User Agents in Digital Libraries and Collaborative Learning Environments
J. Alfredo Sánchez, Gerardo Ayala
Laboratory of Interactive and Cooperative Technologies
Center for Research in Information and Automation Technologies
Universidad de las Américas-Puebla
{alfredo, ayalasan};

1. Introduction

Work on agents at the Laboratory of Interactive and Cooperative Technologies (ICT) of the Universidad de las Américas-Puebla has focused on three main classes of agents: agents as an abstraction used in the construction of collaborative learning environments, agents as autonomous entities that directly interact with users, and agents as processes that migrate (along with their state) between nodes in a network. We refer to these agent categories as programming agents, user agents and network agents, respectively [Sánchez 1996, 1997; Sánchez and Leggett 1997]. We are exploring the application of agents to address problems in managing large, complex and dynamic information spaces (such as digital libraries) and complex social processes (such as collaborative learning). In this paper we provide an overview of various of the agent-related projects we have undertaken.

2. Programming Agents

We have used the notion of agent to model some of the components in a Computer-Supported Collaborative Learning (CSCL) environment termed GRACILE [Ayala and Yano 1998]. CSCL environments should include software elements to assist learners in  applying domain knowledge and generating opportunities for effective collaboration and learning. In order to accomplish this, GRACILE has been constructed in terms of two subclasses of programming agents: domain agents and mediator agents. Since GRACILE is intended to support small groups of students learning a foreign language, domain agents have been designed as deliberative entities that are capable of building and analyzing sentences by applying a set of rules that represent language patterns and expressions. Mediator agents cooperate by exchanging beliefs about the the learners' capabilities and fostering collaboration among them. Some of the agency concepts first explored by GRACILE are now being applied to the construction of a distance learning environment we have called CASSIEL [Llarena 1998].

3. User Agents

The main characteristic of user agents is that they can be perceived by end users as entities with some degree of autonomy to which some well-defined task can be delegated. Most of our work in this area has been conducted in the context of a broad initiative termed The Floristic Digital Library (FDL). A major goal of the FDL initiative is to offer a wide range of services to facilitate the utilization and sharing of a highly distributed repository comprising textual descriptions and discussions, images, illustrations and maps relating to plants around the world. The National Science Foundation-sponsored Flora of North America (FNA) and Flora of China (FOC) projects are participating in this effort along with several related floristic research projects (the FNA project alone involves more than 800 scientists). We are exploring the use of agents in three general areas of the FDL: construction, information browsing and retrieval, and collaboration and communication.

3.1 Construction of the FDL

FDL Architecture. We have developed a system architecture for the floristic digital library that addresses the needs for communication, collaboration and information management among the botanical community members. The architecture for the Floristic Digital Library is an evolution of the system architecture presented by Sánchez et al. [1997] and Sánchez and Leggett [1997], in which agent services play a central role.

Inter-Communication Framework. FDL is by design a highly distributed architecture. Data, services and interfaces may reside practically on any location in the global network. We have defined a framework which will facilitate the communication among the various distributed components and the dynamic incorporation of new components into the architecture [Barceinas et al. 1998]. Given the emphasis of the FDL architecture on agent services, we have produced a prototype of this framework using the Knowledge Query Manipulation Language (KQML) [Finin et al. 1993], a popular inter-agent communication language, and we are working on a more general implementation based on CORBA [Otte et al. 1996].

Assistive environments for massive data entry. We have been working on environments to expedite the process of entering, editing and reviewing manuscripts authored by hundreds of geographically distributed scientists. One of our main efforts in this area is Chrysalis, an environment including a web-accessible set of tools for editing plant morphologic descriptions and user agents that actively participate in the construction of these descriptions by volunteering plant attributes based on similarities among the attributes being entered and taxonomic keys and descriptions already stored in the library [Sánchez et al. 1998a].

3.2 Information Browsing and Retrieval in the FDL

Much work in botany focuses on discussing diverse taxonomic viewpoints that structure existing information about plants in different  ways. Traditionally, however, floristic projects have adopted a single taxonomic point of view that has been agreed upon by an editorial committee. The digital medium makes it possible to provide users with means to traverse information spaces considering multiple taxonomic perspectives. We have designed and prototyped an environment, termed MUTANT (Multi-Taxonomy AgeNTs), in which agents representing differing taxonomic viewpoints serve as guides for users browsing the FDL [Flores 1997, Sánchez et al. 1998b].

Given the volume and complexity of the information spaces comprised by the FDL, we also have
developed personalized information retrieval mechanisms. User agents assist users in dynamically
selecting and filtering those information units that best suit their individual needs and preferences.
We have prototyped agents that perform these tasks on a single FDL node [Cabrera 1997] and mobile agents that traverse multiple nodes, gather information of interest and present it to the user [Pérez Lezama 1998]. For the information retrieval engines of these agents we have relied on well-established techniques such as the vector space and extended boolean models [Salton and McGill 1983].

