Thinh Hung Truong

Thinh Hung Truong

Auto-generated profile, based on publicly available information

Thinh Hung Truong

The University of Melbourne

Research Focus & Metrics

Academic Career Length

9Years (2018-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
5 August 2025 in medRxiv
Verified
Preprint
Open access
PDF file
TTThinh Hung Truong
Thinh Hung Truong
EFEmma Foster
Emma Foster
TFTimothy Fazio
Timothy Fazio
SHSarah Holper
Sarah Holper
Karin VerspoorKarin Verspoor
Karin Verspoor
Information extraction (IE) from specialized clinical texts such as brain MRI reports is important for various clinical and population health contexts. However, this topic is under-explored due to privacy concerns limiting data availability and the inherent complexity and domain-specificity of clinical language. Common methods relying on substantial amounts of training data fail. The recent advances in large language model (LLM) research provide a promising solution to bridge the data scarcity gap, with improved ability to adapt to novel tasks with little supervision. We introduce a new, challenging dataset of 100 expert-annotated brain MRI reports, featuring 152 fine-grained entity types and 4 relation types, characterised by low inter-annotator agreement. This task reflects the inherent complexity and real-world ambiguity of medical text. We evaluate a small, open-weight LLM across span detection, named entity recognition, and relation extraction tasks. We compare few-shot prompting and parameter-efficient fine-tuning against specialized off-the-shelf biomedical IE systems. Our results demonstrate that both few-shot and fine-tuned LLM approaches substantially outperform off-the-shelf baselines. While LLMs show superiority, absolute performance, particularly for complex relations and fine-grained entities, remains modest, correlating with the dataset’s inherent difficulty and the extreme low-resource setting.

5 November 2018 in IOP Conference Series: Materials Science and Engineering
Verified
Conference Paper
Open access
PDF file
ADAn T Dao
An T Dao
TTThinh H Truong
Thinh H Truong
LNLong Nguyen
Long Nguyen
DDDien Dinh
Dien Dinh
Named entities carry essential meanings and information in natural language. Therefore, Named Entity Recognition has many applications in different Natural Language Processing tasks such as Information Retrieval, Information Extraction, Machine Translation, and Question Answering. State-of-the-art Named Entity Recognition systems are based on supervised machine learning algorithms which require huge amounts of training data. The main problem, however, is that constructing named entity annotated corpora is an expensive, labor-intensive, and time-consuming task. Therefore, in this paper, we propose an approach to improve monolingual Named Entity Recognition systems by exploiting an existing unannotated English-Chinese bilingual corpus. The system jointly recognizes named entities in both English and Chinese sentences through the use of bilingual constraints. Experimental results show an improvement in Named Entity Recognition of both Chinese and English compared to the strong baseline StanfordNER. In particular, Chinese Named Entity Recognition improves significantly by 20.81% in term of F 1-score. As for the English language, Named Entity Recognition F 1-score increases slightly from 75.75% to 76.08%. When comparing to the state-of-the-art system in improving Named Entity Recognition based on bilingual resources, we manage to outperform in Chinese Named Entity Recognition task by 5.99% and achieve comparable results for the English side. Our proposed method can also be generalized to apply to resource-limited languages.

1 November 2018
Verified
Conference Paper
Closed access
TDTuan-An Dao
Tuan-An Dao
HTHung-Thinh Truong
Hung-Thinh Truong
LNLong Nguyen
Long Nguyen
DDDien Dinh
Dien Dinh
Bilingual corpora play an essential role in Cross-lingual Natural Language Processing tasks such as Machine Translation, Information Extraction, and Information Retrieval. Nevertheless, manually building such corpora is a time-consuming and expensive task, especially for resource-limited languages like Vietnamese. In this paper, we propose a symmetric Named Entity Alignment method to automatically construct a named entity annotated bilingual English-Vietnamese corpus. Our system uses expansion heuristics for candidate generation and bilingual features for candidate selection. The proposed system outperforms the state-of-the-art method in English-Vietnamese Named Entity Alignment, increasing F1-score from 82.68% to 86.57%. Moreover, Vietnamese Named Entity Recognition performance improves by 23.41% in terms of F1-score compared to Conditional Random Field model (StanfordNER). Our method can be generalized to develop corpora in other resource-limited languages.

7 September 2018
Verified
Conference Paper
Open access
PDF file
TTThinh Truong
Thinh Truong
ADAn Dao
An Dao
LNLong Nguyen
Long Nguyen
DDDien Dinh
Dien Dinh

9 May 2022 in arXiv
Verified
Preprint
Open access
PDF file
TTThinh Hung Truong
Thinh Hung Truong
TBTimothy Baldwin
Timothy Baldwin
TCTrevor Cohn
Trevor Cohn
Karin VerspoorKarin Verspoor
Karin Verspoor
Negation is a common linguistic feature that is crucial in many language understanding tasks, yet it remains a hard problem due to diversity in its expression in different types of text. Recent work has shown that state-of-the-art NLP models underperform on samples containing negation in various tasks, and that negation detection models do not transfer well across domains. We propose a new negation-focused pre-training strategy, involving targeted data augmentation and negation masking, to better incorporate negation information into language models. Extensive experiments on common benchmarks show that our proposed approach improves negation detection performance and generalizability over the strong baseline NegBERT (Khandewal and Sawant, 2020).