In multi-agent argumentation scenarios, considerations such as the strength of attacks on arguments, the number of attacks received, and different criteria for selecting arguments in different argumentative scenarios need to be taken into account. To this end, this paper proposes various multi-agent argumentation semantics. Firstly, various types of multi-agent argumentation semantics including extension-based and ranking-based are proposed to characterize the ways in which arguments are selected by multiple agents in different scenarios. Secondly, methods such as constructing framework-updating models are used to study the relationships between these different semantics, providing theoretical references for the selection of multi-agent argumentation semantics.