About

Lillian Lee is a professor who has been incredibly fortunate to advise a remarkable group of students and postdoctoral researchers throughout her career. Her research trajectory has been significantly influenced by the work and leadership of her students, to whom she expresses deep gratitude. Her students have achieved notable success, giving invited talks and interviews at prestigious institutions and companies worldwide, including Bar-Ilan, CMU, EPFL, Facebook, Google, the Max Planck Institute, Microsoft Research, MIT, Northeastern, Stanford, and Yahoo!, among others. This highlights her role in mentoring individuals who have gone on to make significant contributions in academia and industry. Her mentorship has guided students and postdocs who have received numerous awards and fellowships, such as the Siegel PiTech PhD Impact Fellowship, Best Paper Awards at major conferences like WWW, IJCAI, CSCW, and ACL, as well as prestigious fellowships like the Bloomberg Data Science Fellowship and the Sloan Fellowship. Many of her former students have secured faculty positions at leading universities including MIT, the Max Planck Institute, University of Colorado-Boulder, and the University of Pennsylvania, as well as influential roles in research labs at companies like Google, Yahoo!, and IBM Watson. Lillian Lee's research interests are reflected in the diverse and impactful theses of her students, which cover topics such as computational approaches to linguistic coordination, learning from multimodal web data, document representation, inter-document similarities, sentiment analysis, syntactic dependency analysis, social natural language processing, and computational approaches to understanding human behavior from online social interactions. Her work bridges natural language processing, computational linguistics, and social computing, emphasizing the analysis and understanding of language and behavior in digital contexts.

Research topics

• Artificial Intelligence
• Information Retrieval
• Natural Language Processing
• Computer Science
• Programming language
• Database

Selected publications

• Current Semantic-change Quantification Methods Struggle with Semantic Change Discovery in the Wild

  2025-01-01

  articleOpen access

  Methods for lexical semantic-change detection quantify changes in the meaning of words over time.Prior methods have excelled on established benchmarks consisting of pre-selected target words, chosen ahead of time due to the prohibitive cost of manually annotating all words.However, performance measured on small curated wordsets cannot reveal how well these methods perform at discovering semantic changes among the full corpus vocabulary, which is the actual end goal for many applications.In this paper, we implement a top-k setup to evaluate semantic-change discovery despite lacking complete annotations.(At the same time, we also extend the annotations in the commonly used LiverpoolFC and SemEval-EN benchmarks by 85% and 90%, respectively).We deploy our evaluation setup on a battery of semantic-change detection methods under multiple variations.We find that when presented with a natural distribution of instances, all the methods struggle at ranking known large changes higher than other words in the vocabulary.Furthermore, we manually verify that the majority of words with high detected-change scores in LiverpoolFC do not actually experience meaning changes.In fact, for most of the methods, less than a half of the highest-ranked changes were determined to have changed in meaning.Given the large performance discrepancies between existingbenchmark results and discovery "in the wild", we recommend that researchers direct more attention to semantic-change discovery and include it in their suite of evaluations.Our annotations and code for running evaluations are available at https://github.com/khonzoda/ semantic-change-discovery-emnlp2025.

  PublisherOA PDFDOI

• Hanging in the Balance: Pivotal Moments in Crisis Counseling Conversations

  2025-01-01

  articleOpen access

  During a conversation, there can come certain moments where its outcome hangs in the balance.In these pivotal moments, how one responds can put the conversation on substantially different trajectories leading to significantly different outcomes.Systems that can detect when such moments arise could assist conversationalists in domains with highly consequential outcomes, such as mental health crisis counseling.In this work, we introduce an unsupervised computational method for detecting such pivotal moments as they happen.The intuition is that a moment is pivotal if our expectation of the conversation's outcome varies widely depending on what might be said next.By applying our method to crisis counseling conversations, we first validate it by showing that it aligns with human perception-counselors take significantly longer to respond during moments detected by our method-and with the eventual conversational trajectory-which is more likely to change course at these times.We then use our framework to explore the relation between the counselor's response during pivotal moments and the eventual outcome of the session.

  PublisherOA PDFDOI

• Hanging in the Balance: Pivotal Moments in Crisis Counseling Conversations

  ArXiv.org · 2025-06-04

  preprintOpen access

  During a conversation, there can come certain moments where its outcome hangs in the balance. In these pivotal moments, how one responds can put the conversation on substantially different trajectories leading to significantly different outcomes. Systems that can detect when such moments arise could assist conversationalists in domains with highly consequential outcomes, such as mental health crisis counseling. In this work, we introduce an unsupervised computational method for detecting such pivotal moments as they happen, in an online fashion. Our approach relies on the intuition that a moment is pivotal if our expectation of the outcome varies widely depending on what might be said next. By applying our method to crisis counseling conversations, we first validate it by showing that it aligns with human perception -- counselors take significantly longer to respond during moments detected by our method -- and with the eventual conversational trajectory -- which is more likely to change course at these times. We then use our framework to explore the relation of the counselor's response during pivotal moments with the eventual outcome of the session.

  PublisherOA PDFDOI

• Taking a turn for the better: Conversation redirection throughout the course of mental-health therapy

  arXiv (Cornell University) · 2024-10-09

  preprintOpen access

  Mental-health therapy involves a complex conversation flow in which patients and therapists continuously negotiate what should be talked about next. For example, therapists might try to shift the conversation's direction to keep the therapeutic process on track and avoid stagnation, or patients might push the discussion towards issues they want to focus on. How do such patient and therapist redirections relate to the development and quality of their relationship? To answer this question, we introduce a probabilistic measure of the extent to which a certain utterance immediately redirects the flow of the conversation, accounting for both the intention and the actual realization of such a change. We apply this new measure to characterize the development of patient-therapist relationships over multiple sessions in a very large, widely-used online therapy platform. Our analysis reveals that (1) patient control of the conversation's direction generally increases relative to that of the therapist as their relationship progresses; and (2) patients who have less control in the first few sessions are significantly more likely to eventually express dissatisfaction with their therapist and terminate the relationship.

