Item

Abstract

The rapid proliferation of online social networks has sharpened the demand for audience-specific information dissemination, i.e., maximizing influence within a predefined target group while suppressing spill-over to non-targets. We formalize this requirement as the random-distributed target influence maximization problem and present High Dual-Balanced Influence (HDBI), a powerful algorithm for seed-node selection for such problem. HDBI combines two novel estimators: combined approximate theoretical infection probability, which analytically gauges a candidate’s expected reach inside the target community by coupling local topology with diffusion dynamics; and weighted collective influence, which anticipates collateral exposure by aggregating the propagation capacities of neighboring non-target nodes. These estimators are then used for HDBI which iteratively selects seeds with maximal on-target impact and minimal off-target leakage. Extensive experiments on eight real-world networks of diverse scale and structure, complemented by large synthetic benchmarks, show that HDBI achieves substantially higher target coverage and lower non-target activation than state-of-the-art baselines, while maintaining superior computational efficiency. The proposed framework therefore offers a principled and scalable foundation for precision marketing, public-service messaging, and other applications requiring fine-grained control over influence propagation.