On the Hallucination in Simultaneous Machine Translation

Meizhi Zhong, Kehai Chen, Zhengshan Xue, Lemao Liu, Mingming Yang +1 more
6/11/2024
cs.CL

No official code found. Best implementation: zhongmz/SiMT-Hallucination (2 stars, last commit 2024-06-01).

Abstract

It is widely known that hallucination is a critical issue in Simultaneous Machine Translation (SiMT) due to the absence of source-side information. While many efforts have been made to enhance performance for SiMT, few of them attempt to understand and analyze hallucination in SiMT. Therefore, we conduct a comprehensive analysis of hallucination in SiMT from two perspectives: understanding the distribution of hallucination words and the target-side context usage of them. Intensive experiments demonstrate some valuable findings and particularly show that it is possible to alleviate hallucination by decreasing the over usage of target-side information for SiMT.

View on arXivView PDF

Code Implementations(1)

zhongmz/SiMT-HallucinationCommunity100%

source code of paper "On the Hallucination in Simultaneous Machine Translation"

20C++, LuaJun 1, 20242 years ago

Similar Papers

Mitigating Hallucinated Translations in Large Language Models with Hallucination-focused Preference Optimization

85% match2025

Enhanced Hallucination Detection in Neural Machine Translation through Simple Detector Aggregation

85% match2024

Machine Translation Hallucination Detection for Low and High Resource Languages using Large Language Models

85% match1 repo42024

Cognitive Mirage: A Review of Hallucinations in Large Language Models

83% match1 repo492023

The Dawn After the Dark: An Empirical Study on Factuality Hallucination in Large Language Models

83% match2024

SLPL SHROOM at SemEval2024 Task 06: A comprehensive study on models ability to detect hallucination

83% match2024

Cite this paper

@article{zhong2024hallucination,
  title  = {On the Hallucination in Simultaneous Machine Translation},
  author = {Meizhi Zhong and Kehai Chen and Zhengshan Xue and Lemao Liu and Mingming Yang and Min Zhang},
  year   = {2024},
  eprint = {2406.07239},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2406.07239v1}
}

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/2406.07239.svg)](https://codeofpaper.com/paper/2406.07239)

Discussion