Implementing Powerful Retrieval Capabilities in a Distributed Environment for Libraries and Archives

An on-line distributed environment, which was implemented in the context of the VENIVA project for historical libraries and archives is presented here. Emphasis is given in the presentation of the powerful search capabilities provided to the end-users, which are typical for the end-users (either researchers or ordinary people) of any Digital Library environment. The information managed resides in a number of different relational databases in one or more institutions (i.e. Libraries and Historical Archives). The end-user of the system uses a WWW client to pose traditional boolean queries, similarity queries or complex queries containing both boolean and similarity terms on the contents of the databases. In the case of similarity queries, the end-user can also select the evaluation formula used to rank the objects that the system returns as the answer to his query. This gives a flexibility to experiment with alternative retrieval models without starting the implementation from scratch. A Graphical Query Editor is used in order to construct the queries. The innovative aspect of this work is that the similarity queries are translated by an appropriate component of the server into a series of traditional SQL queries, so that there is no need to have separate systems to support the various services offered. Only a standard relational DBMS (Database Management System) is used in the core of the system. The software layer that has been implemented on top gives all the additional flexibility. The implementation is based on a sound and flexible mathematical retrieval model. The framework, which supports queries containing both boolean and similarity terms in the context of VENIVA, is based on a powerful mathematical tool for the description of similarity queries and different evaluation models. It has been implemented on top of a relational DBMS as a distributed system. This framework aims to integrate Information Retrieval System (IRS) techniques on top of traditional relational DBMS. In particular, a mathematical model for the description of Information Retrieval Systems based on different fuzzy models has been developed [1]. The framework has been implemented [2] on top of a relational DBMS as a distributed system. The general architecture of this system is shown in Fig. 1. Schema editor is a graphical tool, which is used to define the mapping between fuzzy relations and relational tables. It is also responsible to create the relational tables that are used to store the fuzzy relations corresponding to similarity queries. The mappings are then stored in the Map File. Based on these mappings, the Retrieval Engine is able to translate retrieval requests from any Application into SQL queries sent to the Relational DB and give back the results