Processing on-the-fly high volume of data streams is increasingly needed. To cope with the heterogeneity of this data, RDF model is more and more being adopted leading to plethora of RDF Stream Processing (RSP) systems and languages dealing with issues such as continuous querying, incremental reasoning and complex event processing (CEP). However, most of them has implemented centralized approaches and therefore suffer from some limitations as collaboration, sharing, expressiveness and scalability. Multi-agents systems have widely proven their worth and efficiency in particular their intrinsic decentralized property along with their cooperation and communication mechanism. In this paper we propose a new framework MAS4MEAN (Multi-Agent System for streaM rEAsoNing) based on a multi-agents model to embrace their benefits and tackle the challenges of increasing the scalability and ease of deployment in highly dynamic environments. A preliminary experimental evaluation with a real-world dataset show promising results when compared to an existing work.