nav icon

Building multimodal AI that sees, reasons, and communicates with data

Based at York University, our research Lab brings together NLP, visualization, and multimodal LLMs to make data interaction more responsible, interpretable, and human-centered.

RESEARCH THEMES

Multimodal LLMs for Visualization Understanding and Generation

Benchmarks and models for chart comprehension, narrative generation, and visual data communication—shaping how multimodal LLMs interpret, explain, and generate visualizations from and for real analytic workflows.

Multimodal LLMs for chart and visualization reasoning

Representative papers

Agentic AI for Data Science Workflows

Building multimodal agents that plan actions, interact with analytic tools, refine outputs, and collaborate with humans across realistic, end-to-end data science workflows.

Agentic AI operating on analytic dashboards and tools

Representative papers

Human-Centered & Responsible AI for Data Visualization

Designing accessible, trustworthy, and goal-aligned AI systems that collaborate with people, mitigate bias and deception, and support diverse analytical needs in data visualization.

<img src="/images/inverted.png alt="Human-centered and responsible AI for data visualization">

Representative papers

Visual & Interactive Document Analytics

Developing interfaces and visual analytics techniques that help users read, explore, and interpret text-rich documents.

<img src="/images/3.jpg alt="Visual analytics for large text collections">

Representative papers

Language Understanding, Summarization, and Evaluation

Advancing question answering, summarization, domain adaptation, and trustworthy evaluation methods that strengthen the foundations of language-driven analytics.

Language understanding, summarization, and evaluation for analytics

Representative papers