Beggars can't be choosers: Augmenting sparse data for embedding-based product recommendations in retail stores

Matthias Wölbitsch, Simon Walk, Michael Goller, Denis Helic

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

Abstract

Recommender systems are an essential component in many e-commerce platforms to drive sales and guide customers when exploring new products. With the increasing adoption of RFID technology in traditional brick-and-mortar stores, for example, in the form of smart fitting rooms that allow to display recommendations in the integrated mirror, retailers have only recently started to tap into existing product recommendation algorithms. However, due to limited data availability as well as sparsity, for example due to assortments adapted for different demographics, traditional retailers largely struggle to leverage this technology. In this paper we extend the state-of-the-art embedding-based recommender approach prod2vec by processing information about co-purchased products (i.e., shopping baskets) in retail stores. By adding point-of-sale information to shopping baskets we are able to provide recommendations aimed at individual stores, without having to maintain separate models for each location. Furthermore, we experiment with data augmentation methods to overcome the imposed limitations of the available data, and are able to increase the quality of the computed recommendations by more than 6.9%.

Original languageEnglish
Title of host publicationACM UMAP 2019 - Proceedings of the 27th ACM Conference on User Modeling, Adaptation and Personalization
PublisherAssociation of Computing Machinery
Pages104-112
Number of pages9
ISBN (Electronic)9781450360210
DOIs
Publication statusPublished - 7 Jun 2019
Event27th ACM Conference on User Modeling, Adaptation and Personalization, UMAP 2019 - Larnaca, Cyprus
Duration: 9 Jun 201912 Jun 2019

Publication series

NameACM UMAP 2019 - Proceedings of the 27th ACM Conference on User Modeling, Adaptation and Personalization

Conference

Conference27th ACM Conference on User Modeling, Adaptation and Personalization, UMAP 2019
CountryCyprus
CityLarnaca
Period9/06/1912/06/19

Fingerprint

Retail stores
Sales
Recommender systems
Brick
Mortar
Radio frequency identification (RFID)
Mirrors
Availability
Experiments

Keywords

  • Prod2vec
  • Recommender
  • Retail industry
  • Shopping baskets

ASJC Scopus subject areas

  • Software

Cite this

Wölbitsch, M., Walk, S., Goller, M., & Helic, D. (2019). Beggars can't be choosers: Augmenting sparse data for embedding-based product recommendations in retail stores. In ACM UMAP 2019 - Proceedings of the 27th ACM Conference on User Modeling, Adaptation and Personalization (pp. 104-112). (ACM UMAP 2019 - Proceedings of the 27th ACM Conference on User Modeling, Adaptation and Personalization). Association of Computing Machinery. https://doi.org/10.1145/3320435.3320454

Beggars can't be choosers : Augmenting sparse data for embedding-based product recommendations in retail stores. / Wölbitsch, Matthias; Walk, Simon; Goller, Michael; Helic, Denis.

ACM UMAP 2019 - Proceedings of the 27th ACM Conference on User Modeling, Adaptation and Personalization. Association of Computing Machinery, 2019. p. 104-112 (ACM UMAP 2019 - Proceedings of the 27th ACM Conference on User Modeling, Adaptation and Personalization).

