%0 Conference Proceedings
%T Cheap and Cheerful: Trading Speed and Quality for Scalable Social Recommenders
%+ As Scalable As Possible: foundations of large scale dynamic distributed systems (ASAP)
%+ Université de Rennes (UR)
%A Kermarrec, Anne-Marie
%A Taïani, François
%A Tirado Martin, Juan Manuel
%< avec comité de lecture
%( Lecture Notes in Computer Science
%B 15th IFIP International Conference on Distributed Applications and Interoperable Systems (DAIS)
%C Grenoble, France
%Y Alysson Bessani
%Y Sara Bouchenak
%I Springer International Publishing
%3 Distributed Applications and Interoperable Systems
%V LNCS-9038
%P 138-151
%8 2015-06-02
%D 2015
%R 10.1007/978-3-319-19129-4_11
%K Score Function, Social Distance, Multivariate Adaptive Regression Spline, Link Prediction, Social Graph
%Z Computer Science [cs]/Distributed, Parallel, and Cluster Computing [cs.DC]
%Z Computer Science [cs]/Information Retrieval [cs.IR]
%Z Computer Science [cs]/Social and Information Networks [cs.SI]Conference papers
%X Recommending appropriate content and users is a critical feature of on-line social networks. Computing accurate recommendations on very large datasets can however be particularly costly in terms of resources , even on modern parallel and distributed infrastructures. As a result, modern recommenders must generally trade-off quality and computational cost to reach a practical solution. This trade-off has however so far been largely left unexplored by the research community, making it difficult for practitioners to reach informed design decisions. In this paper, we investigate to which extent the additional computing costs of advanced recommendation techniques based on supervised classifiers can be balanced by the gains they bring in terms of quality. In particular , we compare these recommenders against their unsupervised counterparts , which offer lightweight and highly scalable alternatives. We propose a thorough evaluation comparing 11 classifiers against 7 lightweight recommenders on a real Twitter dataset. Additionally, we explore data grouping as a method to reduce computational costs in a distributed setting while improving recommendation quality. We demonstrate how classifiers trained using data grouping can reduce their computing time by 6 while improving recommendations up to 22% when compared with lightweight solutions.
%G English
%Z TC 6
%Z WG 6.1
%2 https://inria.hal.science/hal-01170757/document
%2 https://inria.hal.science/hal-01170757/file/summary.pdf
%L hal-01170757
%U https://inria.hal.science/hal-01170757
%~ INSTITUT-TELECOM
%~ UNIV-RENNES1
%~ CNRS
%~ INRIA
%~ UNIV-UBS
%~ INSA-RENNES
%~ INRIA-RENNES
%~ IRISA
%~ IRISA_SET
%~ INRIA_TEST
%~ TESTALAIN1
%~ IFIP-LNCS
%~ IFIP
%~ CENTRALESUPELEC
%~ IRISA-D1
%~ INRIA2
%~ IFIP-TC
%~ IFIP-WG
%~ IFIP-TC6
%~ IFIP-WG6-1
%~ IFIP-DAIS
%~ UR1-HAL
%~ UR1-MATH-STIC
%~ UR1-UFR-ISTIC
%~ IFIP-DISCOTEC
%~ TEST-UNIV-RENNES
%~ TEST-UR-CSS
%~ UNIV-RENNES
%~ INRIA-RENGRE
%~ IFIP-LNCS-9038
%~ INSTITUTS-TELECOM
%~ ANR
%~ UR1-MATH-NUM