Call for Papers – Special Track on LLMs for Data Science
PAKDD 2025 will be held exclusively in person. All accepted presentations must be delivered on-site.
The 29th Pacific-Asia Conference on Knowledge Discovery and Data Mining (PAKDD) will take place in Sydney, Australia, on June 10-13, 2025.
The rapid advancements in Large Language Models (LLMs) have opened new avenues for innovation and research across various domains, particularly in the field of Data Science. As LLMs continue to evolve, their applications in data analysis, machine learning, natural language processing, and decision-making processes are becoming increasingly profound. The PAKDD 2025 Special Track on Large Language Models for Data Science aims to explore the transformative potential of LLMs for Data Science, bringing together researchers, practitioners, and industry experts to discuss the latest developments, challenges, and opportunities in this rapidly growing area. Novel, high-quality, and original research papers that provide innovative insights into all facets of large language models and their applications in data science, including but not limited to science and algorithms of LLMs, enlarged language models, retrieval-augmented text generation, vision-language pretraining, vision transformers, trustworthiness and societal implications of LLMs, and LLMs on diverse applications are solicited. Papers accepted to the LLM Special Track will be published in the PAKDD proceedings by Springer. At least one author of each accepted paper must register for the conference and present the work.
For up-to-date information on PAKDD 2025, please visit its homepage: https://pakdd2025.org.
Important Dates
- Paper Submission Deadline:
November 30, 2024December 14, 2024 - Paper Acceptance Notification: February 8, 2025
- Camera Ready Papers Due: March 1, 2025
*All deadlines are 23:59 Pacific Standard Time (PST)
Topics
- Novel Design and Architecture of LLMs: Innovations in model scaling, compression, efficient training and inference techniques, and optimization strategies for large language models.
- Pre-training and Fine-tuning Strategies: New methods for pre-training and fine-tuning LLMs on domain-specific or task-specific data, enhancing adaptability and performance.
- Prompt Engineering and Optimization: Techniques for designing and optimizing prompts to elicit desired responses from LLMs, including prompt tuning and dynamic prompt generation.
- Retrieval-Augmented Generation (RAG): Integrating retrieval mechanisms with LLMs to enhance text generation, improve context awareness, and provide more accurate and relevant outputs, especially in information-heavy tasks.
- Multimodal LLMs: Integration of LLMs with various modalities such as vision, speech, or sensor data, and advancements in cross-modal retrieval and generation using LLMs.
- Privacy in LLMs: Techniques for ensuring data privacy in the training and usage of LLMs, including but not limited to federated learning approaches.
- Fairness in LLMs: Approaches to ensuring fairness and equity in the outputs and predictions generated by LLMs, addressing bias and discrimination.
- Safety of LLMs: Methods to ensure the safe deployment and use of LLMs, including alignment strategies and risk mitigation techniques.
- Interpretability and Explainability: Developing methods to make LLMs more interpretable and understandable, along with generating explanations for the predictions and decisions made by these models.
- Zero-shot and Few-shot Learning: Enhancing LLM capabilities for performing tasks with minimal or no labeled data, pushing the boundaries of low-resource learning.
- Retrieval-Augmented Text Generation: Hybrid models that combine LLMs with retrieval systems to improve context understanding in dialogue systems and content creation.
- LLM-based Scientific Discovery: Applications of LLMs in fields such as drug discovery, biology, computational chemistry, molecular dynamics, materials design, and the solution of partial differential equations (PDEs).
- Domain-specific Applications: Exploring the use of LLMs in various domains, including healthcare, education, finance, environmental science, urban computing, and more.
Submission
Submission Details: https://pakdd2025.org/authors-kit
If you have any questions, please feel free to contact us at pakdd2025.llm@gmail.com.
PAKDD 2025 LLM Track Chairs
- Dongjin Song (University of Connecticut, USA)
- Ernestina Menasalvas (Universidad Politécnica de Madrid, Spain)