Resource allocation is an important way to solve the problem of the lack of spectrum resources, and humanity plays a key role in the communication quality of Device-to-Device (D2D) on the licensed radio spectrum. On these bases, in this paper, a double auction scheme of resource allocation with social ties and sentiment classification for D2D communications, which combines the social and physical domains, is proposed. Specifically, the double auction scheme is successfully implemented by the general assignment game algorithm (GA-alg), in which D2D pairs and spectrum resources act as sellers and bidders in each auction round, respectively. Additionally, social ties between users are determined by interaction data and the Kullback–Leibler divergence language model based on sentiment classification. Finally, simulation results not only illustrate that the proposed double auction scheme can achieve an optimal solution with lower complexity, but also demonstrate the effect of social ties on D2D communications.