Should we embed? A study on the online performance of utilizing embeddings for real-time job recommendations

Emanuel Lacic, Markus Reiter-Haas, Tomislav Duricic, Valentin Slawicek, Elisabeth Lex

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

Abstract

In this work, we present the fndings of an online study, where we explore the impact of utilizing embeddings to recommend job postings under real-time constraints. On the Austrian job platform Studo Jobs, we evaluate two popular recommendation scenarios: (i) providing similar jobs and, (ii) personalizing the job postings that are shown on the homepage. Our results show that for recommending similar jobs, we achieve the best online performance in terms of Click-Through Rate when we employ embeddings based on the most recent interaction. To personalize the job postings shown on a user's homepage, however, combining embeddings based on the frequency and recency with which a user interacts with job postings results in the best online performance.

Original languageEnglish
Title of host publicationRecSys 2019 - 13th ACM Conference on Recommender Systems
PublisherAssociation of Computing Machinery
Pages496-500
Number of pages5
ISBN (Electronic)9781450362436
DOIs
Publication statusPublished - 10 Sep 2019
Event13th ACM Conference on Recommender Systems, RecSys 2019 - Copenhagen, Denmark
Duration: 16 Sep 201920 Sep 2019

Publication series

NameRecSys 2019 - 13th ACM Conference on Recommender Systems

Conference

Conference13th ACM Conference on Recommender Systems, RecSys 2019
CountryDenmark
CityCopenhagen
Period16/09/1920/09/19

Keywords

  • BLL Equation
  • Frequency
  • Item Embeddings
  • Job Recommendations
  • Online Evaluation
  • Real-time
  • Recency

ASJC Scopus subject areas

  • Control and Systems Engineering
  • Software
  • Information Systems
  • Computer Science Applications

Cite this

Lacic, E., Reiter-Haas, M., Duricic, T., Slawicek, V., & Lex, E. (2019). Should we embed? A study on the online performance of utilizing embeddings for real-time job recommendations. In RecSys 2019 - 13th ACM Conference on Recommender Systems (pp. 496-500). (RecSys 2019 - 13th ACM Conference on Recommender Systems). Association of Computing Machinery. https://doi.org/10.1145/3298689.3346989

Should we embed? A study on the online performance of utilizing embeddings for real-time job recommendations. / Lacic, Emanuel; Reiter-Haas, Markus; Duricic, Tomislav; Slawicek, Valentin; Lex, Elisabeth.

RecSys 2019 - 13th ACM Conference on Recommender Systems. Association of Computing Machinery, 2019. p. 496-500 (RecSys 2019 - 13th ACM Conference on Recommender Systems).

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

Lacic, E, Reiter-Haas, M, Duricic, T, Slawicek, V & Lex, E 2019, Should we embed? A study on the online performance of utilizing embeddings for real-time job recommendations. in RecSys 2019 - 13th ACM Conference on Recommender Systems. RecSys 2019 - 13th ACM Conference on Recommender Systems, Association of Computing Machinery, pp. 496-500, 13th ACM Conference on Recommender Systems, RecSys 2019, Copenhagen, Denmark, 16/09/19. https://doi.org/10.1145/3298689.3346989
Lacic E, Reiter-Haas M, Duricic T, Slawicek V, Lex E. Should we embed? A study on the online performance of utilizing embeddings for real-time job recommendations. In RecSys 2019 - 13th ACM Conference on Recommender Systems. Association of Computing Machinery. 2019. p. 496-500. (RecSys 2019 - 13th ACM Conference on Recommender Systems). https://doi.org/10.1145/3298689.3346989
Lacic, Emanuel ; Reiter-Haas, Markus ; Duricic, Tomislav ; Slawicek, Valentin ; Lex, Elisabeth. / Should we embed? A study on the online performance of utilizing embeddings for real-time job recommendations. RecSys 2019 - 13th ACM Conference on Recommender Systems. Association of Computing Machinery, 2019. pp. 496-500 (RecSys 2019 - 13th ACM Conference on Recommender Systems).
@inproceedings{ca0fa16c7fd04d1eb539d2771d1e9b10,
title = "Should we embed? A study on the online performance of utilizing embeddings for real-time job recommendations",
abstract = "In this work, we present the fndings of an online study, where we explore the impact of utilizing embeddings to recommend job postings under real-time constraints. On the Austrian job platform Studo Jobs, we evaluate two popular recommendation scenarios: (i) providing similar jobs and, (ii) personalizing the job postings that are shown on the homepage. Our results show that for recommending similar jobs, we achieve the best online performance in terms of Click-Through Rate when we employ embeddings based on the most recent interaction. To personalize the job postings shown on a user's homepage, however, combining embeddings based on the frequency and recency with which a user interacts with job postings results in the best online performance.",
keywords = "BLL Equation, Frequency, Item Embeddings, Job Recommendations, Online Evaluation, Real-time, Recency",
author = "Emanuel Lacic and Markus Reiter-Haas and Tomislav Duricic and Valentin Slawicek and Elisabeth Lex",
year = "2019",
month = "9",
day = "10",
doi = "10.1145/3298689.3346989",
language = "English",
series = "RecSys 2019 - 13th ACM Conference on Recommender Systems",
publisher = "Association of Computing Machinery",
pages = "496--500",
booktitle = "RecSys 2019 - 13th ACM Conference on Recommender Systems",
address = "United States",

}

TY - GEN

T1 - Should we embed? A study on the online performance of utilizing embeddings for real-time job recommendations

AU - Lacic, Emanuel

AU - Reiter-Haas, Markus

AU - Duricic, Tomislav

AU - Slawicek, Valentin

AU - Lex, Elisabeth

PY - 2019/9/10

Y1 - 2019/9/10

N2 - In this work, we present the fndings of an online study, where we explore the impact of utilizing embeddings to recommend job postings under real-time constraints. On the Austrian job platform Studo Jobs, we evaluate two popular recommendation scenarios: (i) providing similar jobs and, (ii) personalizing the job postings that are shown on the homepage. Our results show that for recommending similar jobs, we achieve the best online performance in terms of Click-Through Rate when we employ embeddings based on the most recent interaction. To personalize the job postings shown on a user's homepage, however, combining embeddings based on the frequency and recency with which a user interacts with job postings results in the best online performance.

AB - In this work, we present the fndings of an online study, where we explore the impact of utilizing embeddings to recommend job postings under real-time constraints. On the Austrian job platform Studo Jobs, we evaluate two popular recommendation scenarios: (i) providing similar jobs and, (ii) personalizing the job postings that are shown on the homepage. Our results show that for recommending similar jobs, we achieve the best online performance in terms of Click-Through Rate when we employ embeddings based on the most recent interaction. To personalize the job postings shown on a user's homepage, however, combining embeddings based on the frequency and recency with which a user interacts with job postings results in the best online performance.

KW - BLL Equation

KW - Frequency

KW - Item Embeddings

KW - Job Recommendations

KW - Online Evaluation

KW - Real-time

KW - Recency

UR - http://www.scopus.com/inward/record.url?scp=85073364689&partnerID=8YFLogxK

U2 - 10.1145/3298689.3346989

DO - 10.1145/3298689.3346989

M3 - Conference contribution

T3 - RecSys 2019 - 13th ACM Conference on Recommender Systems

SP - 496

EP - 500

BT - RecSys 2019 - 13th ACM Conference on Recommender Systems

PB - Association of Computing Machinery

ER -