Detecting Multiple Information Sources Based on the Quantum Walk

Abstract

The source detection problem is the reverse of information diffusion. It is identifying information sources of the diffusion process. However, most existing algorithms have limited performance due to various reasons, therefore the application is limited. We propose an approach, namely Quantum Walk based Source Detection (QWSD), to identify the source nodes based on the continuous-time quantum walk (CTQW). The continuous-time quantum walk can be utilized to simulate the diffusion process. Therefore, we utilize the reversed quantum walk to reverse the diffusion, thus revealing the probability for a node to be a source node. It is the first time that the continuous-time quantum walk is applied to the source detection problem. Extensive experiments demonstrate that our algorithm outperforms other source detection algorithms on both synthetic datasets and real-world datasets.