Minimal interaction search in recommender systems

Branislav Kveton; Shlomo Berkovsky

doi:10.1145/2678025.2701367

Minimal interaction search in recommender systems

Branislav Kveton, Shlomo Berkovsky

Research output: Chapter in Book/Report/Conference proceeding › Conference proceeding contribution › peer-review

6 Citations (Scopus)

Abstract

While numerous works study algorithms for predicting item ratings in recommender systems, the area of the user-recommender interaction remains largely under-explored. In this work, we look into user interaction with the recommendation list, aiming to devise a method that allows users to discover items of interest in a minimal number of interactions. We propose generalized linear search (GLS), a combination of linear and generalized searches that brings together the benefits of both approaches. We prove that GLS performs at least as well as generalized search and compare our method to several baselines and heuristics. Our evaluation shows that GLS is liked by the users and achieves the shortest interactions.

Original language	English
Title of host publication	Proceedings of the 20th International Conference on Intelligent User Interfaces, IUI'15
Publisher	Association for Computing Machinery (ACM)
Pages	236-246
Number of pages	11
ISBN (Electronic)	9781450333061
DOIs	https://doi.org/10.1145/2678025.2701367
Publication status	Published - 2015
Externally published	Yes
Event	20th International Conference on Intelligent User Interfaces, IUI 2015 - Atlanta, United States Duration: 29 Mar 2015 → 1 Apr 2015

Conference

Conference	20th International Conference on Intelligent User Interfaces, IUI 2015
Country/Territory	United States
City	Atlanta
Period	29/03/15 → 1/04/15

Keywords

Human-computer interaction
interactive search
generalized binary search
recommender systems
active learning

Access to Document

10.1145/2678025.2701367

Cite this

@inproceedings{fd844615eeb14411b5085741e9b39349,

title = "Minimal interaction search in recommender systems",

abstract = "While numerous works study algorithms for predicting item ratings in recommender systems, the area of the user-recommender interaction remains largely under-explored. In this work, we look into user interaction with the recommendation list, aiming to devise a method that allows users to discover items of interest in a minimal number of interactions. We propose generalized linear search (GLS), a combination of linear and generalized searches that brings together the benefits of both approaches. We prove that GLS performs at least as well as generalized search and compare our method to several baselines and heuristics. Our evaluation shows that GLS is liked by the users and achieves the shortest interactions.",

keywords = "Human-computer interaction, interactive search, generalized binary search, recommender systems, active learning",

author = "Branislav Kveton and Shlomo Berkovsky",

year = "2015",

doi = "10.1145/2678025.2701367",

language = "English",

pages = "236--246",

booktitle = "Proceedings of the 20th International Conference on Intelligent User Interfaces, IUI'15",

publisher = "Association for Computing Machinery (ACM)",

address = "United States",

note = "20th International Conference on Intelligent User Interfaces, IUI 2015 ; Conference date: 29-03-2015 Through 01-04-2015",

}

Kveton, B & Berkovsky, S 2015, Minimal interaction search in recommender systems. in Proceedings of the 20th International Conference on Intelligent User Interfaces, IUI'15. Association for Computing Machinery (ACM), pp. 236-246, 20th International Conference on Intelligent User Interfaces, IUI 2015, Atlanta, Georgia, United States, 29/03/15. https://doi.org/10.1145/2678025.2701367

TY - GEN

T1 - Minimal interaction search in recommender systems

AU - Kveton, Branislav

AU - Berkovsky, Shlomo

PY - 2015

Y1 - 2015

N2 - While numerous works study algorithms for predicting item ratings in recommender systems, the area of the user-recommender interaction remains largely under-explored. In this work, we look into user interaction with the recommendation list, aiming to devise a method that allows users to discover items of interest in a minimal number of interactions. We propose generalized linear search (GLS), a combination of linear and generalized searches that brings together the benefits of both approaches. We prove that GLS performs at least as well as generalized search and compare our method to several baselines and heuristics. Our evaluation shows that GLS is liked by the users and achieves the shortest interactions.

AB - While numerous works study algorithms for predicting item ratings in recommender systems, the area of the user-recommender interaction remains largely under-explored. In this work, we look into user interaction with the recommendation list, aiming to devise a method that allows users to discover items of interest in a minimal number of interactions. We propose generalized linear search (GLS), a combination of linear and generalized searches that brings together the benefits of both approaches. We prove that GLS performs at least as well as generalized search and compare our method to several baselines and heuristics. Our evaluation shows that GLS is liked by the users and achieves the shortest interactions.

KW - Human-computer interaction

KW - interactive search

KW - generalized binary search

KW - recommender systems

KW - active learning

UR - http://www.scopus.com/inward/record.url?scp=84939610756&partnerID=8YFLogxK

U2 - 10.1145/2678025.2701367

DO - 10.1145/2678025.2701367

M3 - Conference proceeding contribution

SP - 236

EP - 246

BT - Proceedings of the 20th International Conference on Intelligent User Interfaces, IUI'15

PB - Association for Computing Machinery (ACM)

T2 - 20th International Conference on Intelligent User Interfaces, IUI 2015

Y2 - 29 March 2015 through 1 April 2015

ER -

Minimal interaction search in recommender systems

Abstract

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this