This research project is in the area of open interaction systems, that is, of software systems in which different programs, regarded as heterogeneous autonomous agents, interact with other agents according to a shared set of rules, in order to achieve their individual goals. A crucial characteristic of this innovative type of systems is that they are open: this means that they are available on the Internet and agents may freely enter and leave them. Major application areas of these systems are e-business and socio-technical systems in the Web 2.0 and 3.0 era.This project is a continuation of the SNF project 200021-121681 with the title "Open Interaction Systems: An approach based on artificial institutions and reasoning agents with an application to collaborative environments".In our past works we have defended the view that open interaction systems can be specified as artificial institutions, that is, as systems in which agents play certain roles, are empowered to perform certain actions, and are subject to a set of norms regulating the interactions. In particular we have proposed the OCeAN metamodel, that we now use for the specification of open interaction frameworks, regarded as a set of artificial institutions. At the meta-model level we mainly investigated the problem of enabling an agent to understand and use at runtime a logical representation of an artificial institution and enabling the interaction system¿s to monitor the agents¿ interaction, in order to apply a form of social control to the actions of those autonomous and self-interested parties. This involves aspects concerning knowledge representation and reasoning.At the architecture level our focus is on the specification of the complete abstract software architecture (with no reference to a specific application domain) for institution-based open interaction systems that is necessary to enable independent parties to find each other, to interact in an open environment, and to have reliable expectations on the future evolution of the state of the system without necessarily regimenting all agent actions. After an extensive study of possible alternatives, we came to appreciate the importance for our goals of integrating Agent Oriented Computing (AOC) with Service Oriented Computing (SOC). We now plan to continue our research by proposing an architecture and an infrastructure for institution-based open interaction systems, which combine elements from both AOC and SOC, and by evaluating such a proposal through the realization of a prototype of an open interaction system in the e-business domain, that is, an e-marketplace. Given that these activities are strictly interconnected we plan to carry out them following an incremental development model.This project touches upon three areas of Computer Science: multiagent systems, semantic web technologies, and service oriented computing. It addresses issues of both basic and applied research, and thus may have a significant impact on practical applications.