MANET-oriented hybrid grids for location-based recommender systems

MANET-oriented hybrid grids for location-based recommender systems

PIP Project 2017-2019

Funded by CONICET

Mobile devices such as smartphones and tablets have experienced an overwhelming and sustained growth in recent years, and by 2021 the number of connected devices is expected to exceed the population world. In addition to this quantitative projection, such devices have gradually experienced remarkable qualitative improvements, given by processors with more cores, higher resolution and durability screens, greater memory and storage capacities, greater variety and accuracy of its sensors, among others. This reality has attracted many researchers from the area of ​​Parallel and Distributed Computing, leading to projects that grant to mobile devices a leading role in the context of new intensive computing environments such as Mobile Clouds, Mobile-edge Grids, and hybrid Grids. In particular, hybrid Grids propose using the phones for downloading computations and data to devices in the neighborhood as well as fixed hardware resources, a paradigm that is suitable for scenarios where a high-quality fixed device-infrastructure connectivity can be assumed and devices exhibit low mobility. However, in many application scenarios, these assumptions are not valid, and therefore it is necessary to maintain a mobile computing environment with a (semi) disconnected fixed infrastructure and also with greater mobility. This in turn can benefit the services based on location / context that make intensive use of computing resources, data or sensors. A family of techniques that naturally fit into this conceptualization are the recommendation algorithms which generate suggestions to users based on their interests, behavior and current context, applicable to a large number of domains such as vehicle traffic management, participatory sensing, tourism, among others. In these systems the main factor is the individual/colective location/context to make recommendations as well as taking advantage of nearby computing resources contributes to deliver recommendations in a timely manner. Specifically, this project plans to adapt the middleware to hybrid grids and associated mechanisms already studied by the group through MANET concepts, with focus on supporting algorithms for recommendation systems in context-aware applications.

PI: Cristian Mateos

Research Team:

  • Daniela Godoy
  • Matias Hirsch
  • Alfredo Teyseyre
Daniela Godoy

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