Social peer‐to‐peer (P2P) networks are usually designed by reflecting a user's interest/behavior for structuring the underlying network. Human interest is affected by various factors such as age, locality, and so on which changes after some time. The behavior when reflected in a network, results in peers moving within the network in order to connect the peer with peers of the same behavior/interest. Especially in community‐based schemes when a peer leaves a community the data that a peer was sharing will not be accessible in the same community anymore. It has an effect on the performance of the network due to the inaccessibility of data and the unavailability of connections, which affect network robustness. We address this issue by considering entities in data in the form of digital memories of a user and structuring network according to entity‐based communities. The simulation results for the proposed entity‐based community are demonstrated, which shows the effect on network performance during varying network size and traffic.