Research Interest
The big picture of my research interests lies in
democratizing data intelligence to empower people and
organizations to derive insights, learn and share
knowledge, and build intelligence to turn data into
action.
Regardless of the various forms of data, understanding,
generation, and interaction are the three
common themes. Data understanding aims to achieve semantic
understanding of various types of data; Data generation
aims for automatic content generation based on users'
needs; and Data interaction aims to create unparalleled
user experiences working with data like recommendation or
information retrieval.
Specifically, I am interested in the following topics:
Development of Large Language Model
,
Semi-structured Data Modeling & Reasoning
,
Causal Inference
.
A few questions that drive my recent research are:
-
how can we get foundation models to efficiently learn
domain knowledge?;
-
how can we advance better models with humans'
collaborations?
-
how can we reduce potential harms (fairness, privacy
and bias)?
-
how can we genuinely advance our understanding of
current LLMs (capabilities and limitations), both
empirically and theoretically?
|
News
- [2024.11]: One co-authored paper is accepted
by KDD'25! Congrats to Yufei!
- [2024.09]: One paper is accepted by
EMNLP'24!
- [2024.08]: Invited to serve as a reviewer for
KDD'24 and ICLR'25!
- [2024.04]: Invited to serve as a reviewer for
NeurIPS'24!
- [2023.10]: One paper is accepted by WSDM'24!
Explore the Microsoft
Research Blog of our work!
- [2023.08]: Start my Ph.D. Journey at National
University of Singapore (NUS) with Ph.D. research
scholarship!
- [2023.03]: Honored to be involved in developing
the Excel
Copilot, which is the "moon-shot"
project of Microsoft!
- [2022.10]: Join MSRA, DKI Group as a research
intern!
- [2022.06]: Join Dartmouth College, Minds,
Machines and Society Lab as a research intern!
- [2022.05]: One paper is accepted by KBS
(journal)!
- [2022.03]: One paper is accepted by
IJCNN'22!
- [2022.02]: Join ICT, VIPL Group as a research
intern!
|
Publications
(selected, * refers to equal contribution)
|
Can Knowledge Graphs Make Large Language Models More
Trustworthy? An Empirical Study over Open-ended Question
Answering
Yuan Sui, Bryan Hooi
Preprint, 2024
arXiv
/
code
This paper presents OKGQA, a new benchmark for evaluating
Knowledge Graph-enhanced LLMs in open-ended question
answering, focusing on reducing hallucinations and
improving reasoning. It also introduces OKGQA-P to test
model performance with perturbed KGs. The work aims to (1)
explore whether KGs can make LLMs more trustworthy in an
open-ended setting, and (2) conduct a comparative analysis
to shed light on methods and future directions for
leveraging KGs to reduce LLMs' hallucination.
|
FiDeLiS: Faithful Reasoning in Large Language Model for
Knowledge Graph Question Answering
Yuan Sui, Yufei He, Nian Liu, Xiaoxin He, Kun Wang,
Bryan Hooi
Preprint, 2024
arXiv
/
code
This paper introduces FiDeLiS, a retrieval-exploration
interactive method that integrates knowledge graphs (KG)
with large language models (LLMs) to enhance reasoning
accuracy and reduce hallucinations. By utilizing the
Path-RAG module to recall relevant KG knowledge and
leveraging LLMsâ deductive reasoning for guiding the
reasoning process, FiDeLiS achieves more reliable and
efficient question answering performance.
|
UniGraph: Learning a Unified CrossâDomain Foundation
Model for TextâAttributed Graphs
Yufei He, Yuan Sui, Xiaoxin He, Bryan Hooi
31st ACM SIGKDD Conference on Knowledge Discovery and
Data Mining (KDD'25), 2024
arXiv
/
code
This paper introduces UniGraph, a framework for
generalizing graph learning across diverse domains using
Text-Attributed Graphs (TAGs).
It proposes a pre-training algorithm based on masked graph
modeling and graph instruction tuning to enable zero-shot
and few-shot learning on unseen domains.
