Machine Translation
Neural and LLM-based translation for low-resource and multilingual settings, document-level and discourse-aware translation, Portuguese–Chinese translation, and the evaluation and quality estimation of translation systems.
Research
From Portuguese–Chinese machine translation to large language models, evaluation and trustworthy AI.
Neural and LLM-based translation for low-resource and multilingual settings, document-level and discourse-aware translation, Portuguese–Chinese translation, and the evaluation and quality estimation of translation systems.
Training, alignment and instruction tuning of LLMs, data selection, reasoning from experience, uncertainty and self-evaluation, and benchmarks that measure what models really know.
Detection of LLM-generated text, model safety under adversarial prompting, bias and debiasing, hallucination mitigation, and explainable and controllable generation.
Multimodal machine translation, vision-language models and their evaluation, speech and affective computing, and representation learning across languages.
Agents for mathematics and science, medical and clinical NLP, personalised education, poetry translation and generation, and other interdisciplinary uses of language technology.
Large parallel corpora such as UM-Corpus and UM-pCorpus, the UM-CAT computer-aided translation platform, UM-NMT, dictionaries, morphological analysers and annotation toolkits.
Supported by the National Natural Science Foundation of China (NSFC), the Science and Technology Development Fund of Macau (FDCT), the University of Macau Multi-Year Research Grant (MYRG) and industry partners.
| Project | Funding | Period | Role |
|---|---|---|---|
| Research and Development of Large Language Models for Intelligent Service Platforms in Information Security | FDCT, Macao Funding Scheme for Key R&D Projects | 2025–2027 | Co-PI |
| Research on Chinese-Oriented Low-Resource Machine Translation Technologies | NSFC-FDCT, FDCT/060/2022/AFJ | 2023–2025 | PI |
| Modularized Semantic Neural Machine Translation: Fundamentals & Application | MOST-FDCT, FDCT/070/2022/AMJ | 2022–2025 | PI |
| Pretraining-based Automatic Evaluation of Machine Translation | MYRG, MYRG-GRG2023-00006-FST-UMDF | 2024–2025 | PI |
| Large Language Models for Machine Translation with Enhanced Discourse Modeling | Tencent AI Lab Rhino-Bird Research Program | 2023–2024 | PI |
| Research of Non-Autoregressive Neural Machine Translation | MYRG, MYRG2020-00054-FST | 2021–2023 | PI |
| Guiding Self-Attention Based Neural Machine Translation with Linguistic Prior Knowledge | FDCT, 0101/2019/A2 | 2019–2022 | PI |
| Research of Chinese-Portuguese Machine Translation Based on Deep Learning | NSFC-FDCT, FDCT/045/2017/AFJ | 2018–2020 | Co-PI |
| Research on Domain Adaptation for Low-Resource Chinese-Portuguese Machine Translation in Big Data Era: Methodology and Its Implementation | NSFC, 61702203 | 2017–2020 | PI |
| Morpho-Syntactic Modeling for Chinese-Portuguese Statistical Machine Translation | MYRG, MYRG2017-00087-FST | 2018–2020 | PI |
| Research on Linguistic Prediction Model with Graph-based Semi-Supervised Learning | MYRG, MYRG2015-00188-FST | 2015–2018 | PI |
| Heterogeneous Domain Adaptation Model for SMT | FDCT, 057/2014/A | 2015–2017 | PI |
| Domain Adaptation in Statistical Machine Translation Based on Graph Propagation | MYRG, MYRG2015-00175-FST | 2015–2018 | PI |
| Model-Driven Software Development from Requirement Description to Code Generation | FDCT, 103/2015/A3 | 2016–2019 | Co-PI |
| AuTema-ReOrgS – Automatic Reorganization of Structures | MYRG, MYRG2016-00109-FAH | 2017–2019 | Co-PI |
| Construction of Portuguese-Chinese Parallel Treebank (PCTreeBank) Based on Unsupervised Learning Approach | MYRG, MYRG076(Y1-L2)-FST13-WF | 2013–2016 | PI |
