PRESTO - a Polyhedric Recommender Engine based on Situation and Time-aware cOntexts

Purpose - The main objective of this system is to suggest the right and suitable resources, people, groups and activities within OPENNESS platform to the end users to personalize and recommend the right information[Resnick97], items and services to the end users, based on their implicit and explicit preferences. OPENNESS is the main outcome of the italian research project VINCENTE (A Virtual collective INtelligenCe ENvironment to develop sustainable Technology Entrepreneurship ecosystems) and is an Enteprise Social Software for knowledge workers whose aim is to connect users that share their knowledge creating and finding solutions and new services for young innovative entrepreneurs and already existing SMEs. As described in [Elia2014] the four pillars of the OPENNESS platform are people, activities, resources and actions. The actions represent the means to link the other three pillars among them and to connect the communities, the knowledge base, the achieved objectives and the completed/uncompleted activities. Design/methodology/approach - In this paper we present a new implicit personalization and recommendation approach, based on a hybrid system, built on top of an arithmetical model, a semantic network, a distributional semantics context and time aware recommender [Musto2012, Basile2015] and a goal-oriented model in order to infer relationships with other user groups and behavioural patterns. Originality/value - This methodology puts in evidence how the knowledge about interests, preferences and goals of a user is the basis of the concepts of recommendation and personalization. In a collective intelligence environment we can extend the user profile information for profiling system with additional data that comes with the collaborative environment, like the group the user belongs, what or who the users follow and like, what a user is talking about more. The audit log of the platform is the sensing feature of the recommender, is the sensor that listens to the resource social life and analysing it, the system in able to provide the most suitable resources, activities and groups to join. Practical implications - The hybrid system proposed in this paper, goes beyond the individual systems that deal only with recommendation or practical reasoning by exploiting the potential that their integration ensures. The system is able to work even on small data volumes or without any explicit input from the user (thanks to the configurable weights available for user actions) and on inferences made by exploiting the cascade of inferences triggered by the implicit profiling system. We need to consider that PRACTIONIST (a framework for developing agent systems according to the Belief-Desire-Intention model as described in Morreale2006), the semantic network and the recommendation algorithms CASPERI and STARS taken individually or even by their sub-components, can still be used in a modular way for working on their specific field of application.