Abstract: The spread of HIV is the result of coevolution of individual behaviors,intervention measures and social networks．This paper presented a method based on bipartite dynamic weighting scale-free agent network for investigating HIV heterosexual transmission and the effectiveness of HIV prevention interventions．According to different high-risk behavior,the female is divided into the general population and female sex workers (FSW),and the male is divided into the general population and clients of female sex worker (CSW) in the bipartite network．A bipartite scale-free network configuration generation algorithm was constructed for describing the relationships of heterosexual contacts．A new weighting algorithm was developed to construct weighting network by adding sex act explicitly on heterosexual contact network．The dynamics of heterosexual network is supported through heterosexual network stable,casual and temporarypartnership dissolution and formation．In the model,agent represents males and females who interact each other within risk networks engaging in sexual behavior over time．Agent describes individual high-risk behavior,course of HIV disease and social structure．Agent also defines the behavior rules for adaption to the HIV prevention interventions．Exploratory simulation results are consistent with empirical studies,which demonstrate the effectiveness of a combination of interventions,including voluntary counseling and HIV test,condom promotion for safe sexual behavior and initiating antiretroviral treatment.

Key words: Agent,Bipartite scale-free network,Intervention,HIV