ECLAIR: an Extensible Class Library for Information Retrieval

Abstract

This paper describes ECLAIR, an extensible object-oriented class library, which can be used to construct information retrieval systems and applications. ECLAIR is implemented in C++ and uses an object-oriented database management system for persistent object storage. The library provides data structures and operations necessary to allow the efficient indexing, storage and retrieval of data items based on their text content using automatic indexing and best-match retrieval techniques. By exploiting the object-oriented paradigm, we have been able to develop a library which is usable by any application requiring information retrieval functionality, and further, is extensible such that many of the functions provided by the library can be changed as required by an application. Additionally, applications can exploit the features of the underlying object-oriented database management system, which include support for persistence of data (objects), modelling complex objects, concurrent access to data, and reliable processing of data in the presence of system failures. In the paper we consider how to deliver information retrieval functionality to the developers of application systems and we evaluate current approaches. The design and implementation of ECLAIR is described in detail. A number of experiments applying ECLAIR are reported.