PulseTrendingJournal ClubResearchersJournalsExplore
Instagram
HomeTrendingJournal ClubExplore
Synapse
⌘+K
Synapse
June 17, 2026International Journal of Pattern Recognition and Artificial Intelligence

Semantic Alignment–Driven Cross-Domain Knowledge Representation Learning with Large Language Models

View Full Paper
Ask AI
Bookmark
Share

Authors

JZJianing ZhangMWMeixian WangYYYan Yan

Discussion

Loading...

Member takes

Overview

Randomized trial demonstrates improved translation accuracy in low-resource domains, indicating enhanced semantic fidelity.

Key Points

  • This research aims to improve the accuracy of cross-domain neural machine translation by addressing semantic drift.
  • Developed a semantic alignment framework using multilingual LLM encoders.
  • Implemented a dual-view semantic alignment mechanism combining distance minimization and contrastive discrimination.
  • Introduced an adaptive alignment strategy to adjust the alignment strength during training.
  • Achieved significant improvements in BLEU, ChrF, METEOR, and COMET scores across various benchmarks.
  • Demonstrated particularly strong gains in datasets with high domain mismatch.
  • Enhanced semantic fidelity and robustness in neural machine translation.

Cite This Study

Zhang et al. (2026) studied this question.

synapsesocial.com/papers/6a323a2ad50b63ecad205603https://doi.org/10.1142/s0218001426510055
View Full Paper
Ask AI
Bookmark
Share