A Study of Neural Matching Models for Cross-lingual IR

TL;DR

This paper explores neural matching models for cross-lingual information retrieval, evaluating various architectures and interaction methods using cross-lingual embeddings, aiming to develop an end-to-end CLIR system.

Contribution

It provides a comprehensive analysis of neural models and interaction strategies for CLIR, highlighting effective approaches for leveraging cross-lingual embeddings.

Findings

Neural models outperform traditional methods in CLIR tasks.

Different interaction representations significantly impact retrieval effectiveness.

Insights into word-pair similarity distributions inform model design.

Abstract

In this study, we investigate interaction-based neural matching models for ad-hoc cross-lingual information retrieval (CLIR) using cross-lingual word embeddings (CLWEs). With experiments conducted on the CLEF collection over four language pairs, we evaluate and provide insight into different neural model architectures, different ways to represent query-document interactions and word-pair similarity distributions in CLIR. This study paves the way for learning an end-to-end CLIR system using CLWEs.