Latent features based prediction on new users' tastes

Ming Chu Chen, Hung-Yu Kao

Research output: Contribution to journalConference article

Abstract

Recommendation systems have become popular in recent years. A key challenge in such systems is how to effectively characterize new users' tastes - an issue that is generally known as the cold-start problem. New users judge the system by the ability to immediately provide them with what they consider interesting. A general method for solving the cold-start problem is to elicit information about new users by having them provide answers to interview questions. In this paper, we present Matrix Factorization K-Means (MFK), a novel method to solve the problem of interview question construction. MFK first learns the latent features of the user and the item through observed rating data and then determines the best interview questions based on the clusters of latent features. We can determine similar groups of users after obtaining the responses to the interview questions. Such recommendation systems can indicate new users' tastes according to their responses to the interview questions. In our experiments, we evaluate our methods using a public dataset for recommendations. The results show that our method leads to better performance than other baselines.

Original languageEnglish
Pages (from-to)176-187
Number of pages12
JournalLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume8443 LNAI
Issue numberPART 1
DOIs
Publication statusPublished - 2014 Jan 1
Event18th Pacific-Asia Conference on Advances in Knowledge Discovery and Data Mining, PAKDD 2014 - Tainan, Taiwan
Duration: 2014 May 132014 May 16

Fingerprint

Recommender systems
Factorization
Prediction
Recommendation System
Matrix Factorization
K-means
Experiments
Immediately
Recommendations
Baseline
Evaluate
Experiment

All Science Journal Classification (ASJC) codes

  • Theoretical Computer Science
  • Computer Science(all)

Cite this

@article{a568100dbb0041ad8a222b1a529f015c,
title = "Latent features based prediction on new users' tastes",
abstract = "Recommendation systems have become popular in recent years. A key challenge in such systems is how to effectively characterize new users' tastes - an issue that is generally known as the cold-start problem. New users judge the system by the ability to immediately provide them with what they consider interesting. A general method for solving the cold-start problem is to elicit information about new users by having them provide answers to interview questions. In this paper, we present Matrix Factorization K-Means (MFK), a novel method to solve the problem of interview question construction. MFK first learns the latent features of the user and the item through observed rating data and then determines the best interview questions based on the clusters of latent features. We can determine similar groups of users after obtaining the responses to the interview questions. Such recommendation systems can indicate new users' tastes according to their responses to the interview questions. In our experiments, we evaluate our methods using a public dataset for recommendations. The results show that our method leads to better performance than other baselines.",
author = "Chen, {Ming Chu} and Hung-Yu Kao",
year = "2014",
month = "1",
day = "1",
doi = "10.1007/978-3-319-06608-0_15",
language = "English",
volume = "8443 LNAI",
pages = "176--187",
journal = "Lecture Notes in Computer Science",
issn = "0302-9743",
publisher = "Springer Verlag",
number = "PART 1",

}

Latent features based prediction on new users' tastes. / Chen, Ming Chu; Kao, Hung-Yu.

In: Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), Vol. 8443 LNAI, No. PART 1, 01.01.2014, p. 176-187.

Research output: Contribution to journalConference article

TY - JOUR

T1 - Latent features based prediction on new users' tastes

AU - Chen, Ming Chu

AU - Kao, Hung-Yu

PY - 2014/1/1

Y1 - 2014/1/1

N2 - Recommendation systems have become popular in recent years. A key challenge in such systems is how to effectively characterize new users' tastes - an issue that is generally known as the cold-start problem. New users judge the system by the ability to immediately provide them with what they consider interesting. A general method for solving the cold-start problem is to elicit information about new users by having them provide answers to interview questions. In this paper, we present Matrix Factorization K-Means (MFK), a novel method to solve the problem of interview question construction. MFK first learns the latent features of the user and the item through observed rating data and then determines the best interview questions based on the clusters of latent features. We can determine similar groups of users after obtaining the responses to the interview questions. Such recommendation systems can indicate new users' tastes according to their responses to the interview questions. In our experiments, we evaluate our methods using a public dataset for recommendations. The results show that our method leads to better performance than other baselines.

AB - Recommendation systems have become popular in recent years. A key challenge in such systems is how to effectively characterize new users' tastes - an issue that is generally known as the cold-start problem. New users judge the system by the ability to immediately provide them with what they consider interesting. A general method for solving the cold-start problem is to elicit information about new users by having them provide answers to interview questions. In this paper, we present Matrix Factorization K-Means (MFK), a novel method to solve the problem of interview question construction. MFK first learns the latent features of the user and the item through observed rating data and then determines the best interview questions based on the clusters of latent features. We can determine similar groups of users after obtaining the responses to the interview questions. Such recommendation systems can indicate new users' tastes according to their responses to the interview questions. In our experiments, we evaluate our methods using a public dataset for recommendations. The results show that our method leads to better performance than other baselines.

UR - http://www.scopus.com/inward/record.url?scp=84901255962&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84901255962&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-06608-0_15

DO - 10.1007/978-3-319-06608-0_15

M3 - Conference article

VL - 8443 LNAI

SP - 176

EP - 187

JO - Lecture Notes in Computer Science

JF - Lecture Notes in Computer Science

SN - 0302-9743

IS - PART 1

ER -