Hybrid Distributed Infrastructures for Recommendation Algorithms

Hybrid Distributed Infrastructures for Recommendation Algorithms

PICT Project 2018

Funded by ANPCyT

Distributed computing systems such as Clouds, Clusters or Grids group several computers to provide vast computational capabilities to user applications creating the illusion of virtual super computers. In recent years, the application of distributed computing has grown tremendously, enabling significant advances in various areas of knowledge. This growth was motivated by the need for computational resources to run applications of increasing complexity. In particular, recommendation algorithms in social networks and collaborative tagging systems pose several challenge to distributed computing due to their high processing and storage requirements. Recently, the possibility of integrating devices such as smartphones and tablets as resources of distributed infrastructures gave rise to mobile clouds, hybrid grids and ad-hoc clusters of smartphones, which are known as hybrid environments. In these scenarios, cloud computing and service-oriented architectures (SOA) constitute the fundamental technological pillars. The first pillar represents a high-scalability computing paradigm that exposes a wide range of computing resources in a variety of ways (IaaS, PaaS, SaaS - Infrastructure/Platform/Software as a Service). The second pillar, the SOAs, allows the design of distributed applications through the assembly of existing software, or services, that can be invoked remotely. Technologically, this software is usually implemented through Web Services, whose main objective is to achieve reusable and interoperable services.

One of the main challenges that arise from the emergence of these distributed systems is how to deal with the heterogeneity in terms of hardware and software of computing resources. In this project it is proposed to investigate new methods and techniques to achieve such systems, focusing on the problem of how to take advantage of the joint computing capabilities of fixed resources (PCs and servers) and mobile ones (smartphones and tablets). In particular, efforts will be focused on the requirements posed by the algorithms used in Social Recommendation Systems (SRSs), more specifically the recommendation in social networks and collaborative tagging systems.

PI: Alejandro Zunino

Research Team:

  • Daniela Godoy
  • Matías Hirsch
  • Cristian Mateos
  • Juan Manuel Rodriguez
  • Antonela Tommasel
Daniela Godoy

My research interests include recommender systems, social networks, text mining and social networks.