Noisy Parallel Data Alignment

Ruoyu Xie, Antonios Anastasopoulos
1/23/2023
cs.CLcs.LG

No official code found. Best implementation: ruoyuxie/noisy_parallel_data_alignment (7 stars, BSD-3-Clause, last commit 2023-03-04).

Abstract

An ongoing challenge in current natural language processing is how its major advancements tend to disproportionately favor resource-rich languages, leaving a significant number of under-resourced languages behind. Due to the lack of resources required to train and evaluate models, most modern language technologies are either nonexistent or unreliable to process endangered, local, and non-standardized languages. Optical character recognition (OCR) is often used to convert endangered language documents into machine-readable data. However, such OCR output is typically noisy, and most word alignment models are not built to work under such noisy conditions. In this work, we study the existing word-level alignment models under noisy settings and aim to make them more robust to noisy data. Our noise simulation and structural biasing method, tested on multiple language pairs, manages to reduce the alignment error rate on a state-of-the-art neural-based alignment model up to 59.6%.

View on arXivView PDF

Code Implementations(1)

ruoyuxie/noisy_parallel_data_alignmentCommunity100%
BSD-3-Clause

Enhanced awesome-align for low-resource languages and noise simulation: https://arxiv.org/abs/2301.09685

71PythonJan 27, 20233 years agoBSD-3-Clause
low-resource-languageslow-resource-nlpnoisenoisy-datanueral-machine-translation+4 more

Similar Papers

Word Alignment by Fine-tuning Embeddings on Parallel Corpora

81% match1 repo3882021

BinaryAlign: Word Alignment as Binary Sequence Labeling

81% match1 repo122024

Multilingual Word-Level Forced Alignment with Self-Supervised Representations and Learned Dynamic Programming

80% match2026

WSPAlign: Word Alignment Pre-training via Large-Scale Weakly Supervised Span Prediction

80% match2 repos122023

SOTAlign: Semi-Supervised Alignment of Unimodal Vision and Language Models via Optimal Transport

80% match2026

SimAlign: High Quality Word Alignments without Parallel Training Data using Static and Contextualized Embeddings

80% match1 repo3972020

Cite this paper

@article{xie2023noisy,
  title  = {Noisy Parallel Data Alignment},
  author = {Ruoyu Xie and Antonios Anastasopoulos},
  year   = {2023},
  eprint = {2301.09685},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2301.09685v2}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2301.09685.svg)](https://codeofpaper.com/paper/2301.09685)

Discussion