Data Mining Algorithms for Internet Data: from Transport to Application Layer

Nowadays we live in a data-driven world. Advances in data generation, collection and storage technology have enabled organizations to gather data sets of massive size. Data mining is a discipline that blends traditional data analysis methods with sophisticated algorithms to handle the challenges posed by these new types of data sets. The Internet is a complex and dynamic system with new protocols and applications that arise at a constant pace. All these characteristics designate the Internet a valuable and challenging data source and application domain for a research activity, both looking at Transport layer, analyzing network tra c flows, and going up to Application layer, focusing on the ever-growing next generation web services: blogs, micro-blogs, on-line social networks, photo sharing services and many other applications (e.g., Twitter, Facebook, Flickr, etc.). In this thesis work we focus on the study, design and development of novel algorithms and frameworks to support large scale data mining activities over huge and heterogeneous data volumes, with a particular focus on Internet data as data source and targeting network tra c classification, on-line social network analysis, recommendation systems and cloud services and Big data.