A large language model context aggregation method, device, equipment, storage medium and program product

Abstract

This disclosure provides a method, apparatus, device, storage medium, and program product for large language model context aggregation. The method includes: responding to a session splitting instruction for a target session; splitting the target session into at least two sub-sessions based on context information of the target session; responding to an explicit session citation instruction for the sub-sessions; determining the citation pattern corresponding to the explicit session citation instruction; performing parallel analysis on the sub-sessions based on the citation pattern to obtain parallel analysis results; generating an extraction strategy based on the parallel analysis results; generating context extraction prompts for the citation sessions; thereby obtaining target fragments of the citation sessions; aggregating the citation sessions based on the target fragments according to a preset aggregation algorithm to obtain an aggregated context package; and generating an aggregated answer based on the aggregated context package. This achieves collaborative splitting of complex tasks and automatic information integration, improving the efficiency of complex topic research, analysis, and multi-person collaboration, and solving the problems of difficult information synchronization and inefficient conclusion merging.

Images