A Distributed Real-Time Recommender System for Big Data Streams

Clicks: 45
ID: 283117
2022
Article Quality & Performance Metrics
Overall Quality
Not rated
Combines reader engagement with the AI quality analysis. This article has not been analysed, so there is no overall score — reader engagement is measured and shown alongside.
AI Quality Assessment
Not analyzed
Readership in this journal
Steady

Ranked #605 of 803 articles by views in arXiv

Most read Least read

Bar heights use a square-root scale. Only the 120 most-read articles are drawn; the journal has 803 in total.

Mint this article as an NFT
Not yet minted

Create a permanent, verifiable on-chain record of this article on the Scimatic Network. The NFT is held in your Journament account, and you can withdraw it to your own wallet at any time.

5 SUSD one-off · no wallet required
Abstract
In today's data-driven world, recommender systems (RS) play a crucial role to support the decision-making process. As users become continuously connected to the internet, they become less patient and less tolerant to obsolete recommendations made by an RS, e.g., movie recommendations on Netflix or books to read on Amazon. This, in turn, requires continuous training of the RS to cope with both the online fashion of data and the changing nature of user tastes and interests, known as concept drift. Streaming (online) RS has to address three requirements: continuous training and recommendation, handling concept drifts, and ability to scale. Streaming recommender systems proposed in the literature mostly, address the first two requirements and do not consider scalability. That is because they run the training process on a single machine. Such a machine, no matter how powerful it is, will eventually fail to cope with the volume of the data, a lesson learned from big data processing. To tackle the third challenge, we propose a Splitting and Replication mechanism for building distributed streaming recommender systems. Our mechanism is inspired by the successful shared-nothing architecture that underpins contemporary big data processing systems. We have applied our mechanism to two well-known approaches for online recommender systems, namely, matrix factorization and item-based collaborative filtering. We have implemented our mechanism on top of Apache Flink. We conducted experiments comparing the performance of the baseline (single machine) approach with our distributed approach. Evaluating different data sets, improvement in processing latency, throughput, and accuracy have been observed. Our experiments show online recall improvement by 40\% with more than 50\% less memory consumption.
Reference Key
hassan2022a Use this key to autocite in the manuscript while using SciMatic Manuscript Manager or Thesis Manager
Authors Heidy Hazem; Ahmed Awad; Ahmed Hassan
Journal arXiv
Year 2022
DOI
DOI not found
URL
Keywords

Citations

No citations found. To add a citation, contact the admin at info@scimatic.org

No comments yet. Be the first to comment on this article.