DeepRec

A deep neural network approach to recommendation with item embedding and weighted loss function

Wen Zhang, Yuhang Du, Taketoshi Yoshida, Ye Yang

Research output: Contribution to journalArticle

1 Citation (Scopus)

Abstract

Traditional collaborative filtering techniques suffer from the data sparsity problem in practice. That is, only a small proportion of all items in the recommender system occur in a user's rated item list. However, in order to retrieve items meeting a user's interest, all possible candidate items should be investigated. To address this problem, this paper proposes a recommendation approach called DeepRec, based on feedforward deep neural network learning with item embedding and weighted loss function. Specifically, item embedding learns numerical vectors for item representation, and weighted loss function balances popularity and novelty of recommended items. Moreover, it introduces two strategies, i.e. sampling by random (Ran-Strategy) and sampling by distribution (Pro-Strategy), to leave one item as output and the remaining as input from each user's historically rated item list. Max-pooling and average-pooling are employed to combine individual item vectors to derive users’ input vectors for feedforward deep neural network learning. Experiments on the App dataset and the Last.fm dataset demonstrate that the proposed DeepRec approach is superior to state-of-the-art techniques in recommending Apps and songs in terms of accuracy and diversity as well as complexity.

Original languageEnglish (US)
Pages (from-to)121-140
Number of pages20
JournalInformation sciences
Volume470
DOIs
StatePublished - Jan 1 2019

Fingerprint

Loss Function
Recommendations
Neural Networks
Application programs
Pooling
Feedforward Neural Networks
Sampling
Collaborative filtering
Recommender systems
Sampling Strategy
Collaborative Filtering
Recommender Systems
Sparsity
Proportion
Deep neural networks
Neural networks
Loss function
Output
Experiments
Demonstrate

All Science Journal Classification (ASJC) codes

  • Software
  • Information Systems and Management
  • Artificial Intelligence
  • Theoretical Computer Science
  • Control and Systems Engineering
  • Computer Science Applications

Cite this

@article{c8e16ed46537476598e420e550156d4f,
title = "DeepRec: A deep neural network approach to recommendation with item embedding and weighted loss function",
abstract = "Traditional collaborative filtering techniques suffer from the data sparsity problem in practice. That is, only a small proportion of all items in the recommender system occur in a user's rated item list. However, in order to retrieve items meeting a user's interest, all possible candidate items should be investigated. To address this problem, this paper proposes a recommendation approach called DeepRec, based on feedforward deep neural network learning with item embedding and weighted loss function. Specifically, item embedding learns numerical vectors for item representation, and weighted loss function balances popularity and novelty of recommended items. Moreover, it introduces two strategies, i.e. sampling by random (Ran-Strategy) and sampling by distribution (Pro-Strategy), to leave one item as output and the remaining as input from each user's historically rated item list. Max-pooling and average-pooling are employed to combine individual item vectors to derive users’ input vectors for feedforward deep neural network learning. Experiments on the App dataset and the Last.fm dataset demonstrate that the proposed DeepRec approach is superior to state-of-the-art techniques in recommending Apps and songs in terms of accuracy and diversity as well as complexity.",
author = "Wen Zhang and Yuhang Du and Taketoshi Yoshida and Ye Yang",
year = "2019",
month = "1",
day = "1",
doi = "https://doi.org/10.1016/j.ins.2018.08.039",
language = "English (US)",
volume = "470",
pages = "121--140",
journal = "Information Sciences",
issn = "0020-0255",
publisher = "Elsevier Inc.",

}

DeepRec : A deep neural network approach to recommendation with item embedding and weighted loss function. / Zhang, Wen; Du, Yuhang; Yoshida, Taketoshi; Yang, Ye.

In: Information sciences, Vol. 470, 01.01.2019, p. 121-140.

Research output: Contribution to journalArticle

TY - JOUR

T1 - DeepRec

T2 - A deep neural network approach to recommendation with item embedding and weighted loss function

AU - Zhang, Wen

AU - Du, Yuhang

AU - Yoshida, Taketoshi

AU - Yang, Ye

PY - 2019/1/1

Y1 - 2019/1/1

N2 - Traditional collaborative filtering techniques suffer from the data sparsity problem in practice. That is, only a small proportion of all items in the recommender system occur in a user's rated item list. However, in order to retrieve items meeting a user's interest, all possible candidate items should be investigated. To address this problem, this paper proposes a recommendation approach called DeepRec, based on feedforward deep neural network learning with item embedding and weighted loss function. Specifically, item embedding learns numerical vectors for item representation, and weighted loss function balances popularity and novelty of recommended items. Moreover, it introduces two strategies, i.e. sampling by random (Ran-Strategy) and sampling by distribution (Pro-Strategy), to leave one item as output and the remaining as input from each user's historically rated item list. Max-pooling and average-pooling are employed to combine individual item vectors to derive users’ input vectors for feedforward deep neural network learning. Experiments on the App dataset and the Last.fm dataset demonstrate that the proposed DeepRec approach is superior to state-of-the-art techniques in recommending Apps and songs in terms of accuracy and diversity as well as complexity.

AB - Traditional collaborative filtering techniques suffer from the data sparsity problem in practice. That is, only a small proportion of all items in the recommender system occur in a user's rated item list. However, in order to retrieve items meeting a user's interest, all possible candidate items should be investigated. To address this problem, this paper proposes a recommendation approach called DeepRec, based on feedforward deep neural network learning with item embedding and weighted loss function. Specifically, item embedding learns numerical vectors for item representation, and weighted loss function balances popularity and novelty of recommended items. Moreover, it introduces two strategies, i.e. sampling by random (Ran-Strategy) and sampling by distribution (Pro-Strategy), to leave one item as output and the remaining as input from each user's historically rated item list. Max-pooling and average-pooling are employed to combine individual item vectors to derive users’ input vectors for feedforward deep neural network learning. Experiments on the App dataset and the Last.fm dataset demonstrate that the proposed DeepRec approach is superior to state-of-the-art techniques in recommending Apps and songs in terms of accuracy and diversity as well as complexity.

UR - http://www.scopus.com/inward/record.url?scp=85052747217&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85052747217&partnerID=8YFLogxK

U2 - https://doi.org/10.1016/j.ins.2018.08.039

DO - https://doi.org/10.1016/j.ins.2018.08.039

M3 - Article

VL - 470

SP - 121

EP - 140

JO - Information Sciences

JF - Information Sciences

SN - 0020-0255

ER -