in Computing, Vol. 1 No. 2, pp. 1-17.

Shoham, Y. (1997), "An overview of agent-oriented programming", in Bradshaw, J.M. (Ed.), Software Agents, MIT Press, Menlo Park, CA, pp. 271-290.

Sundsted, T. (1998), “An introduction to agents”, JavaWorld, available at: www.javaworld.com/javaworld/jv-06-1998/jw-06-howto_p.html (accessed 11 January 2006).

Wellman, M., Durfee, E., and Birmingham, W. (1996), “The digital library as community of information agents”, UMDL Digital Library Expert Essay, available at: http://ai.eecs.umich.edu/people/wellman/pubs/expert96.html (accessed 20 January 2006).

Wooldridge, M. (1997), “Agent based software engineering”, IEEE Proceedings on Software Engineering, Vol. 144 No. 1, pp 26-37.

Wooldridge, M. and Jennings, N. (1998), Agent Technology: Foundations, Applications, and Markets, Springer Verlag, Berlin.

Xue, X., Xiaodong, L., Qiping, S., and Wang, Y. (2005), “An agent-based framework for supply chain coordination in construction”, Automation in Construction, Vol. 14 No. 3, pp. 413-430.

Zellermayer, M., Salomon, G., Globerson, T., and Givon, H. (1991), “Enhancing Writing-Related Metacognitions Through a Computerized Writing Partner”, American Educational Research Journal, Vol. 28 No. 2, pp. 373-391.

Zick, L. (2000), “The work of information mediators: A comparison of librarians and intelligent software agents”, First Monday, Vol. 5 No. 5, pp. 1-14, available at: http://firstmonday.org/issue5_5/zick/index.html (accessed 20 January 2006).