Congratulations to Prof Ivor Tsang, Director, Distinguished Principal Scientist and Dr Pan Yuangang, Senior Scientist, on their paper acceptance at the IEEE International Conference on Data Engineering (ICDE)!

Their paper, titled “Boosting with Fewer Tokens: Multi-Query Optimisation for LLMs Using Node Text and Neighbour Cues”, highlights two multi-query optimisation strategies—token pruning and query boosting—based on single-query information theory analysis. Token pruning reduces token usage by removing redundant tokens from saturated queries, while query boosting improves performance by enriching unexecuted queries with pseudo-labels through strategic scheduling.

Held from 19 – 23 May at Hong Kong SAR, China, ICDE addresses research issues on designing, building, managing, and evaluating advanced data intensive systems and applications.