  PublisherOA PDFDOI

• Taking a turn for the better: Conversation redirection throughout the course of mental-health therapy

  2024-01-01

  articleOpen access

  Mental-health therapy involves a complex conversation flow in which patients and therapists continuously negotiate what should be talked about next.For example, therapists might try to shift the conversation's direction to keep the therapeutic process on track and avoid stagnation, or patients might push the discussion towards issues they want to focus on.How do such patient and therapist redirections relate to the development and quality of their relationship?To answer this question, we introduce a probabilistic measure of the extent to which a certain utterance immediately redirects the flow of the conversation, accounting for both the intention and the actual realization of such a change.We apply this new measure to characterize the development of patienttherapist relationships over multiple sessions in a very large, widely-used online therapy platform.Our analysis reveals that (1) patient control of the conversation's direction generally increases relative to that of the therapist as their relationship progresses; and (2) patients who have less control in the first few sessions are significantly more likely to eventually express dissatisfaction with their therapist and terminate the relationship.

  PublisherOA PDFDOI

• Supplementary Figure 3 from Bridging the Gap between Preclinical and Clinical Studies Using Pharmacokinetic–Pharmacodynamic Modeling: An Analysis of GDC-0973, a MEK Inhibitor

  2023-03-31

  preprintOpen access

  <p>PDF file, 66KB, Plot of observed versus predicted GDC-0973 plasma concentrations following fitting of PK model (Figure 1A) to mean GDC-0973 plasma concentration-time data from patients in cohort 1.</p>

  PublisherDOI

• Polarity Dataset v2.0

  Zenodo (CERN European Organization for Nuclear Research) · 2023-05-12

  datasetOpen accessSenior author

  Preliminary steps were taken to remove rating information from the text files, but only the rating information upon which the rating decision was based is guaranteed to have been removed. Thus, if the original review contains several instances of rating information, potentially given in different forms, those not recognized as valid ratings remain part of the review text. The reviews are split into sentences in the .csv file, which are labeled with the review they come from, as well as the sentiment of the overall review.

  PublisherDOI

• Supplementary Tables 1 - 3, Figure 1 from Cabozantinib (XL184), a Novel MET and VEGFR2 Inhibitor, Simultaneously Suppresses Metastasis, Angiogenesis, and Tumor Growth

  2023-04-03

  supplementary-materialsOpen access

  <p>PDF file - 104KB</p>

  PublisherOA PDFDOI

• Data from Cabozantinib (XL184), a Novel MET and VEGFR2 Inhibitor, Simultaneously Suppresses Metastasis, Angiogenesis, and Tumor Growth

  2023-04-03

  preprintOpen access

  <div>Abstract<p>The signaling pathway of the receptor tyrosine kinase MET and its ligand hepatocyte growth factor (HGF) is important for cell growth, survival, and motility and is functionally linked to the signaling pathway of VEGF, which is widely recognized as a key effector in angiogenesis and cancer progression. Dysregulation of the MET/VEGF axis is found in a number of human malignancies and has been associated with tumorigenesis. Cabozantinib (XL184) is a small-molecule kinase inhibitor with potent activity toward MET and VEGF receptor 2 (VEGFR2), as well as a number of other receptor tyrosine kinases that have also been implicated in tumor pathobiology, including RET, KIT, AXL, and FLT3. Treatment with cabozantinib inhibited MET and VEGFR2 phosphorylation <i>in vitro</i> and in tumor models <i>in vivo</i> and led to significant reductions in cell invasion <i>in vitro</i>. In mouse models, cabozantinib dramatically altered tumor pathology, resulting in decreased tumor and endothelial cell proliferation coupled with increased apoptosis and dose-dependent inhibition of tumor growth in breast, lung, and glioma tumor models. Importantly, treatment with cabozantinib did not increase lung tumor burden in an experimental model of metastasis, which has been observed with inhibitors of VEGF signaling that do not target MET. Collectively, these data suggest that cabozantinib is a promising agent for inhibiting tumor angiogenesis and metastasis in cancers with dysregulated MET and VEGFR signaling. <i>Mol Cancer Ther; 10(12); 2298–308. ©2011 AACR</i>.</p></div>

  PublisherDOI

• Supplementary Figure 1 from Bridging the Gap between Preclinical and Clinical Studies Using Pharmacokinetic–Pharmacodynamic Modeling: An Analysis of GDC-0973, a MEK Inhibitor

  2023-03-31

  preprintOpen access

  <p>PDF file, 57KB, Structure of GDC-0973.</p>

  PublisherDOI

Recent grants

• Graph-based Approaches to Text Processing

  NSF · $450k · 2003–2007

Frequent coauthors

• Cristian Danescu-Niculescu-Mizil

  36 shared

• Tianze Shi

  Peking University

  26 shared

• Jack Hessel

  24 shared

• Chenhao Tan

  22 shared

• Oren Kurland

  Technion – Israel Institute of Technology

  19 shared

• Bo Pang

  Shanghai Jiao Tong University

  16 shared

• Fernando Pereira

  16 shared

• Bo Pang

  Xiangtan University

  16 shared

Awards & honors

• Sloan Research Fellowship Alfred P. Sloan Foundation (2002)
• ACM Fellow (2018)
• ACL Fellow (2017)
• AAAI Fellow (2013)