Restricted Boltzmann Machine based active learning for sparse recommendation

Weiqing Wang, Hongzhi Yin, Zi Huang, Xiaoshuai Sun, Nguyen Quoc Viet Hung

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

Abstract

In recommender systems, users’ preferences are expressed as ratings (either explicit or implicit) for items. In general, more ratings associated with users or items are elicited, more effective the recommendations are. However, almost all user rating datasets are sparse in the real-world applications. To acquire more ratings, the active learning based methods have been used to selectively choose the items (called interview items) to ask users for rating, inspired by that the usefulness of each rating may vary significantly, i.e., different ratings may bring a different amount of information about the user’s tastes. Nevertheless, existing active learning based methods, including both static methods and decision-tree based methods, encounter the following limitations. First, the interview item set is predefined in the static methods, and they do not consider the user’s responses when asking the next question in the interview process. Second, the interview item set in the decision tree based methods is very small (i.e., usually less than 50 items), which leads to that the interview items cannot fully reflect or capture the diverse user interests, and most items do not have the opportunity to obtain additional ratings. Moreover, these decision tree based methods tend to choose popular items as the interview items instead of items with sparse ratings (i.e., sparse items), resulting in “Harry Potter Effect” (http://bickson.blogspot.com.au/2012/09/harry-potter-effect-on-recommendations.html). To address these limitations, we propose a new active learning framework based on RBM (Restricted Boltzmann Machines) to add ratings for sparse recommendation in this paper. The superiority of this method is demonstrated on two publicly available real-life datasets.

Original languageEnglish
Title of host publicationDatabase Systems for Advanced Applications
Subtitle of host publication23rd International Conference, DASFAA 2018, Gold Coast, QLD, Australia, May 21-24, 2018, Proceedings, Part I
EditorsJian Pei, Yannis Manolopoulos, Shazia Sadiq, Jianxin Li
Place of PublicationCham Switzerland
PublisherSpringer
Pages100-115
Number of pages16
ISBN (Electronic)9783319914527
ISBN (Print)9783319914510
DOIs
Publication statusPublished - 2018
Externally publishedYes
EventDatabase Systems for Advanced Applications 2018 - Gold Coast, Australia
Duration: 21 May 201824 May 2018
Conference number: 23rd
https://www.springer.com/us/book/9783319914541 (Proceedings)

Publication series

NameLecture Notes in Computer Science
PublisherSpringer
Volume10827
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

ConferenceDatabase Systems for Advanced Applications 2018
Abbreviated titleDASFAA 2018
CountryAustralia
CityGold Coast
Period21/05/1824/05/18
Internet address

Cite this

Wang, W., Yin, H., Huang, Z., Sun, X., & Hung, N. Q. V. (2018). Restricted Boltzmann Machine based active learning for sparse recommendation. In J. Pei, Y. Manolopoulos, S. Sadiq, & J. Li (Eds.), Database Systems for Advanced Applications : 23rd International Conference, DASFAA 2018, Gold Coast, QLD, Australia, May 21-24, 2018, Proceedings, Part I (pp. 100-115). (Lecture Notes in Computer Science ; Vol. 10827 ). Cham Switzerland : Springer. https://doi.org/10.1007/978-3-319-91452-7_7
Wang, Weiqing ; Yin, Hongzhi ; Huang, Zi ; Sun, Xiaoshuai ; Hung, Nguyen Quoc Viet. / Restricted Boltzmann Machine based active learning for sparse recommendation. Database Systems for Advanced Applications : 23rd International Conference, DASFAA 2018, Gold Coast, QLD, Australia, May 21-24, 2018, Proceedings, Part I. editor / Jian Pei ; Yannis Manolopoulos ; Shazia Sadiq ; Jianxin Li. Cham Switzerland : Springer, 2018. pp. 100-115 (Lecture Notes in Computer Science ).
@inproceedings{c7bb28441ec646899e2cf7f6a26fad14,
title = "Restricted Boltzmann Machine based active learning for sparse recommendation",
abstract = "In recommender systems, users’ preferences are expressed as ratings (either explicit or implicit) for items. In general, more ratings associated with users or items are elicited, more effective the recommendations are. However, almost all user rating datasets are sparse in the real-world applications. To acquire more ratings, the active learning based methods have been used to selectively choose the items (called interview items) to ask users for rating, inspired by that the usefulness of each rating may vary significantly, i.e., different ratings may bring a different amount of information about the user’s tastes. Nevertheless, existing active learning based methods, including both static methods and decision-tree based methods, encounter the following limitations. First, the interview item set is predefined in the static methods, and they do not consider the user’s responses when asking the next question in the interview process. Second, the interview item set in the decision tree based methods is very small (i.e., usually less than 50 items), which leads to that the interview items cannot fully reflect or capture the diverse user interests, and most items do not have the opportunity to obtain additional ratings. Moreover, these decision tree based methods tend to choose popular items as the interview items instead of items with sparse ratings (i.e., sparse items), resulting in “Harry Potter Effect” (http://bickson.blogspot.com.au/2012/09/harry-potter-effect-on-recommendations.html). To address these limitations, we propose a new active learning framework based on RBM (Restricted Boltzmann Machines) to add ratings for sparse recommendation in this paper. The superiority of this method is demonstrated on two publicly available real-life datasets.",
author = "Weiqing Wang and Hongzhi Yin and Zi Huang and Xiaoshuai Sun and Hung, {Nguyen Quoc Viet}",
year = "2018",
doi = "10.1007/978-3-319-91452-7_7",
language = "English",
isbn = "9783319914510",
series = "Lecture Notes in Computer Science",
publisher = "Springer",
pages = "100--115",
editor = "Pei, {Jian } and Manolopoulos, {Yannis } and Sadiq, {Shazia } and Li, {Jianxin }",
booktitle = "Database Systems for Advanced Applications",

