Learning to question: Leveraging user preferences for shopping advice

Mahashweta Das, Gianmarco Morales, Aristides Gionis, Ingmar Weber

Research output: Chapter in Book/Report/Conference proceedingConference contribution

12 Citations (Scopus)

Abstract

We present ShoppingAdvisor, a novel recommender system that helps users in shopping for technical products. ShoppingAdvisor leverages both user preferences and technical product attributes in order to generate its suggestions. The system elicits user preferences via a tree-shaped owchart, where each node is a question to the user. At each node, ShoppingAdvisor suggests a ranking of products matching the preferences of the user, and that gets progressively refined along the path from the tree's root to one of its leafs. In this paper we show (i) how to learn the structure of the tree, i.e., which questions to ask at each node, and (ii) how to produce a suitable ranking at each node. First, we adapt the classical top-down strategy for building decision trees in order to find the best user attribute to ask at each node. Differently from decision trees, ShoppingAdvisor partitions the user space rather than the product space. Second, we show how to employ a learning-To-rank approach in order to learn, for each node of the tree, a ranking of products appropriate to the users who reach that node. We experiment with two real-world datasets for cars and cameras, and a synthetic one. We use mean reciprocal rank to evaluate ShoppingAdvisor, and show how the performance increases by more than 50% along the path from root to leaf. We also show how collaborative recommendation algorithms such as k-nearest neighbor benefits from feature selection done by the ShoppingAdvisor tree. Our experiments show that ShoppingAdvisor produces good quality interpretable recommendations, while requiring less input from users and being able to handle the cold-start problem.

Original languageEnglish
Title of host publicationKDD 2013 - 19th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining
PublisherAssociation for Computing Machinery
Pages203-211
Number of pages9
VolumePart F128815
ISBN (Electronic)9781450321747
DOIs
Publication statusPublished - 11 Aug 2013
Event19th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, KDD 2013 - Chicago, United States
Duration: 11 Aug 201314 Aug 2013

Other

Other19th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, KDD 2013
CountryUnited States
CityChicago
Period11/8/1314/8/13

Fingerprint

Decision trees
Recommender systems
Feature extraction
Railroad cars
Experiments
Cameras

Keywords

  • Collaborative content
  • Learning
  • Ranking
  • Recommendation

ASJC Scopus subject areas

  • Software
  • Information Systems

Cite this

Das, M., Morales, G., Gionis, A., & Weber, I. (2013). Learning to question: Leveraging user preferences for shopping advice. In KDD 2013 - 19th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (Vol. Part F128815, pp. 203-211). [2487653] Association for Computing Machinery. https://doi.org/10.1145/2487575.2487653

Learning to question : Leveraging user preferences for shopping advice. / Das, Mahashweta; Morales, Gianmarco; Gionis, Aristides; Weber, Ingmar.

KDD 2013 - 19th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. Vol. Part F128815 Association for Computing Machinery, 2013. p. 203-211 2487653.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Das, M, Morales, G, Gionis, A & Weber, I 2013, Learning to question: Leveraging user preferences for shopping advice. in KDD 2013 - 19th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. vol. Part F128815, 2487653, Association for Computing Machinery, pp. 203-211, 19th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, KDD 2013, Chicago, United States, 11/8/13. https://doi.org/10.1145/2487575.2487653
Das M, Morales G, Gionis A, Weber I. Learning to question: Leveraging user preferences for shopping advice. In KDD 2013 - 19th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. Vol. Part F128815. Association for Computing Machinery. 2013. p. 203-211. 2487653 https://doi.org/10.1145/2487575.2487653
Das, Mahashweta ; Morales, Gianmarco ; Gionis, Aristides ; Weber, Ingmar. / Learning to question : Leveraging user preferences for shopping advice. KDD 2013 - 19th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. Vol. Part F128815 Association for Computing Machinery, 2013. pp. 203-211
@inproceedings{54945f177f024990851a72dc81972687,
title = "Learning to question: Leveraging user preferences for shopping advice",
abstract = "We present ShoppingAdvisor, a novel recommender system that helps users in shopping for technical products. ShoppingAdvisor leverages both user preferences and technical product attributes in order to generate its suggestions. The system elicits user preferences via a tree-shaped owchart, where each node is a question to the user. At each node, ShoppingAdvisor suggests a ranking of products matching the preferences of the user, and that gets progressively refined along the path from the tree's root to one of its leafs. In this paper we show (i) how to learn the structure of the tree, i.e., which questions to ask at each node, and (ii) how to produce a suitable ranking at each node. First, we adapt the classical top-down strategy for building decision trees in order to find the best user attribute to ask at each node. Differently from decision trees, ShoppingAdvisor partitions the user space rather than the product space. Second, we show how to employ a learning-To-rank approach in order to learn, for each node of the tree, a ranking of products appropriate to the users who reach that node. We experiment with two real-world datasets for cars and cameras, and a synthetic one. We use mean reciprocal rank to evaluate ShoppingAdvisor, and show how the performance increases by more than 50{\%} along the path from root to leaf. We also show how collaborative recommendation algorithms such as k-nearest neighbor benefits from feature selection done by the ShoppingAdvisor tree. Our experiments show that ShoppingAdvisor produces good quality interpretable recommendations, while requiring less input from users and being able to handle the cold-start problem.",
keywords = "Collaborative content, Learning, Ranking, Recommendation",
author = "Mahashweta Das and Gianmarco Morales and Aristides Gionis and Ingmar Weber",
year = "2013",
month = "8",
day = "11",
doi = "10.1145/2487575.2487653",
language = "English",
volume = "Part F128815",
pages = "203--211",
booktitle = "KDD 2013 - 19th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining",
publisher = "Association for Computing Machinery",

