Pralekha: Cross-Lingual Document Alignment for Indic Languages

TL;DR

Pralekha introduces a large-scale Indic language document alignment benchmark and a novel chunk-based alignment metric, DAC, which improves efficiency and accuracy for parallel document mining in machine translation.

Contribution

The paper presents Pralekha, a new benchmark with over 3 million aligned document pairs for Indic languages, and proposes DAC, a fine-grained, efficient alignment metric that outperforms pooling-based methods.

Findings

DAC achieves 2-3x faster alignment

DAC outperforms pooling-based baselines in accuracy

MT models trained on DAC-aligned data perform better

Abstract

Mining parallel document pairs for document-level machine translation (MT) remains challenging due to the limitations of existing Cross-Lingual Document Alignment (CLDA) techniques. Existing methods often rely on metadata such as URLs, which are scarce, or on pooled document representations that fail to capture fine-grained alignment cues. Moreover, the limited context window of sentence embedding models hinders their ability to represent document-level context, while sentence-based alignment introduces a combinatorially large search space, leading to high computational cost. To address these challenges for Indic languages, we introduce Pralekha, a benchmark containing over 3 million aligned document pairs across 11 Indic languages and English, which includes 1.5 million English-Indic pairs. Furthermore, we propose Document Alignment Coefficient (DAC), a novel metric for fine-grained document alignment. Unlike pooling-based methods, DAC aligns documents by matching smaller chunks and computes similarity as the ratio of aligned chunks to the average number of chunks in a pair. Intrinsic evaluation shows that our chunk-based method is 2-3x faster while maintaining competitive performance, and that DAC achieves substantial gains over pooling-based baselines. Extrinsic evaluation further demonstrates that document-level MT models trained on DAC-aligned pairs consistently outperform those using baseline alignment methods. These results highlight DAC's effectiveness for parallel document mining. The dataset and evaluation framework are publicly available to support further research.