About Me
I am a Ph.D. candidate in the School of Humanities at Tsinghua University. My major is computational linguistics, combining artificial intelligence and natural languages. I am interested in computational lexical semantics (including polysemy, ambiguity and uncertainty) and AI interpretability.
News
[July 2024] I have finished my defense of Thesis Proposal Defense. Thank you for the insightful comments from three committee members!
[June 2024] I will review one paper from ACL workshop of Rep4NLP.`
[June 2024] I submitted a paper of thesis proposal to ACL SRW! Expected to helpful comments.
[May 2024] The paper has been accepted by Findings: ACL 2024
[May 2024] I lead a reading group on computational linguistics. Hope we can learn something together.
[May 2024] I have applied CSC-funded the joint Phd program, cosupervised by Professor Roberto Navigli at Sapienza University of Rome
[Febrary 2024] One paper is under review by ACL 2024
[December 2023] I have received Excellent Comprehensive Scholarship of Tsinghua University (First Prize).
[December 2023] I have finished two Linguistic lessions (auditing).
[December 2023] I attended the CDH 2023 and the paper was awarded Outstanding Creative Paper Award on Interactive and automatic mural captioning.
[November 2023] One paper titled "Evaluation of the semantics and agential degree of Chinese subject-object reversible sentences based on large language model" was presented and awarded the first prize.
[July 2023] Five papers in EMNLP 2023 have been reviewed.
[June 2023] A talk titled "To know, or not to know? Language, Uncertainty and Artificial Intelligence" at Shanxi University of Finance and Economics.
[May 2023] One paper has been accepted by ACL findings 2023.
Education
Tsinghua University (PhD Candidate)
Sept 2022 - May 2026 (Expected)
Major Computational Linguistics
courses: Linguistic typology (by Rui Guo, auditing), Morphology (by XiuFang Dong, auditing), Probability and Random Mathematics (by Jun Ye, auditing), Natural Language Processing (by Yue Zhang, online auditing), Topics in Chinese Morphology and Syntax (by Dun Deng, A), Research on Corpus Linguistics (by Ying Liu, A), Research of Seclected Topics on Chinese Semantics (by Bing Qiu, A, Thesis: An Empirical Study on Semantic Relations within Chinese Compound Words based on Word Embeddings)
Other interesting courses: Appreciation of Western Opera (by Yi Ding, auditing), History of Western Music (by Xiao Kang, auditing), Appreciation of Peking Opera (by Mengmei Zhou)
Road to Thesis:
A Survey of Word Sense Disambiguation (Thesis of Comprehensive Examination, A)
Aspect of uncertainty in WSD (Conference paper)
Research proposal for thesis (RP) [May, 2024]
Thesis proposal defense for Final Thesis (Chinese RP, PPT) [July, 2024]
Southern University of Science and Technology (Master)
Sept 2019 - July 2022
Major: Computer Science
courses: Advanced Algorithm (A), Bayesian Data Analysis (A), Machine Learning (A)
Other interesting courses: Translation and Appreciation of Chinese Classical Poetry （by Mengwen Zhu, A), Classics and the Process of Canonization (by Mengwen Zhu, auditing)
Thesis: Towards Human-like Diverse Video Captioning Via a Latent Generative Model
Dalian University of Technology (Undergraduate)
Sept 2015 - July 2019
Major: Digital Media Techonology
Experiences
Research Intern in THUNLP, cosupervised by Maosong Sun. Reasearch focuses on LLM interpretability and semantic representation.
October 2023 -
Reviewer of EMNLP 2023
Publications & Projects
How well do distributed representations convey contextual lexical semantics: a Thesis Proposal
Arxiv, Under Review
[Paper]
Fantastic Semantics and Where to Find Them: Investigating Which Layers of Generative LLMs Reflect Lexical Semantics
Findings of ACL 2024
[Paper]
Ambiguity Meets Uncertainty: Investigating Uncertainty Estimation for Word Sense Disambiguation
Findings of ACL 2023
Show, Tell and Rephrase: Diverse Video Captioning via Two-Stage Progressive Training
TMM 2022
Talks
To know, or not to know? Language, Uncertainty and Artificial Intelligence
At Shanxi University of Finance and Economnics, June 12th, [PPT]
In this talk, I reviewed my recent research from a philosophical perspective of epistemology: What does the model not know? (Known Knowns) What does the model unknow? (Known Unknowns).
Evaluation agentiveness of Subject-object reversible sentences in LLMs
At a linguistic seminar, November 26th, [PPT]
In this talk, I employ difference vetors in LLMs to represent agentiveness of different components (SVO) in a reversible but meaning-preserving sentence pair.
Other Resources:
- A comprehensive repo about lexical ambiguity: Awesome Word Sense Disambiguation.