Transfer of resource allocation between overlapping and embedded communities in multiagent social networks.

In multiagent social networks, resource allocation aims to replicate and distribute resources to optimize the efficiency of agents’ resource access. Traditionally, each community in a network is considered separately, producing certain repetitive loads and computationally expensive processing. Observing that relations between communities are significantly simpler than their internal structures, we conclude that relation-based resource transfer could be economical compared to allocating from scratch. Therefore, in this study, we propose a transfer of resource allocation (TRA) method based on overlapping and embedded relations among communities. Our proposed method first identifies the most influential community in a network, with the greatest impact on other communities, and selects it as an original community, allocating its associated resources. The resource distribution is then transferred gradually to other communities based on relations between the completed communities and pending communities. When no transfer is available, another influential community is selected, and the previous steps are repeated until all communities are considered. We present experimental results demonstrating that our proposed TRA method allocated computational resources to agents with a lower cost than traditional methods, with acceptable performance.