MuG: A Multilevel Graph Representation for Big Data Interpretation

Our society is oriented towards data production. The increasingly massive spread of mobile devices and the Internet of Things is transforming our society into a data factory. Data, however, does not immediately lead to knowledge and, in fact we can become overwhelmed with a mass of information that is difficult to understand: often the desire to predict the future from data analysis turns into the nightmare of data overload. There are numerous approaches, automatic and manual, present in the literature that try to interpret data by extracting information. Among the various methodologies proposed, none seems to have resolved the problem in a definitive and universal way, perhaps because every data analysis problem needs to be faced from a different point of view. This paper introduces an approach for the interpretation of data from sensors located within a city. Three graphs (Ontologies, Context Dimension Tree and Bayesian Networks) were chosen for the representation of the scenarios both from the point of view of the sensors involved and of the services and events connected to the data. Through the Ontologies and the Context Dimension Tree it is possible to analyze the scenario from a syntactic and semantic point of view constructing Bayes networks that enable the estimation of the probability that some events happen. A first empirical analysis conducted on some districts of London seems to confirm the effectiveness of the proposed method.