Evaluating recommender behavior for new users

Daniel Kluver; Joseph A. Konstan

doi:10.1145/2645710.2645742

Evaluating recommender behavior for new users

Daniel Kluver, Joseph A. Konstan

Computer Science and Engineering

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

47 Scopus citations

Abstract

The new user experience is one of the important problems in recommender systems. Past work on recommending for new users has focused on the process of gathering information from the user. Our work focuses on how different algorithms behave for new users. We describe a methodology that we use to compare representatives of three common families of algorithms along eleven different metrics. We find that for the first few ratings a baseline algorithm performs better than three common collaborative filtering algorithms. Once we have a few ratings, we find that Funk's SVD algorithm has the best overall performance. We also find that ItemItem, a very commonly deployed algorithm, performs very poorly for new users. Our results can inform the design of interfaces and algorithms for new users.

Original language	English (US)
Title of host publication	RecSys 2014 - Proceedings of the 8th ACM Conference on Recommender Systems
Publisher	Association for Computing Machinery
Pages	121-128
Number of pages	8
ISBN (Electronic)	9781450326681
DOIs	https://doi.org/10.1145/2645710.2645742
State	Published - Oct 6 2014
Event	8th ACM Conference on Recommender Systems, RecSys 2014 - Foster City, United States Duration: Oct 6 2014 → Oct 10 2014

Publication series

Name	RecSys 2014 - Proceedings of the 8th ACM Conference on Recommender Systems

Other

Other	8th ACM Conference on Recommender Systems, RecSys 2014
Country/Territory	United States
City	Foster City
Period	10/6/14 → 10/10/14

Bibliographical note

Publisher Copyright:
Copyright © 2014 ACM.

Keywords

Evaluation
New user experience
New user problem
Profile size
Recommender systems
User cold start

Publisher link

10.1145/2645710.2645742

Find It

Find It at U of M Libraries

Cite this

Evaluating recommender behavior for new users. / Kluver, Daniel ; Konstan, Joseph A.
RecSys 2014 - Proceedings of the 8th ACM Conference on Recommender Systems. Association for Computing Machinery, 2014. p. 121-128 (RecSys 2014 - Proceedings of the 8th ACM Conference on Recommender Systems).

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

Kluver, D & Konstan, JA 2014, Evaluating recommender behavior for new users. in RecSys 2014 - Proceedings of the 8th ACM Conference on Recommender Systems. RecSys 2014 - Proceedings of the 8th ACM Conference on Recommender Systems, Association for Computing Machinery, pp. 121-128, 8th ACM Conference on Recommender Systems, RecSys 2014, Foster City, United States, 10/6/14. https://doi.org/10.1145/2645710.2645742

@inproceedings{52e203fd70f3468da032964f58697d6e,

title = "Evaluating recommender behavior for new users",

abstract = "The new user experience is one of the important problems in recommender systems. Past work on recommending for new users has focused on the process of gathering information from the user. Our work focuses on how different algorithms behave for new users. We describe a methodology that we use to compare representatives of three common families of algorithms along eleven different metrics. We find that for the first few ratings a baseline algorithm performs better than three common collaborative filtering algorithms. Once we have a few ratings, we find that Funk's SVD algorithm has the best overall performance. We also find that ItemItem, a very commonly deployed algorithm, performs very poorly for new users. Our results can inform the design of interfaces and algorithms for new users.",

keywords = "Evaluation, New user experience, New user problem, Profile size, Recommender systems, User cold start",

author = "Daniel Kluver and Konstan, {Joseph A.}",

note = "Publisher Copyright: Copyright {\textcopyright} 2014 ACM.; 8th ACM Conference on Recommender Systems, RecSys 2014 ; Conference date: 06-10-2014 Through 10-10-2014",

year = "2014",

month = oct,

day = "6",

doi = "10.1145/2645710.2645742",

language = "English (US)",

series = "RecSys 2014 - Proceedings of the 8th ACM Conference on Recommender Systems",

publisher = "Association for Computing Machinery",

pages = "121--128",

booktitle = "RecSys 2014 - Proceedings of the 8th ACM Conference on Recommender Systems",

}

TY - GEN

T1 - Evaluating recommender behavior for new users

AU - Kluver, Daniel

AU - Konstan, Joseph A.

PY - 2014/10/6

Y1 - 2014/10/6

N2 - The new user experience is one of the important problems in recommender systems. Past work on recommending for new users has focused on the process of gathering information from the user. Our work focuses on how different algorithms behave for new users. We describe a methodology that we use to compare representatives of three common families of algorithms along eleven different metrics. We find that for the first few ratings a baseline algorithm performs better than three common collaborative filtering algorithms. Once we have a few ratings, we find that Funk's SVD algorithm has the best overall performance. We also find that ItemItem, a very commonly deployed algorithm, performs very poorly for new users. Our results can inform the design of interfaces and algorithms for new users.

AB - The new user experience is one of the important problems in recommender systems. Past work on recommending for new users has focused on the process of gathering information from the user. Our work focuses on how different algorithms behave for new users. We describe a methodology that we use to compare representatives of three common families of algorithms along eleven different metrics. We find that for the first few ratings a baseline algorithm performs better than three common collaborative filtering algorithms. Once we have a few ratings, we find that Funk's SVD algorithm has the best overall performance. We also find that ItemItem, a very commonly deployed algorithm, performs very poorly for new users. Our results can inform the design of interfaces and algorithms for new users.

KW - Evaluation

KW - New user experience

KW - New user problem

KW - Profile size

KW - Recommender systems

KW - User cold start

UR - http://www.scopus.com/inward/record.url?scp=84908879792&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84908879792&partnerID=8YFLogxK

U2 - 10.1145/2645710.2645742

DO - 10.1145/2645710.2645742

M3 - Conference contribution

AN - SCOPUS:84908879792

T3 - RecSys 2014 - Proceedings of the 8th ACM Conference on Recommender Systems

SP - 121

EP - 128

BT - RecSys 2014 - Proceedings of the 8th ACM Conference on Recommender Systems

PB - Association for Computing Machinery

T2 - 8th ACM Conference on Recommender Systems, RecSys 2014

Y2 - 6 October 2014 through 10 October 2014

ER -

Evaluating recommender behavior for new users

Abstract

Publication series

Other

Bibliographical note

Keywords

Publisher link

Find It

Other files and links

Fingerprint

Cite this