A Large-Scale Benchmark for Vietnamese Sentence Paraphrases

Sang Quang Nguyen, Kiet Van Nguyen
2/11/2025
cs.CL

No official code found. Best implementation: ngwgsang/ViSP (5 stars, last commit 2025-07-08).

Abstract

This paper presents ViSP, a high-quality Vietnamese dataset for sentence paraphrasing, consisting of 1.2M original-paraphrase pairs collected from various domains. The dataset was constructed using a hybrid approach that combines automatic paraphrase generation with manual evaluation to ensure high quality. We conducted experiments using methods such as back-translation, EDA, and baseline models like BART and T5, as well as large language models (LLMs), including GPT-4o, Gemini-1.5, Aya, Qwen-2.5, and Meta-Llama-3.1 variants. To the best of our knowledge, this is the first large-scale study on Vietnamese paraphrasing. We hope that our dataset and findings will serve as a valuable foundation for future research and applications in Vietnamese paraphrase tasks.

View on arXivView PDF

Code Implementations(1)

ngwgsang/ViSPCommunity100%
50Feb 2, 20251 years ago

Similar Papers

Crossing Linguistic Horizons: Finetuning and Comprehensive Evaluation of Vietnamese Large Language Models

83% match1 repo52024

Towards Comprehensive Vietnamese Retrieval-Augmented Generation and Large Language Models

82% match2024

VlogQA: Task, Dataset, and Baseline Models for Vietnamese Spoken-Based Machine Reading Comprehension

82% match1 repo12024

Coreference Resolution for Vietnamese Narrative Texts

81% match2025

Improving Vietnamese-English Medical Machine Translation

81% match2024

A Vietnamese Dataset for Text Segmentation and Multiple Choices Reading Comprehension

81% match2025

Cite this paper

@article{nguyen2025largescale,
  title  = {A Large-Scale Benchmark for Vietnamese Sentence Paraphrases},
  author = {Sang Quang Nguyen and Kiet Van Nguyen},
  year   = {2025},
  eprint = {2502.07188},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2502.07188v1}
}

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/2502.07188.svg)](https://codeofpaper.com/paper/2502.07188)

Discussion