Social Computing in JaCaMo

Social Computing (SC) requires agents to reason seamlessly both on their social relationships and on their goals, beliefs. We claim the need to explicitly represent the social state and social relationships as resources, available to agents. We built a framework, based on JaCaMo, where this vision is realized and SC is implemented through social commitments and commitment protocols. 1 PROPOSAL AND MOTIVATION Many systems, developed to support human users, require a transition from an individualistic to a societal perspective. For instance, Socio-Technical Systems (STS) are large-scale, multi-party, crossorganizational systems, which help stakeholders to interact and to use shared resources [11]. Such systems perform a social computation which is the sum of the independent contributions of autonomous, and heterogeneous, parties [10]. Traditional approaches to software engineering do not fit the needs of such systems, because they do not help capturing the social aspects of the computation, like the social relationships between the parties. The way suggested by STS is to foresee a specific layer that contains the regulations that norm the system behavior. This direction is followed by normative MAS, e.g. [7], which enrich MASs by representing the norms that rule the system. However, such approaches lack proper abstractions for capturing the social state, i.e. the set of relationships and dependencies that are created and exist along the course of events, which are the foundations on which the social behavior of the parties is established. Social relationships connect the interacting parties, they have a normative value (in that they allow agents to have expectations on one another), and they can be verified based just on the observable behavior of the agents. From a Software Engineering perspective, the advantage of explicitly representing the social state is to allow the realization of systems with a high degree of decoupling and of modularity of their components, avoiding to “hard code” the logic of interaction inside the code of the agents, whose executions are kept aligned by the social state itself. Most of Multi-Agent frameworks and platforms do not explicitly account for the social state. We propose an agent framework, 2COMM4JASON, that, instead, does so by explicitly representing the social state through the social relationships and the rules that cause it to evolve along the interaction. Agents and social relationships are first-class entities that interact in a bi-directional manner. Social relationships are created by the execution of interaction protocols and provide expectations on the agents’ behavior. On the other hand, existing social relationships affect the decisions and the behaviors of the agents they involve. Our proposal exploits the Agents&Artifacts (A&A) meta-model [8] and reifies the social state as a set of resources, in a way that al1 Università di Torino, Dipartimento di Informatica, Italy, email: {matteo.baldoni,cristina.baroglio,federico.capuzzimati}@unito.it lows agents to seamlessly reason on them and on their beliefs, goals, etc. Thus parties can dynamically recognize, accept, refuse rules and relationships, as advised in [4]. Moreover, the social state, as a resource, can be used to monitor the interaction. Social relationships are modeled as commitments and the rules that cause the social state to evolve are modeled as commitment-based interaction protocols. The framework builds upon the JaCaMo platform [2], and the Jason language is extended so to allow reasoning on commitments.