Parallelizing random walk with restart for large-scale query recommendation

Meng Fen Chiang*, Tsung Wei Wang, Wen-Chih Peng

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Random Walk with Restart (abbreviated as RWR) has been widely employed in Web search and recommendation systems and several performance enhancement approaches for RWR have been proposed to save storage costs and improve the on-line response time. In this paper, we explore and implement RWR for query recommendation in Yahoo! Asia Knowledge Plus, which contains a huge amount of Question and Answer Web pages (abbreviated as QA). From user click logs, we first discover temporal following patterns that indicates frequent QA browsing behaviors of users within a pre-defined time window. In light of temporal following patterns, a graph structure is built for RWR. Since users may submit their queries at the same time, we design a parallel approach for the implementation of RWR in a cloud computing environment. Empirical results on Yahoo! Asia Knowledge Plus dataset demonstrates that our RWR-based recommendation is able to effectively and efficiently recommend related QA Web pages.

Original languageEnglish
Title of host publicationProceedings of the 2010 Workshop on Massive Data Analytics on the Cloud, MDAC 2010, in Association with the 19th Annual World Wide Web Conference, WWW2010
DOIs
StatePublished - 16 Jul 2010
Event2010 Workshop on Massive Data Analytics on the Cloud, MDAC 2010, in Association with the 19th Annual World Wide Web Conference, WWW2010 - Raleigh, NC, United States
Duration: 26 Apr 201026 Apr 2010

Publication series

NameACM International Conference Proceeding Series

Conference

Conference2010 Workshop on Massive Data Analytics on the Cloud, MDAC 2010, in Association with the 19th Annual World Wide Web Conference, WWW2010
CountryUnited States
CityRaleigh, NC
Period26/04/1026/04/10

Keywords

  • parallel computing
  • recommender systems
  • temporal relation

Fingerprint Dive into the research topics of 'Parallelizing random walk with restart for large-scale query recommendation'. Together they form a unique fingerprint.

Cite this