}

TY - GEN

T1 - Learning to question

T2 - Leveraging user preferences for shopping advice

AU - Das, Mahashweta

AU - Morales, Gianmarco

AU - Gionis, Aristides

AU - Weber, Ingmar

PY - 2013/8/11

Y1 - 2013/8/11

N2 - We present ShoppingAdvisor, a novel recommender system that helps users in shopping for technical products. ShoppingAdvisor leverages both user preferences and technical product attributes in order to generate its suggestions. The system elicits user preferences via a tree-shaped owchart, where each node is a question to the user. At each node, ShoppingAdvisor suggests a ranking of products matching the preferences of the user, and that gets progressively refined along the path from the tree's root to one of its leafs. In this paper we show (i) how to learn the structure of the tree, i.e., which questions to ask at each node, and (ii) how to produce a suitable ranking at each node. First, we adapt the classical top-down strategy for building decision trees in order to find the best user attribute to ask at each node. Differently from decision trees, ShoppingAdvisor partitions the user space rather than the product space. Second, we show how to employ a learning-To-rank approach in order to learn, for each node of the tree, a ranking of products appropriate to the users who reach that node. We experiment with two real-world datasets for cars and cameras, and a synthetic one. We use mean reciprocal rank to evaluate ShoppingAdvisor, and show how the performance increases by more than 50% along the path from root to leaf. We also show how collaborative recommendation algorithms such as k-nearest neighbor benefits from feature selection done by the ShoppingAdvisor tree. Our experiments show that ShoppingAdvisor produces good quality interpretable recommendations, while requiring less input from users and being able to handle the cold-start problem.

AB - We present ShoppingAdvisor, a novel recommender system that helps users in shopping for technical products. ShoppingAdvisor leverages both user preferences and technical product attributes in order to generate its suggestions. The system elicits user preferences via a tree-shaped owchart, where each node is a question to the user. At each node, ShoppingAdvisor suggests a ranking of products matching the preferences of the user, and that gets progressively refined along the path from the tree's root to one of its leafs. In this paper we show (i) how to learn the structure of the tree, i.e., which questions to ask at each node, and (ii) how to produce a suitable ranking at each node. First, we adapt the classical top-down strategy for building decision trees in order to find the best user attribute to ask at each node. Differently from decision trees, ShoppingAdvisor partitions the user space rather than the product space. Second, we show how to employ a learning-To-rank approach in order to learn, for each node of the tree, a ranking of products appropriate to the users who reach that node. We experiment with two real-world datasets for cars and cameras, and a synthetic one. We use mean reciprocal rank to evaluate ShoppingAdvisor, and show how the performance increases by more than 50% along the path from root to leaf. We also show how collaborative recommendation algorithms such as k-nearest neighbor benefits from feature selection done by the ShoppingAdvisor tree. Our experiments show that ShoppingAdvisor produces good quality interpretable recommendations, while requiring less input from users and being able to handle the cold-start problem.

KW - Collaborative content

KW - Learning

KW - Ranking

KW - Recommendation

UR - http://www.scopus.com/inward/record.url?scp=85006080387&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85006080387&partnerID=8YFLogxK

U2 - 10.1145/2487575.2487653

DO - 10.1145/2487575.2487653

M3 - Conference contribution

AN - SCOPUS:85006080387

VL - Part F128815

SP - 203

EP - 211

BT - KDD 2013 - 19th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining

PB - Association for Computing Machinery

ER -