Agent-based Adaptive Selection and Interaction to Z39.50 Servers Camino Fern´ andez Ignacio Aedo Paloma D´ ıaz DEI - Laboratory Universidad Carlos III de Madrid Av. Universidad 30. Legan´ es E-28911 Spain camino,pdp @inf.uc3m.es, aedo@ia.uc3m.es Vicente Matell´ an Systems and Communications Group (GSyC) Universidad Rey Juan Carlos C/ Tulip´ an s/n. M´ ostoles E-28933 Spain vmo@gsyc.escet.urjc.es Abstract Digital libraries and OPACs are offering an enormous collection of information resources which can be accessed through Internet by quite different users thanks to the exis- tence of standard protocols for information retrieval such as Z39.50. However, some users may find difficulty in getting the items they are looking for in such a huge bulk of hetero- geneous resources since they have to know where and how they have to search for. Thus, one of the main reasons for the disregard of the Z39.50 service has traditionally been the absence of appropriate interfaces and the cost of mak- ing alternative searches. WAY-Z39.50 tries to help spread- ing the use of these servers by providing easy web access to them, as well as, by facilitating the access to multiple servers at once. The prototype described implements the WAY general model based on an agent-based architecture that provides adaptive searching in different servers using also an adaptive interface. Keywords: Z39.50, adaptive information access, adap- tive interface, adaptive searches, agent technologies, Java 1 Introduction Digital libraries and OPACs (Online Public Access Cat- alogue) are offering an enormous collection of information resources which can be accessed through Internet by quite different users thanks to the existence of standard protocols for information retrieval such as Z39.50 [19]. Z39.50 is a session-oriented and stateful network protocol whose devel- opment began in the early 80’s. It has evolved from its first version, Z39.50-1988, that was the basis of the WAIS [5] (Wide Area Information Systems) protocol, to the current one: Z39.50-1995 (Version 3). This protocol provides a unified access to structured information catalogs (mainly libraries’ catalogs) and supports sophisticated searches in such catalogs. However, some users may find difficult to get the items they are looking for in such a huge bulk of heterogeneous resources. First, users have to decide which are the most appropriate information servers for their queries and, there- fore, they need to know which are the available choices for every particular server. Second, users have to deal with dif- ferent interfaces, and even though the idea of providing a unified interface widely accessible, as the web is, has been a goal in the Z39.50 community [12], the interfaces to these servers have not been really accessible [18] or easy to use. Indeed, different servers provide different search fields and they also may use different languages to interact with the user. In order to solve these problems, several approaches can be taken dealing with such issues as multilinguality [8] and the use of adaptivity [21]. In this context, a web-browser accessible prototype named WAY-Z39.50 of adaptive system to access Z39.50 servers is presented. WAY-Z39.50 is an implementation of WAY model [9], which is a general model of architecture whose main goal is to improve information access systems through the fulfillment of three tasks: adaptation of the in- terface to the user, adaptation of the searching process and extraction and maintenance of updated information about the available information servers [4]. Next section describes the software architecture of WAY- Z39.50, making special emphasis on design an implemen-