원문정보
초록
영어
Online Social Networks (OSNs) accumulate a large amount of user-generated data and Social Recommender Systems (SRSs) can help users discover information they are interested. However, most of the existing SRSs do not have good scalabilities to process huge volumes of data. Aiming to this problem we design a social recommender system named SRSH, which is based on Hadoop parallel computing platform. SRSH provides second-degree friends, similar users, user community and content recommendation modules, which can meet user needs of finding potential friends and attractive content. Especially, every core methods existing in these modules above can be implemented using MapReduce parallel programming framework and run in Hadoop cluster. We have conducted extensive related experiments on the realistic dataset and the experimental results show that SRSH scales well and has the ability of dealing with the problem of recommendation in the large-scale OSN.
목차
1. Introduction
2. Related Work
3. System Overview
4. Core algorithms implemented using MapReduce
4.1 Second-degree Friends Recommendation (SDFR)
4.2. Similar Users Recommendation (SUR)
4.3. User Community Recommendation (UCR)
4.4. Content Recommendation (CR)
5. Experimental Results and Analysis
6. Conclusions and Future work
Acknowledgements
References