Semantic path-based learning for review volume prediction

Ujjwal Sharma, Stevan Rudinac, Marcel Worring, Joris Demmers, Willemijn van Dolen

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

Abstract

Graphs offer a natural abstraction for modeling complex real-world systems where entities are represented as nodes and edges encode relations between them. In such networks, entities may share common or similar attributes and may be connected by paths through multiple attribute modalities. In this work, we present an approach that uses semantically meaningful, bimodal random walks on real-world heterogeneous networks to extract correlations between nodes and bring together nodes with shared or similar attributes. An attention-based mechanism is used to combine multiple attribute-specific representations in a late fusion setup. We focus on a real-world network formed by restaurants and their shared attributes and evaluate performance on predicting the number of reviews a restaurant receives, a strong proxy for popularity. Our results demonstrate the rich expressiveness of such representations in predicting review volume and the ability of an attention-based model to selectively combine individual representations for maximum predictive power on the chosen downstream task.

Original languageEnglish
Title of host publicationAdvances in Information Retrieval
Subtitle of host publication42nd European Conference on IR Research, ECIR 2020, Proceedings, Part 1
EditorsJoemon M. Jose, Emine Yilmaz, João Magalhães, Pablo Castells, Nicola Ferro, Mário J. Silva, Flávio Martins
Place of PublicationCham Switzerland
PublisherSpringer
Pages821-835
Number of pages15
Edition1st
ISBN (Electronic)9783030454395
ISBN (Print)9783030454388
DOIs
Publication statusPublished - 2020
EventEuropean Conference on Information Retrieval 2020 - Lisbon, Portugal
Duration: 14 Apr 202017 Apr 2020
Conference number: 42nd

Publication series

NameLecture Notes in Computer Science
PublisherSpringer
Volume12035 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

ConferenceEuropean Conference on Information Retrieval 2020
Abbreviated titleECIR 2020
CountryPortugal
CityLisbon
Period14/04/2017/04/20

Keywords

  • Deep learning on graphs
  • Heterogeneous information networks
  • Metapaths
  • Venue popularity prediction

Cite this