Skip to main content

Social Analytics

The research activities are related to technologies to access, retrieve, extract, process information for very large databases with structured and unstructured data, with particular reference to the analysis of social networks. The research activities are:

  1. Techniques of statistical analysis for information flows by probabilistic computation or by sampling, for example, using the MinHashing -based counter or the Flajolet-Martin algorithm;

  2. Analysis of the degree of connectivity of a social network, through the estimation of measures for large graphs, such as the diameter or the average distance between the nodes of the graph, using counting techniques to estimate the size of distinct pairs of reachable nodes at arbitrary distance;

  3. Sentiment Analysis applied to social networks with cumulative counting techniques, that is techniques that do not classify and count messages but that quantifies directly the categories sizes, by sampling on a set of sentiment words (features);

  4. Definition of models for highly scalable platforms of Data Analytics, with particular attention to the use of:

    1. predictive models, for example based on Holt-Winters method, Naive Bayes, SVM, linear regression, logistic regression classifiers

    2. models for the discovery and visualization of relationships between entities;

  5. Analysis of the platforms enabling the elaboration of information flows in real time;

  6. Analysis of the Spark ecosystem for statistical data processing on distributed R (SparkR);

  7. Distributed deployment of Big Data technologies in PaaS modality on a cluster of machines.