| Computational Psycholinguistic Models of Second Language Processing and Production | MYRG, MYRG064(Y1-L2)-FSH13-CCM | 2013–2016 | Co-PI |
| Blind Biosignal Classification (B2SC) Framework for a Multi-Function Health Alert Device | MYRG, MYRG070(Y1-L2)-FST12-CS | 2012–2015 | Co-PI |
| AuTema-Post-Editing (AuTema-PostEd) | MYRG, MYRG058(Y1-L2)-FSH12-ALL | 2012–2015 | Co-PI |
| Autema – SynTree – Annotation and Analysis of Bilingual Syntactic Trees for Chinese/Portuguese | MYRG, MYRG102(Y1-L2)-FSH11-ALL | 2011–2014 | Co-PI |
| Syntax Modeling & Parsing for Portuguese-Chinese Machine Translation and Establishment of Macau Key Research Laboratory in NLP & MT | FDCT, 057/2009/A2 | 2010–2013 | PI |
| Grammatical Inference for Analyzing the Second Language Acquisition of Spoken Portuguese | MYRG, RG060/09-10S/11R/CS/FST | 2010–2012 | Co-PI |
| Data Mining Model in Analyzing the Acquisition of Portuguese Knowledge for Macau Students | MYRG, RG054/08-09S/CS/FST | 2009–2010 | Co-PI |
| AuTema-Dis II | MYRG, RG035/09-10S/ALL/FSH | 2010–2011 | Co-PI |
| R&D of Knowledge-based Machine-Aided Bi-Directional Paragraph Translation in Portuguese-Chinese and English-Chinese | FDCT, 041/2005/A | 2006–2009 | PI |
The University of Macau's online computer-aided translation platform for Portuguese, Chinese and English.
Online neural machine translation for Portuguese and Chinese.
A large multi-domain English–Chinese parallel corpus for statistical machine translation (Tian et al., LREC 2014).
A large Portuguese–Chinese parallel corpus (Chao et al., LREC 2018).
The Portuguese–Chinese bidirectional electronic dictionary with more than 100,000 entries and trilingual pronunciation.
A machine-aided translation system for Portuguese and Chinese embedded in Microsoft Word, now superseded by UM-CAT.
An online parallel treebank annotation and construction platform with syntactic tree creation and parallel tree alignment (Xing et al., 2016).
Portuguese morphological variation analyser with a lemmatiser and verb conjugator.
English morphological variation analyser with a lemmatiser, inflector and verb conjugator.
A multilingual sentence boundary detection model based on the i+Learning incremental learning algorithm (Wong et al., 2014).
Implementations of the lab's papers, including Meta-Curriculum, Norm-Based Curriculum Learning, Uncertainty-Aware Curriculum Learning and Self-Paced Learning for NMT.
Best Paper Award, First Workshop on VLM4RWD at NeurIPS 2025 (VisAidMath).
University-level Teaching Excellence Award 2023/2024 and FST Teaching Excellence Award, University of Macau (Prof. Derek F. Wong).
Ranked 1st in Discourse-Level Literary Translation, Conference on Machine Translation (WMT 2023).
Yu Sheng: Human-in-Loop Classical Chinese Poetry Generation System — First Prize, 17th Pan-Pearl River Delta+ University IT Project Competition; Third Prize, 10th 'Sharing Cup' Innovation Competition.
Excellence Award, Tencent AI Lab Rhino-Bird Research Program.
Second Prize, Technological Invention Award, Macao Science and Technology Awards 2022 — Linguistics-Driven Neural Machine Translation Model and Its Application in Chinese-Portuguese Machine Translation.
Ranked 1st (multilingual and English–Russian) and 2nd (English–German and Chinese–English) in Quality Estimation, WMT 2022.
Ranked 1st in Evaluation Metrics for Chinese/Czech/German/Japanese–English, and 2nd for Hausa/Icelandic–English, WMT 2021 (with Alibaba DAMO Academy).
Ranked 1st and 2nd for Chinese–English newswire translation, 13th China Workshop on Machine Translation (CWMT 2017).
Second-Class Prize, Science and Technology Progress Award, 1st Macao Science and Technology Awards — Technologies and Applications of Portuguese-Chinese Machine Translation System.