Intelligence clue evolution inference method for fragmented intelligence

Abstract

This application relates to a method for inferring the evolution of intelligence leads in fragmented intelligence. The method includes: first, acquiring fragmented information from multiple sources; constructing a fragmented network structure using a graph neural network; analyzing the fracture characteristics and locating missing regions in the fragmented network; filling in the missing links using a graph neural network to obtain repaired lead chain structure data; then, calculating the fragment correlation degree based on this data; and finally, weighted integration using an attention mechanism to form a lead correlation framework; and finally, constructing a global lead relationship graph using a clustering algorithm, weighted evaluation of highly correlated leads to obtain a lead reliability assessment result. This method completes fragment correlation modeling and lead repair using a graph neural network, solving the problems of fragmented and missing correlations in multi-source intelligence leads; it optimizes key lead selection by combining an attention mechanism; and it achieves global relationship construction and reliability determination through clustering and weighted evaluation, effectively improving the accuracy, systematicity, and processing efficiency of inferring the evolution of fragmented intelligence leads.

Images