AI-Ready Data Preparation is the fourth lecture of ScienceOne.

Details of the Lecture

In this session, Prof. LIU Qiang will introduce the concept of AI-Ready Data for AI for Science (AI4S), highlighting key attributes like representativeness, completeness, and low redundancy. It presents a data preparation pipeline involving cleaning, alignment, and annotation. The talk also covers the ScienceOne methodology, which integrates modality-specific models with LLMs and chain-of-thought reasoning to improve scientific tasks such as material property prediction and molecular verification. Case studies on spectral data understanding show over 99% accuracy in molecular structure reconstruction.

Speaker Profile

LIU Qiang is a professor at the Institute of Automation, Chinese Academy of Sciences. His research focuses on data mining and AI for Science (AI4S). He has received prestigious awards, including the Microsoft Research Fellowship and the Dean's Award of the Chinese Academy of Sciences. He has published over 90 papers in CCF-A/B class and IEEE/ACM transactions, with more than 12,000 citations and a maximum of 1,300 citations for a single first-authored paper. As a core member, he built domain-specific modality models for the ScienceOne Foundation Model. His work is integrated into NVIDIA's Graph Library and Baidu's PaddlePaddle. His team has established collaborations with or achieved technology transfer to multiple companies, including Tencent, Alibaba, and Ant Group

Date & Time

15:00–16:00 (Beijing Time, GMT+8)

May 19, 2026

Organizers

● Alliance of National and International Science Organizations for the Belt and Road Regions (ANSO)

● Institute of Automation, Chinese Academy of Sciences (CASIA)

Online Access (Microsoft Teams)

Meeting Link: https://teams.live.com/meet/9354152305035?p=budhuGBXVAVxn01Pvc

Meeting ID: 935 415 230 503 5

Passcode: Zw2RM9