PASCAL - Pattern Analysis, Statistical Modelling and Computational Learning

A Resource-Light Approach to Cross-Language Information Retrieval
Jean-Michel Renders, Eric Gaussier and Cyril Goutte
In: [rejected](2006).

Abstract

This paper aims at describing how the combination of light resources such as standard bilingual dictionaries and multilingual corpora can be formalized and exploited in the general and efficient framework of the Relevance Model (RM) approach to Cross-language Information Retrieval (CLIR). This combination of light resources is then compared with state-of-the-art CLIR methods and, particularly, those based on merging different Machine Translation systems and retrieval engines. It is shown experimentally that using an appropriate mix of resources extracted from standard dictionaries, general parallel corpora and (possibly) specialised comparable corpora in the RM framework allows to achieve performance at least equal to these current state-of-the-art methods.

EPrint Type:Conference or Workshop Item (Paper)
Project Keyword:Project Keyword UNSPECIFIED
Subjects:Natural Language Processing
Information Retrieval & Textual Information Access
ID Code:1476
Deposited By:Cyril Goutte
Deposited On:28 November 2005