# Cross-language Sentence Selection via Data Augmentation and Rationale Training

Source: [Cross-language Sentence Selection via Data Augmentation and Rationale Training](<https://www.sota2.com/research/paper/68fa6a1d1e209131cea25f84>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2025-12-24T15:58:07.536Z

Authors: Yanda Chen, Chris Kedzie, Suraj Nair, Petra Galu\\v{s}\\v{c}\\'akov\\'a, Rui Zhang, Douglas W. Oard, Kathleen McKeown

Published: 2021-06-04T07:08:47.000Z

## Abstract

This paper proposes an approach to cross-language sentence selection in a low-resource setting. It uses data augmentation and negative sampling techniques on noisy parallel sentence data to directly learn a cross-lingual embedding-based query relevance model. Results show that this approach performs as well as or better than multiple state-of-the-art machine translation + monolingual retrieval systems trained on the same parallel data. Moreover, when a rationale training secondary objective is applied to encourage the model to match word alignment hints from a phrase-based statistical machine translation model, consistent improvements are seen across three language pairs (English-Somali, English-Swahili and English-Tagalog) over a variety of state-of-the-art baselines.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Cross-language Document Retrieval on MATERIAL Tagalog 1.0 (Analysis)](<https://www.sota2.com/research/sota/cross-language-document-retrieval-on-material-tagalog-1-0-analysis.md>): Cross-language Document Retrieval; MATERIAL Tagalog 1.0 (Analysis); MAP: 61.1.

- [Cross-language Document Retrieval on MATERIAL Tagalog 1.0 (dev)](<https://www.sota2.com/research/sota/cross-language-document-retrieval-on-material-tagalog-1-0-dev.md>): Cross-language Document Retrieval; MATERIAL Tagalog 1.0 (dev); MAP: 59.

- [Cross-Language Document Retrieval on MATERIAL Somali (Analysis)](<https://www.sota2.com/research/sota/cross-language-document-retrieval-on-material-somali-analysis.md>): Cross-language Document Retrieval; MATERIAL Somali (Analysis); MAP (Text): 35.4.

- [Cross-Language Document Retrieval on MATERIAL Somali (dev)](<https://www.sota2.com/research/sota/cross-language-document-retrieval-on-material-somali-dev.md>): Cross-language Document Retrieval; MATERIAL Somali (dev); MAP (Text): 29.5.

- [Cross-Language Document Retrieval on MATERIAL Somali (Eval)](<https://www.sota2.com/research/sota/cross-language-document-retrieval-on-material-somali-eval.md>): Cross-language Document Retrieval; MATERIAL Somali (Eval); MAP (Text): 13.1.

- [Cross-Language Document Retrieval on MATERIAL Swahili (Analysis)](<https://www.sota2.com/research/sota/cross-language-document-retrieval-on-material-swahili-analysis.md>): Cross-language Document Retrieval; MATERIAL Swahili (Analysis); MAP (Text): 48.3.

- [Cross-Language Document Retrieval on MATERIAL Swahili (dev)](<https://www.sota2.com/research/sota/cross-language-document-retrieval-on-material-swahili-dev.md>): Cross-language Document Retrieval; MATERIAL Swahili (dev); MAP (Text): 40.3.

- [Cross-Language Document Retrieval on MATERIAL Swahili (eval)](<https://www.sota2.com/research/sota/cross-language-document-retrieval-on-material-swahili-eval.md>): Cross-language Document Retrieval; MATERIAL Swahili (eval); MAP (Text): 22.7.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2106.02293>)

[PDF](<https://arxiv.org/pdf/2106.02293.pdf>)
