Research( Yilin Xia ) :-
Declarative(Query), Declarative(Graph), Declarative(AI).
I am a final-year Ph.D. candidate co-advised by Dr. Bertram Ludäscher and Dr. Matthew Turk in the School of Information Sciences at the University of Illinois at Urbana-Champaign. I also collaborate closely with Dr. Shawn Bowers (Gonzaga University), Dr. Evgeny Skvortsov (Google), and the ArviZ Team.
My research addresses fundamental challenges in querying, reasoning over, and explaining graph-structured knowledge. I utilize Neuro-Symbolic Approaches that combine symbolic logic with large language models, enabling scalable analytics, hybrid retrieval, and explainable multi-agent interaction. My current research interests include:
1
Unstructured Knowledge to Graph
Multi-agent graph building from unstructured knowledge base
DaSH@VLDB'25
2
Declarative Query Language
Query language for graph transformation
TGD@EDBT'25
EDBT'24
Datalog2.0'24
3
Graph Reasoning & Explanation
Argumentation-based reasoning and explanation with provenance
ICAIL'25
PW@SIGMOD'25
COMMA'24
Neuro-Symbolic(NeSy)
Combining symbolic logic with LLMs throughout the pipeline, exploring three
of six types
Type 2: Logica
Type 3: ArguAgents
Type 4: DeepGit
Featured Publications
Latest News
January, 2026 We have received the Gemini Academic Program Award ($10K GCP credit) from Google
December, 2025 We have received the Tinker Research Grant $5,000 from Thinking Machines Lab
July, 2025 We have received the GitHub Secure Open Source Fund
June, 2025 DeepGit got accepted by DaSH@VLDB'25!
Mar, 2025 AF-XRAY got accepted by ICAIL'25 demo track! Let's enhance legal reasoning with AFs!
Mar, 2025 DeepGit received Google Academic Research Grants, marking our first industry funding support for the project
Jan, 2025 LogicLM has been accepted by EDBT 2025 Demo Track
Sept, 2024 We got two papers on Logica accepted by ICLP workshops Datalog2.0 & PEG
July, 2024 Attended SciPy'24 and presented two posters on JupyterLab NodeEditor (JLNE) and Arviz Dashboard
June, 2024 Our demo paper "Layered Visualization of Argumentation Frameworks" accepted for COMMA 2024
Feb, 2024 Our demo paper "Logica: Declarative Data Science for Mere Mortals" accepted for EDBT 2024
Jan, 2024 Our paper "MAIDR: Making Statistical Visualizations Accessible with Multimodal Data Representation" has been accepted for CHI 2024
Nov, 2023 Received $10,000 funding from NumFOCUS to work on project "Dashboard-assisted exploratory analysis of Bayesian models"
Nov, 2023 Our work "Games, Queries, and Argumentation Frameworks: Time for a Family Reunion!" has been accepted to AI^3 at AIxIA
Oct, 2023 Awarded $1,000 Conference Travel Grant from United States Research Software Engineer Association(US-RSE)
July, 2023 One workshop paper was accepted by XLoKR at KR2023
July 10-14, 2023 Thrilled to be part of Princeton University INTERSECT Program, supported by an NSF travel grant
May, 2023 Got Outstanding Review for IEEE VIS 2023
March, 2023 Our first (VISxProv)*Domain workshop proposal accepted by IEEE VIS 2023. Check out the workshop website
Feb 22nd, 2023 I led a seminar reading discussion on Knowledge Graphs in Action: a Tour of Extensions and Real-World Applications of the Vadalog System at UIUC iSchool
Jan, 2023 Our paper CrowdIDEA: Blending Crowd Intelligence and Data Analytics to Empower Causal Reasoning" has been accepted to CHI 2023
Funding & Grants
Academic Program Award
Google Gemini
$10K GCP credit
2026
Secure Open Source Fund
GitHub
$10K grant and $100K Azure credit
2025
Tinker Research Grant
Thinking Machines Lab
$5K credit
2025
Travel Grant
Logic Programming and Non-monotonic Reasoning
2024
Small Development Grant
NumFOCUS
$10K Grant
2023
CS Research Mentorship Program Scholar
Google
2022
Featured Projects
CrowdIDEA
CrowdIDEA integrates crowd beliefs, data analytics, and causal diagrams into a tool for stimulating causal reasoning. Its design encourages exploration of causal relationships using collective insights
ArviZ Dashboard
Designed for Bayesian statisticians, this project transforms static visualizations into interactive dashboards with tools like Panel and Bokeh. It enables dynamic model comparison within a unified interface.
MAIDR
MAIDR offers non-visual access to statistical visualizations for users with visual impairments. It supports interaction through braille, text, sonification, and review (BTS + R).