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

Wölbitsch, M, Walk, S, Goller, M & Helic, D 2019, Beggars can't be choosers: Augmenting sparse data for embedding-based product recommendations in retail stores. in ACM UMAP 2019 - Proceedings of the 27th ACM Conference on User Modeling, Adaptation and Personalization. ACM UMAP 2019 - Proceedings of the 27th ACM Conference on User Modeling, Adaptation and Personalization, Association of Computing Machinery, pp. 104-112, 27th ACM Conference on User Modeling, Adaptation and Personalization, UMAP 2019, Larnaca, Cyprus, 9/06/19. https://doi.org/10.1145/3320435.3320454
Wölbitsch M, Walk S, Goller M, Helic D. Beggars can't be choosers: Augmenting sparse data for embedding-based product recommendations in retail stores. In ACM UMAP 2019 - Proceedings of the 27th ACM Conference on User Modeling, Adaptation and Personalization. Association of Computing Machinery. 2019. p. 104-112. (ACM UMAP 2019 - Proceedings of the 27th ACM Conference on User Modeling, Adaptation and Personalization). https://doi.org/10.1145/3320435.3320454
Wölbitsch, Matthias ; Walk, Simon ; Goller, Michael ; Helic, Denis. / Beggars can't be choosers : Augmenting sparse data for embedding-based product recommendations in retail stores. ACM UMAP 2019 - Proceedings of the 27th ACM Conference on User Modeling, Adaptation and Personalization. Association of Computing Machinery, 2019. pp. 104-112 (ACM UMAP 2019 - Proceedings of the 27th ACM Conference on User Modeling, Adaptation and Personalization).
@inproceedings{1e8c5dda5bc74544bd69affb56a4adb3,
title = "Beggars can't be choosers: Augmenting sparse data for embedding-based product recommendations in retail stores",
abstract = "Recommender systems are an essential component in many e-commerce platforms to drive sales and guide customers when exploring new products. With the increasing adoption of RFID technology in traditional brick-and-mortar stores, for example, in the form of smart fitting rooms that allow to display recommendations in the integrated mirror, retailers have only recently started to tap into existing product recommendation algorithms. However, due to limited data availability as well as sparsity, for example due to assortments adapted for different demographics, traditional retailers largely struggle to leverage this technology. In this paper we extend the state-of-the-art embedding-based recommender approach prod2vec by processing information about co-purchased products (i.e., shopping baskets) in retail stores. By adding point-of-sale information to shopping baskets we are able to provide recommendations aimed at individual stores, without having to maintain separate models for each location. Furthermore, we experiment with data augmentation methods to overcome the imposed limitations of the available data, and are able to increase the quality of the computed recommendations by more than 6.9{\%}.",
keywords = "Prod2vec, Recommender, Retail industry, Shopping baskets",
author = "Matthias W{\"o}lbitsch and Simon Walk and Michael Goller and Denis Helic",
year = "2019",
month = "6",
day = "7",
doi = "10.1145/3320435.3320454",
language = "English",
series = "ACM UMAP 2019 - Proceedings of the 27th ACM Conference on User Modeling, Adaptation and Personalization",
publisher = "Association of Computing Machinery",
pages = "104--112",
booktitle = "ACM UMAP 2019 - Proceedings of the 27th ACM Conference on User Modeling, Adaptation and Personalization",
address = "United States",

}

TY - GEN

T1 - Beggars can't be choosers

T2 - Augmenting sparse data for embedding-based product recommendations in retail stores

AU - Wölbitsch, Matthias

AU - Walk, Simon

AU - Goller, Michael

AU - Helic, Denis

PY - 2019/6/7

Y1 - 2019/6/7

N2 - Recommender systems are an essential component in many e-commerce platforms to drive sales and guide customers when exploring new products. With the increasing adoption of RFID technology in traditional brick-and-mortar stores, for example, in the form of smart fitting rooms that allow to display recommendations in the integrated mirror, retailers have only recently started to tap into existing product recommendation algorithms. However, due to limited data availability as well as sparsity, for example due to assortments adapted for different demographics, traditional retailers largely struggle to leverage this technology. In this paper we extend the state-of-the-art embedding-based recommender approach prod2vec by processing information about co-purchased products (i.e., shopping baskets) in retail stores. By adding point-of-sale information to shopping baskets we are able to provide recommendations aimed at individual stores, without having to maintain separate models for each location. Furthermore, we experiment with data augmentation methods to overcome the imposed limitations of the available data, and are able to increase the quality of the computed recommendations by more than 6.9%.

AB - Recommender systems are an essential component in many e-commerce platforms to drive sales and guide customers when exploring new products. With the increasing adoption of RFID technology in traditional brick-and-mortar stores, for example, in the form of smart fitting rooms that allow to display recommendations in the integrated mirror, retailers have only recently started to tap into existing product recommendation algorithms. However, due to limited data availability as well as sparsity, for example due to assortments adapted for different demographics, traditional retailers largely struggle to leverage this technology. In this paper we extend the state-of-the-art embedding-based recommender approach prod2vec by processing information about co-purchased products (i.e., shopping baskets) in retail stores. By adding point-of-sale information to shopping baskets we are able to provide recommendations aimed at individual stores, without having to maintain separate models for each location. Furthermore, we experiment with data augmentation methods to overcome the imposed limitations of the available data, and are able to increase the quality of the computed recommendations by more than 6.9%.

KW - Prod2vec

KW - Recommender

KW - Retail industry

KW - Shopping baskets

UR - http://www.scopus.com/inward/record.url?scp=85068055413&partnerID=8YFLogxK

U2 - 10.1145/3320435.3320454

DO - 10.1145/3320435.3320454

M3 - Conference contribution

T3 - ACM UMAP 2019 - Proceedings of the 27th ACM Conference on User Modeling, Adaptation and Personalization

SP - 104

EP - 112

BT - ACM UMAP 2019 - Proceedings of the 27th ACM Conference on User Modeling, Adaptation and Personalization

PB - Association of Computing Machinery

ER -