I am Dr Jean Lee,
an Associate Lecturer
at the School of Computing and
Information Systems (CIS)
of the University of Melbourne.

Dr Jean Lee is an Associate Lecturer at the School of Computing and Information Systems (CIS) at the University of Melbourne. She holds a PhD in Computer Science (NLP) and a Master of Data Science from the University of Sydney, supervised by Dr. Caren Han and Dr. Josiah Poon. Her current research focuses on AI in Education, Large Language Models, Multimodal Learning, and AI applications. Jean's research has been published in top-tier AI/NLP conferences (e.g. ACL, AAAI, SIGIR, NCAA, COLING), and she has participated in Google Research and NVIDIA workshops as a tutor/mentor. She also reviews for several international AI conferences and journals (e.g. IJCAI, AJCAI, ICPR, EMNLP, IEEE TCSS). Based on her successful research progress during her PhD, she was awarded a Research Training Program (RTP) Scholarship by the Australian Government. Prior to academia, Jean passed the U.S Uniform Certified Public Accountancy Examination (a.k.a. AICPA), received an MBA from Seoul National University, and worked in management consulting firms, including Accenture and KPMG.

Google Scholar | ORCID 0000-0002-7457-028X

Academic Talks

Overview of Large Language Models.

Part 1 - Background, Techniques, and Evolutionary Trends

Overview of LLMs Slide

About

Large language models (LLMs), like ChatGPT, have showcased remarkable capabilities in addressing various natural language processing (NLP) tasks, attracting significant attention across diverse domains. In this talk, I provided an overview of LLMs, including their background, techniques, and evolutionary trends. In the upcoming Part 2, I will summarize evaluation methods and applications of LLMs.

Publications / Preprints

order by date descending

Large Language Models in Finance (FinLLMs).

Accepted by NCAA (Q1) 2025.

Journal | Preprint | Github

Abstract

...Despite the extensive research into general-domain LLMs, and their immense potential in finance, Financial LLM (FinLLM) research remains limited. This survey provides a comprehensive overview of FinLLMs, including their history, techniques, performance, and opportunities and challenges...

M3-VRD: Multimodal Multi-task Multi-teacher Visually-Rich Form Document Understanding.

Accepted by ACL 2024.

Paper | Github

Abstract

...designed to leverage insights from both fine-grained and coarse-grained levels by facilitating a nuanced correlation between token and entity representations, addressing the complexities inherent in form documents. Additionally, we introduce new inter-grained and cross-grained loss functions to further refine...

StockEmotions: Discover Investor Emotions for Financial Sentiment Analysis and Multivariate Time Series.

Accepted by AAAI 2023 Bridge.

Paper | Github | PaperWithCode

Abstract

...a new dataset for detecting emotions in the stock market that consists of 10k English comments collected from StockTwits. Inspired by behavioral finance, it proposes 12 fine-grained emotion classes that span the roller coaster of investor emotion. Unlike existing financial sentiment datasets, StockEmotions presents granular features such as investor sentiment classes, fine-grained emotions, emojis, and time series data....

K-MHaS: A Multi-label Hate Speech Detection Dataset in Korean Online News Comment.

Accepted by COLING 2022.

Paper | Presentation | Github

Abstract

Online Hate speech detection has become important with the growth of digital devices, but resources in languages other than English are extremely limited. We introduce K-MHaS, a new multi-label dataset for hate speech detection that effectively handles Korean language patterns. The dataset consists of 109k utterances from news comments and provides multi-label classification...

CONDA: a CONtextual Dual-Annotated dataset for in-game toxicity understanding and detection.

Accepted by ACL-IJCNLP 2021.

Paper | Presentation | Github

Abstract

Traditional toxicity detection models have focused on the single utterance level without deeper understanding of context. We introduce CONDA, a new dataset for in-game toxic language detection enabling joint intent classification and slot filling analysis, which is the core task of Natural Language Understanding. We propose a robust dual semantic-level toxicity framework...

FedNLP: An interpretable NLP System to Decode Federal Reserve Communications.

Accepted by SIGIR 2021.

Paper | Presentation | Demo | Github

Abstract

The Federal Reserve System plays a significant role in affecting monetary policy and financial conditions worldwide. ...we present FedNLP, an interpretable multi-component Natural Language Processing system to decode Federal Reserve communications. This system is designed for end-users to explore how NLP techniques can assist their holistic understanding of the Fed's communications with NO coding...

Tutoring & Workshop Experience

Worked as a Casual Academic (TA, RA) at the University of Sydney in following units and workshops: Award Recipient for Outstanding Achievement in Teaching - Feedback for Teaching (FFT) Student Survey (2023)

• BUSS6002 Data Science in Business (Postgraduate course, 2023-2024)
• QBUS6600 Data Analytics for Business Capstone (Postgraduate course, 2024)
• COMP5046 Natural Language Processing (Postgraduate course, 2021-2022)
• ENGG1810 Introduction to Engineering Computing (Undergraduate course, 2022)
• DATA1002 Informatics: Data and Computation (Undergraduate course, 2022)
• NVIDIA LLM Bootcamp with CSIRO's National AI Centre (Workshop Tutor, 2024)
• GoogleExploreCSR @ USYD (Workshop Tutor, 2021-2022)
• Local Arrangement Committee @ IJCAI2024 (Student Volunteer Chair, 2024)
• Toxic Language Detection @ AJCAI2022 (Workshop Organisor, 2022)

Professional Experience

Leveraging with my working experience and data science skills, my goal is to pursue my research career in AI specialising in NLP. A brief summary of my career is below: