Responsibility · Reliability · Robustness in MLLM
Co-located with ACM SIGKDD 2026 — the premier international conference on Knowledge Discovery and Data Mining. Held at ICC Jeju, South Korea.
Multimodal Large Language Models (MLLMs), integrating text, images, audio, and video, are rapidly becoming central to data analysis, pattern summarization, and hypothesis generation. However, growing evidence suggests that biases, vulnerabilities, and opaque decision processes in these models can fundamentally reshape the outcomes of data mining.
RespMultimodal 2026 focuses on framing bias, fairness, interpretability, and robustness not as abstract ethical concerns but as core data mining challenges. We explicitly seek work that explores how MLLMs affect discovery validity, introduce spurious cross-modal correlations, and influence data-driven decision-making within the KDD community's scope.
All submissions must include a clear Responsible AI component — such as fairness, reliability, or transparency. Work focusing solely on unimodal LLMs is out of scope.
We invite submissions on topics including, but not limited to, the following areas. All submissions must pertain to Multimodal LLMs.
Two submission tracks — choose based on the maturity and nature of your work.
For mature research, novel methodologies, or comprehensive empirical studies. We encourage submissions that provide rigorous technical contributions to multimodal learning and data mining, including novel algorithms, large-scale evaluations, or in-depth case studies in high-stakes domains.
For early-stage ideas, provocative position statements, and vision papers to spark high-energy discussion. We especially welcome reports on "negative results" — sharing what didn't work and why is often as valuable as a success story.
\documentclass[sigconf]{acmart} format. Submissions that deviate significantly from the format or page limits may be rejected without review.Distinguished researchers from academia and industry. Details to be announced.
