Abstract:  The overreaction theory is introduced into a multi-agent simulation model, individual adoption Internet product interaction method is designed, and an Internet product diffusion simulation model in online social networks is constructed from the perspective of evolutionary games. Multi-agent simulation experiments were performed on the simulation software Anylogic 6.4.1. The results show that: (1) The higher the income of information service products, the product diffusion will fluctuate significantly at an earlier time, and the higher the level of product diffusion, but at the same time stability is poor and diffusion may shrink significantly; (2) when the  transmission cost is large, the rejector will be dominant most of the time; when the penalty cost is large, the receiver will be dominant most of the time; (3) the stronger the overreaction, the worse the diffusion stability, but the greater the chance of diffusion; (4) the smaller the information noise in the network, the higher the diffusion level. Through observing individual’s acceptance and rejection of Internet information service products and studying the effects of various utility parameters on the results of product diffusion in the network, certain reference for Internet product suppliers in advertising marketing is provided.

Key words:  , evolutionary game, overreaction, multi-agent simulation, Internet product diffusion