Experiments on various tasks demonstrate that UniGraph
outperforms state-of-the-art methods in cross-domain graph
learning.
|
TAP4LLM: Table Provider on Sampling, Augmenting, and
Packing Semi-structured Data for Large Language
Model Reasoning
Yuan Sui, Jiaru Zou, Mengyu Zhou, Xinyi He, Lun
Du, Shi Han, Dongmei Zhang
Conference on Empirical Methods in Natural Language
Processing (EMNLP'24), 2023
arXiv
/
poster
/
slide
TAP4LLM presents a framework for effective table reasoning
by decomposing large tables, augmenting them with semantic
and statistical metadata, and intelligently packing the
information for LLM processing. This approach addresses
performance issues with huge tables and complex questions
by ensuring essential information is well-organized and
enriched.
|
Table meets LLM: Can Large Language Models
Understand Structured Table Data? A Benchmark and
Empirical Study
Yuan Sui, Mengyu Zhou, Mingjie Zhou, Shi Han and
Dongmei Zhang
Conference on Web Search and Data Mining (WSDM'24), Long
Paper, 2023
arXiv
/
code
/
blog
/
poster
/
slide
This study introduces a benchmark to evaluate the structural
understanding capabilities of LLMs on table data through
seven distinct foundation tasks. Evaluations on GPT-3.5 and
GPT-4 reveal that input formatting and structural prompting
significantly impact performance. The proposed
self-augmentation structural prompting methods improve LLM
performance on multiple tabular tasks, providing a
foundation for future research in table
comprehension by LLMs.
|
Why is Cross-Lingual Fine-Tuning Inferior
to Multi-Lingual Fine-Tuning? An Empirical
Study
Weicheng Ma, Junhwi Kim, Yuan Sui, Chunyuan Deng,
Lili Wang and Soroush Vosoughi
Preprint, 2023
arXiv
/
code
The paper investigates why cross-lingual fine-tuning
underperforms compared to multi-lingual fine-tuning,
It proposes target-language text-domain adaptation
and feature augmentation to enhance cross-lingual models,
effectively narrowing the performance gap. These methods
offer practical strategies for improving cross-lingual
fine-tuning, especially for low-resource languages.
|
Intelligent Predictive Maintenance of
Hydraulic Systems based on Virtual Knowledge Graph
Wei Yan, Yu Shi, Zengyan Ji, Yuan Sui, Zhenzhen
Tian, Wanjing Wang, Qiushi Cao
Engineering Applications of Artificial Intelligence
, 2023 (IF=8)
This research proposes a virtual knowledge graph-based
approach for the digital modeling and intelligent predictive
maintenance of hydraulic systems in manufacturing. By
integrating heterogeneous data from sensing networks and
structuring domain knowledge, the approach facilitates
effective data access, integration, and predictive
analytics in life-cycle of predictive maintenance.
|
Causality-aware Enhanced Model for Multi-hop
Question Answering over Knowledge Graphs
Yuan Sui, Shanshan Feng, Huaxiang Zhang, Jian
Cao, Liang Hu, Nengjun Zhu
Knowledge-Based Systems
(KBS), 2022 (IF=8.139)
The paper presents CF-KGQA, a causal filter model that
improves knowledge graph-based question answering by
addressing spurious relations through causal
interference in
the relation representation space. By employing a causal
pairwise aggregator and a disentangled latent factor
aggregator, CF-KGQA reduces erroneous relation
representations and enhances robustness on edge cases.
|
Trigger-GNN: A Trigger-Based Graph Neural Network for
Nested Named Entity Recognition
Yuan Sui, Fanyang Bu, Yingting Hu, Wei Yan,
Liang
Zhang
International Joint Conference on Neural Networks
(IJCNN'22), Long Paper, 2022 (oral)
Trigger-GNN introduces a graph neural network that leverages
entity triggers and complementary annotations to improve
nested named entity recognition. By encoding entity triggers
and utilizing an efficient message-passing architecture, the
model gains the capability to effectively identifies and
categorizes complex hierarchical
entity structures.
|
Academic Service
- Conference Reviewer: NeurIPS'24, KDD'25,
ICLR'25, AISTATS'25
- Journal Reviewer: Knowledge-based Systems
(KBS), Neural Computing and Applications
|
Teaching
- 2024/2025 semester 1, Knowledge Discovery and
Data Mining (CS54225/CS5425)
|
|