Alimohammad Beigi
About Me
Hi, I'm Alimohammad (Ali) Beigi! I'm a Ph.D. candidate in Computer Science at Arizona State University, where I am fortunate to be mentored by Dr. Huan Liu and work in the DMML Lab. I also serve as a teaching assistant and instructor at ASU.
I hold a bachelor's degree in Mathematics and Applications from the University of Tehran and a master's degree in Computer Science from Shahid Beheshti University.
My research sits at the intersection of causality, machine learning, NLP, and trustworthy AI. I work on projects including:
- Causal Feature Selection: Redefining CFS to go beyond Markov Blanket discovery and support predictive modeling, fairness, treatment-effect estimation, and domain generalization.
- LLMs for Fact-Checking: Building multimodal fact-checking systems that generate and answer relevant fact-checking questions using LLMs and VLMs.
- Digital Twins for Decision-Making: Combining causal discovery, reinforcement learning, and LLM-augmented reasoning for battlefield and disaster-response scenarios.
- Disaster Assessment: Leveraging BERT-based embeddings, multimodal news/Reddit data, and county-level property damage labels (SHELDUS) to model and predict disaster impacts.
Beyond academia, I explore applied AI and entrepreneurship. Currently, I am working on a Remote Therapeutic Monitoring (RTM) mobile app for chronic pain management, integrating CBT-style prompts, daily symptom tracking, and LLM-powered chat flows.
Outside research, I am a BMW enthusiast, enjoy travel, creative writing, and building communities through collaboration and mentorship.
If you have any questions about my work, research, academia, or programming, feel free to reach out!
Education
PhD in
Computer Science
Arizona State University
- Graduate Research Associate, Data Mining & Machine Learning Lab
(DMML), advised by Dr. Huan Liu
- Causal discovery with LLM-augmented heuristics to recover cause–effect relations and guide inference.
- Designed an interventional disaster assessment system based on causal graphs for simulation, attribution, and counterfactual strategies.
- Developed LRQ-FACT: a multimodal fact-checking framework using LLMs and VLMs to generate fact-checking questions, improving accuracy by up to 10% on MMFakeBench.
- Advanced model attribution for LLM-generated disinformation with supervised contrastive learning and domain generalization (7% accuracy gain).
- Instructor & TA: CSE 472 (Social Media Mining), CSE 511 (Data Processing at Scale), FSE 100 (Intro to Engineering).
- Reviewer: ICDM, WSDM, SDM, AAAI, SBP-BRiMS.
MSc in
Computer Science
Shahid Beheshti University
- Graduate Research Associate, advised by Dr. Ali Katanforoush
- Thesis: Development of a Sequential Recommender System by BERT
- Redesigned Transformer Self-Attention to improve sequential recommender performance by 13%.
- Integrated Self-supervised Q-Networks with Soft Actor-Critic for reinforcement-learning recommender models.
- Focus areas: AI, Machine Learning, Data Mining, Neural Networks.
BSc in
Mathematics and Applications
University of Tehran
- Specialized in Optimization (Linear and Non-Linear), Algebra, and Numerical Linear Algebra.
- Built strong mathematical foundations that shaped later work in machine learning and causality.
Projects
LRQ-FACT: Multimodal Fact-Checking with LLM/VLM-Generated Questions
Framework where LLMs/VLMs generate relevant fact-checking questions and answer
them via RAG over trusted sources, improving F1 by up to 10% on MMFakeBench; includes
ablations, random-question baselines, and runtime analysis (BigData’25).
Interventional Disaster Assessment via Causal Attribution
Real-time disaster assessment using causal graphs for scenario simulation, attribution
scoring, and practical counterfactual strategies; leverages multimodal signals (news,
Reddit) aligned with SHELDUS county-level labels (CIKM’25).
LLM-Heuristic Local Causal Discovery
Augments local structure learning with LLM heuristics to disambiguate edges when CI
tests are brittle.
Model Attribution in LLM-Generated Disinformation
Domain-generalization + supervised contrastive learning pipeline to separate human vs.
LLM-generated text and infer the likely source model—achieving a 7% accuracy gain
(DSAA’24).
Digital Twin for Decision-Making
A pipeline that fuses causal discovery (PC, GES, NOTEARS), RL objectives, and
LLM-augmented graph extraction to support scenario exploration and policy testing for
battlefield/disaster-response settings.
CBT-based Remote Therapeutic Monitoring (RTM) App for Chronic Pain
Co-building an RTM mobile app with clinicians and engineers: daily symptom tracking,
CBT-style prompts, and LLM-powered chat flows. Focus on patient–provider loops, CMS RTM
billing flows, concurrency-safe chat, and user-specific memory.
Publications
Large language models for data annotation and synthesis: A survey
Z Tan, D Li, S Wang, , Zhen Tan,
Dawei Li, Song Wang,
Bohan Jiang, Amrita Bhattacharjee, Mansooreh Karami, Jundong Li, Lu Cheng, Huan Liu
Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, 2024
From generation to judgment: Opportunities and challenges of LLM-as-a-judge
Dawei Li, Bohan Jiang, Liangjie Huang, , Chengshuai Zhao,
Zhen Tan, Amrita Bhattacharjee, Yuxuan Jiang, Canyu Chen, Tianhao Wu, Kai Shu, Lu Cheng,
Huan Liu
Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing, 2025
Model Attribution in LLM-Generated Disinformation: A Domain Generalization Approach with
Supervised Contrastive Learning
, Z Tan, N Mudiam, C Chen, K Shu, H
Liu
2024 IEEE 11th International Conference on Data Science and Advanced Analytics (DSAA)
Can LLMs Improve Multimodal Fact-Checking by Asking Relevant Questions?
, Bohan Jiang, Dawei Li, Zhen Tan,
Pouya Shaeri,
Tharindu Kumarage, Amrita Bhattacharjee, Huan Liu
arXiv preprint arXiv:2410.04616, 2024
FediverseSharing: A Novel Dataset on Cross-Platform Interaction Dynamics between Threads
and Mastodon Users
U Jeong, , A Tahir, SX Tang, HR
Bernard, H Liu
🏆 Best Paper Award: Advances in Social Network Analysis and Mining (ASONAM), 2025
Sentiment and Social Signals in the Climate Crisis: A Survey on Analyzing Social Media
Responses to Extreme Weather Events
P Shaeri, Y Mohammadpour, , A
Middel, H Liu
arXiv preprint arXiv:2504.18837, 2025
Tri-Accel: Curvature-Aware Precision-Adaptive and Memory-Elastic Optimization for
Efficient GPU Usage
M Sheibanian, P Shaeri, , RT Woo, A
Keluskar
arXiv preprint arXiv:2508.16905, 2025