8 citations · 33 across the 23 of their papers we have counts for
25 papers
When Do Multi-Agent Systems Help? An Information Bottleneck Perspective
Wendi Yu, Lianhao Zhou, Xiangjue Dong +6
LLM powered multi-agent systems (MAS) have emerged as a promising paradigm for complex tasks. However, their advantages over single-agent systems (SAS) remain unclear, with perform…
DisasterBench: Benchmarking LLM Planning under Typed Tool Interface Constraints
Zhitong Chen, Kai Yin, Weifeng Zhang +7
Disasters cause severe societal impacts, demanding rapid coordination of heterogeneous AI tools, from satellite analysis to flood prediction and damage assessment, into coherent mu…
SocialPulse: An Open-Source Subreddit Sensemaking Toolkit
Stephanie Birkelbach, Maria Teleki, Peter Carragher +3
Understanding how online communities discuss and make sense of complex social issues is a central challenge in social media research, yet existing tools for large-scale discourse a…
DisastQA: A Comprehensive Benchmark for Evaluating Question Answering in Disaster Management
Zhitong Chen, Kai Yin, Xiangjue Dong +7
Accurate question answering (QA) in disaster management requires reasoning over uncertain and conflicting information, a setting poorly captured by existing benchmarks built on cle…
CHOIR: Collaborative Harmonization fOr Inference Robustness
Xiangjue Dong, Cong Wang, Maria Teleki +2
Persona-assigned Large Language Models (LLMs) can adopt diverse roles, enabling personalized and context-aware reasoning. However, even minor demographic perturbations in personas,…
Language Models as Semantic Augmenters for Sequential Recommenders
Mahsa Valizadeh, Xiangjue Dong, Rui Tuo +1
Large Language Models (LLMs) excel at capturing latent semantics and contextual relationships across diverse modalities. However, in modeling user behavior from sequential interact…
DMRetriever: A Family of Models for Improved Text Retrieval in Disaster Management
Kai Yin, Xiangjue Dong, Chengkai Liu +4
Effective and efficient access to relevant information is essential for disaster management. However, no retrieval model is specialized for disaster management, and existing genera…
Z-Scores: A Metric for Linguistically Assessing Disfluency Removal
Maria Teleki, Sai Janjur, Haoran Liu +10
Evaluating disfluency removal in speech requires more than aggregate token-level scores. Traditional word-based metrics such as precision, recall, and F1 (E-Scores) capture overall…
Conversational Speech Reveals Structural Robustness Failures in SpeechLLM Backbones
Maria Teleki, Sai Janjur, Haoran Liu +11
LLMs serve as the backbone in SpeechLLMs, yet their behavior on spontaneous conversational input remains poorly understood. Conversational speech contains pervasive disfluencies --…
DisastIR: A Comprehensive Information Retrieval Benchmark for Disaster Management
Kai Yin, Xiangjue Dong, Chengkai Liu +5
Effective disaster management requires timely access to accurate and contextually relevant information. Existing Information Retrieval (IR) benchmarks, however, focus primarily on…
Masculine Defaults via Gendered Discourse in Podcasts and Large Language Models
Maria Teleki, Xiangjue Dong, Haoran Liu +1
Masculine defaults are widely recognized as a significant type of gender bias, but they are often unseen as they are under-researched. Masculine defaults involve three key parts: (…
A Survey on LLM Inference-Time Self-Improvement
Xiangjue Dong, Maria Teleki, James Caverlee
Techniques that enhance inference through increased computation at test-time have recently gained attention. In this survey, we investigate the current state of LLM Inference-Time…
ReasoningRec: Bridging Personalized Recommendations and Human-Interpretable Explanations through LLM Reasoning
Millennium Bismay, Xiangjue Dong, James Caverlee
This paper presents ReasoningRec, a reasoning-based recommendation framework that leverages Large Language Models (LLMs) to bridge the gap between recommendations and human-interpr…
Disclosure and Mitigation of Gender Bias in LLMs
Xiangjue Dong, Yibo Wang, Philip S. Yu +1
Large Language Models (LLMs) can generate biased responses. Yet previous direct probing techniques contain either gender mentions or predefined gender stereotypes, which are challe…
The Neglected Tails in Vision-Language Models
Shubham Parashar, Zhiqiu Lin, Tian Liu +5
Vision-language models (VLMs) excel in zero-shot recognition but their performance varies greatly across different visual concepts. For example, although CLIP achieves impressive a…
: A Distribution-Aware Adversarial Attack against Language Models
Yibo Wang, Xiangjue Dong, James Caverlee +1
Language models can be manipulated by adversarial attacks, which introduce subtle perturbations to input data. While recent attack methods can achieve a relatively high attack succ…
Probing Explicit and Implicit Gender Bias through LLM Conditional Text Generation
Xiangjue Dong, Yibo Wang, Philip S. Yu +1
Large Language Models (LLMs) can generate biased and toxic responses. Yet most prior work on LLM gender bias evaluation requires predefined gender-related phrases or gender stereot…
CoPT: Mitigating Bias in Pre-trained Language Models through Counterfactual Contrastive Prompt Tuning
Xiangjue Dong, Ziwei Zhu, Zhuoer Wang +2
Pre-trained Language Models are widely used in many important real-world applications. However, recent studies show that these models can encode social biases from large pre-traini…
Everything Perturbed All at Once: Enabling Differentiable Graph Attacks
Haoran Liu, Bokun Wang, Jianling Wang +3
As powerful tools for representation learning on graphs, graph neural networks (GNNs) have played an important role in applications including social networks, recommendation system…
PromptAttack: Probing Dialogue State Trackers with Adversarial Prompts
Xiangjue Dong, Yun He, Ziwei Zhu +1
A key component of modern conversational systems is the Dialogue State Tracker (or DST), which models a user's goals and needs. Toward building more robust and reliable DSTs, we in…
Closed-book Question Generation via Contrastive Learning
Xiangjue Dong, Jiaying Lu, Jianling Wang +1
Question Generation (QG) is a fundamental NLP task for many downstream applications. Recent studies on open-book QG, where supportive answer-context pairs are provided to models, h…
Weakly Supervised Concept Map Generation through Task-Guided Graph Translation
Jiaying Lu, Xiangjue Dong, Carl Yang
Recent years have witnessed the rapid development of concept map generation techniques due to their advantages in providing well-structured summarization of knowledge from free tex…
Emora: An Inquisitive Social Chatbot Who Cares For You
Sarah E. Finch, James D. Finch, Ali Ahmadvand +9
Inspired by studies on the overwhelming presence of experience-sharing in human-human conversations, Emora, the social chatbot developed by Emory University, aims to bring such exp…
XD at SemEval-2020 Task 12: Ensemble Approach to Offensive Language Identification in Social Media Using Transformer Encoders
Xiangjue Dong, Jinho D. Choi
This paper presents six document classification models using the latest transformer encoders and a high-performing ensemble model for a task of offensive language identification in…