}

Wang, W, Yin, H, Huang, Z, Sun, X & Hung, NQV 2018, Restricted Boltzmann Machine based active learning for sparse recommendation. in J Pei, Y Manolopoulos, S Sadiq & J Li (eds), Database Systems for Advanced Applications : 23rd International Conference, DASFAA 2018, Gold Coast, QLD, Australia, May 21-24, 2018, Proceedings, Part I. Lecture Notes in Computer Science , vol. 10827 , Springer, Cham Switzerland , pp. 100-115, Database Systems for Advanced Applications 2018, Gold Coast, Australia, 21/05/18. https://doi.org/10.1007/978-3-319-91452-7_7

Restricted Boltzmann Machine based active learning for sparse recommendation. / Wang, Weiqing; Yin, Hongzhi; Huang, Zi; Sun, Xiaoshuai; Hung, Nguyen Quoc Viet.

Database Systems for Advanced Applications : 23rd International Conference, DASFAA 2018, Gold Coast, QLD, Australia, May 21-24, 2018, Proceedings, Part I. ed. / Jian Pei; Yannis Manolopoulos; Shazia Sadiq; Jianxin Li. Cham Switzerland : Springer, 2018. p. 100-115 (Lecture Notes in Computer Science ; Vol. 10827 ).

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

TY - GEN

T1 - Restricted Boltzmann Machine based active learning for sparse recommendation

AU - Wang, Weiqing

AU - Yin, Hongzhi

AU - Huang, Zi

AU - Sun, Xiaoshuai

AU - Hung, Nguyen Quoc Viet

PY - 2018

Y1 - 2018

N2 - In recommender systems, users’ preferences are expressed as ratings (either explicit or implicit) for items. In general, more ratings associated with users or items are elicited, more effective the recommendations are. However, almost all user rating datasets are sparse in the real-world applications. To acquire more ratings, the active learning based methods have been used to selectively choose the items (called interview items) to ask users for rating, inspired by that the usefulness of each rating may vary significantly, i.e., different ratings may bring a different amount of information about the user’s tastes. Nevertheless, existing active learning based methods, including both static methods and decision-tree based methods, encounter the following limitations. First, the interview item set is predefined in the static methods, and they do not consider the user’s responses when asking the next question in the interview process. Second, the interview item set in the decision tree based methods is very small (i.e., usually less than 50 items), which leads to that the interview items cannot fully reflect or capture the diverse user interests, and most items do not have the opportunity to obtain additional ratings. Moreover, these decision tree based methods tend to choose popular items as the interview items instead of items with sparse ratings (i.e., sparse items), resulting in “Harry Potter Effect” (http://bickson.blogspot.com.au/2012/09/harry-potter-effect-on-recommendations.html). To address these limitations, we propose a new active learning framework based on RBM (Restricted Boltzmann Machines) to add ratings for sparse recommendation in this paper. The superiority of this method is demonstrated on two publicly available real-life datasets.

AB - In recommender systems, users’ preferences are expressed as ratings (either explicit or implicit) for items. In general, more ratings associated with users or items are elicited, more effective the recommendations are. However, almost all user rating datasets are sparse in the real-world applications. To acquire more ratings, the active learning based methods have been used to selectively choose the items (called interview items) to ask users for rating, inspired by that the usefulness of each rating may vary significantly, i.e., different ratings may bring a different amount of information about the user’s tastes. Nevertheless, existing active learning based methods, including both static methods and decision-tree based methods, encounter the following limitations. First, the interview item set is predefined in the static methods, and they do not consider the user’s responses when asking the next question in the interview process. Second, the interview item set in the decision tree based methods is very small (i.e., usually less than 50 items), which leads to that the interview items cannot fully reflect or capture the diverse user interests, and most items do not have the opportunity to obtain additional ratings. Moreover, these decision tree based methods tend to choose popular items as the interview items instead of items with sparse ratings (i.e., sparse items), resulting in “Harry Potter Effect” (http://bickson.blogspot.com.au/2012/09/harry-potter-effect-on-recommendations.html). To address these limitations, we propose a new active learning framework based on RBM (Restricted Boltzmann Machines) to add ratings for sparse recommendation in this paper. The superiority of this method is demonstrated on two publicly available real-life datasets.

UR - http://www.scopus.com/inward/record.url?scp=85048061080&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-91452-7_7

DO - 10.1007/978-3-319-91452-7_7

M3 - Conference Paper

SN - 9783319914510

T3 - Lecture Notes in Computer Science

SP - 100

EP - 115

BT - Database Systems for Advanced Applications

A2 - Pei, Jian

A2 - Manolopoulos, Yannis

A2 - Sadiq, Shazia

A2 - Li, Jianxin

PB - Springer

CY - Cham Switzerland

ER -

Wang W, Yin H, Huang Z, Sun X, Hung NQV. Restricted Boltzmann Machine based active learning for sparse recommendation. In Pei J, Manolopoulos Y, Sadiq S, Li J, editors, Database Systems for Advanced Applications : 23rd International Conference, DASFAA 2018, Gold Coast, QLD, Australia, May 21-24, 2018, Proceedings, Part I. Cham Switzerland : Springer. 2018. p. 100-115. (Lecture Notes in Computer Science ). https://doi.org/10.1007/978-3-319-91452-7_7