3.3 Collaboration and Communication in the FDL.

We conceive of the FDL as a virtual collaboration and communication space for a wide range of users interested in plants. However, group awareness and collaboration are not easily achieved in a highly distributed environment such as the FDL's. We have developed Agora, an environment that enables group awareness and collaboration by extending existing facilities with recommendation and alerting agents, as well as various synchronous and asynchronous communication interfaces [Sánchez et al. 1998c]. Services in Agora include both content-based and collaborative recommendations [Balabanovic and Shoham 1997], awareness agents that notify users when experts or other users with overlapping interests are also in the library, and communication interfaces through which users can share information asynchronously or interact with other users in the library in real time.

4. Network Agents

As noted in Section 3.2, some of the work we have been doing on user agents overlaps with network agents in that some of the personalized information retrieval agents are mobile processes that traverse a collection of servers to gather information of interest for users. Prior to this work, we conducted a thorough survey of the field and produced a comparative study of various tools and environments available for the construction of network (or mobile) agents [Castellanos and Sandoval 1997].


This work is supported in part by grants from the Andrew W. Mellon Foundation, the National Science Foundation (DEB-9505383 and DEB-9626806), and the Research and Development Network in Informatics (REDII) of the Mexican National Council for Science and Technology (Conacyt).


Ayala, G. and Yano, Y. 1998. A collaborative learning environment based on intelligent agents. Expert Systems with Applications 14, 129-137.

Balabanovic, M., and Shoham, Y. 1997. Special issue on recommender systems. Communications of the ACM 40, 3 (March), 66-72.

Barceinas, A., Sánchez, J. A., and Schnase, J. L. 1998. MICK: A KQML inter-agent communication framework in a digital library. In Memorias del Simposium Internacional de Computación (CIC'98, Mexico City, November). Forthcoming.

Cabrera, J. 1997. Integración de servicios y agentes de usuario en la recuperación de información en una biblioteca digital. B.Sc. Thesis. Department of Computer Systems Engineering, Universidad de las Américas-Puebla, Cholula, Pue. 72820, Mexico. (

Castellanos, E. O., and Sandoval, E. 1997. Un estudio sobre los agentes móviles y su aplicación. B.Sc. Thesis. Department of Computer Systems Engineering, Universidad de las Américas-Puebla, Cholula, Pue. 72820, Mexico (http://ict/people/odette/html/Tesis.html).

Finin, T., Weber, J., Wiederhold, G., Genesereth, M., Fritzson, R., McGuire, J., Shapiro, S., and Beck, C. 1993.  Specification of the KQML agent communication language.  Tech.  Rep. EIT TR 92-04, Enterprise Integration Technologies, Palo Alto, Calif.

Flores, C. 1997. Agentes de usuario como guías en bibliotecas digitales. M. Sc. Thesis. Department of Computer Systems Engineering, Universidad de las Américas-Puebla, Cholula, Pue. 72820, Mexico. (

Llarena, J. 1998. Software de apoyo al facilitador en una red de aprendizaje en intranet. B.Sc. Thesis. Department of Computer Systems Engineering, Universidad de las Américas-Puebla, Cholula, Puebla.

Otte, R., Patrick, P., and Roy, M.. 1996. Understanding CORBA. Prentice Hall.

Pérez Lezama, C. 1998. Agentes móviles en bibliotecas digitales. M.Sc. Thesis. Department of Computer Systems Engineering, Universidad de las Américas-Puebla, Cholula, Pue. 72820, Mexico. (

Salton, G., and McGill, M. 1983.  Introduction to Modern Information Retrieval. McGraw-Hill, New York, N.Y.

Sánchez, J. A., López, C. A., and Schnase, J. L. 1998a. An agent-based approach to the construction of floristic digital libraries. In Proceedings of the Third ACM International Conference on Digital Libraries (DL'98, Pittsburgh, PA., July). 210-216.

Sánchez, J. A., Flores, C. A., and Schnase, J. L. 1998b. Mutant agents as guides for multiple taxonomies. Submitted to the 3rd. International Conference on Autonomous Agents (Agents'99, Seattle, forthcoming).

Sánchez, J. A., Fernández, L., and Schnase, J. L. 1998c. Agora: Enhancing awareness and collaboration in floristic digital libraries. In Proceedings of the Fourth CYTED-RITOS International Workshop on Groupware (CRIWG'98, Rio de Janeiro). 85-95.

Sánchez, J. A. 1997. A Taxonomy of agents. Tech. Rep. ICT-97-1. Laboratory of Interactive and Cooperative Technologies, Department of Computer Systems Engineering, Universidad de las Américas-Puebla. Cholula, Puebla, México.

Sánchez, J. A. and Leggett, J. A. 1997. Agent services for users of digital libraries. Journal of Networks and Computer Applications 21, 1, 45-58.

Sánchez, J. A., Leggett, J. J., and Schnase, J. L. 1997. AGS: Introducing agents as services provided by digital libraries. In Proceedings of the Second ACM International Conference on Digital Libraries (Philadelphia, PA., July), 75-82.

Sánchez, J. A. 1996. Agent Services. Ph.D. Dissertation. Department of Computer Science, Texas A&M University, College Station, Tex., August.