TU Darmstadt / ULB / TUbiblio

Contextual Models for Sequential Recommendation

Tavakol, Maryam (2019)
Contextual Models for Sequential Recommendation.
Technische Universität Darmstadt
Ph.D. Thesis, Primary publication

Abstract

Recommender systems aim to capture the interests of users in order to provide them with tailored recommendations for items or services they might like. User interests are often unique and depend on many unobservable factors including internal moods or external events. This phenomenon creates a broad range of tasks for recommendation systems that are difficult to address altogether. Nevertheless, analyzing the historical activities of users sheds light on the characteristic traits of individual behaviors in order to enable qualified recommendations.

In this thesis, we deal with the problem of comprehending the interests of users, searching for pertinent items, and ranking them to recommend the most relevant items to the users given different contexts and situations. We focus on recommendation problems in sequential scenarios, where a series of past events influences the future decisions of users. These events are either the developed preferences of users over a long span of time or highly influenced by the zeitgeist and common trends. We are among the first to model recommendation systems in a sequential fashion via exploiting the short-term interests of users in session-based scenarios.

We leverage reinforcement learning techniques to capture underlying short- and long-term user interests in the absence of explicit feedback and develop novel contextual approaches for sequential recommendation systems. These approaches are designed to efficiently learn models for different types of recommendation tasks and are extended to continuous and multi-agent settings. All the proposed methods are empirically studied on large-scale real-world scenarios ranging from e-commerce to sport and demonstrate excellent performance in comparison to baseline approaches.

Item Type: Ph.D. Thesis
Erschienen: 2019
Creators: Tavakol, Maryam
Type of entry: Primary publication
Title: Contextual Models for Sequential Recommendation
Language: English
Referees: Fürnkranz, Prof. Dr. Jo­hannes ; Brefeld, Prof. Dr. Ulf
Date: 29 April 2019
Place of Publication: Darmstadt
Refereed: 25 April 2019
URL / URN: https://tuprints.ulb.tu-darmstadt.de/8667
Abstract:

Recommender systems aim to capture the interests of users in order to provide them with tailored recommendations for items or services they might like. User interests are often unique and depend on many unobservable factors including internal moods or external events. This phenomenon creates a broad range of tasks for recommendation systems that are difficult to address altogether. Nevertheless, analyzing the historical activities of users sheds light on the characteristic traits of individual behaviors in order to enable qualified recommendations.

In this thesis, we deal with the problem of comprehending the interests of users, searching for pertinent items, and ranking them to recommend the most relevant items to the users given different contexts and situations. We focus on recommendation problems in sequential scenarios, where a series of past events influences the future decisions of users. These events are either the developed preferences of users over a long span of time or highly influenced by the zeitgeist and common trends. We are among the first to model recommendation systems in a sequential fashion via exploiting the short-term interests of users in session-based scenarios.

We leverage reinforcement learning techniques to capture underlying short- and long-term user interests in the absence of explicit feedback and develop novel contextual approaches for sequential recommendation systems. These approaches are designed to efficiently learn models for different types of recommendation tasks and are extended to continuous and multi-agent settings. All the proposed methods are empirically studied on large-scale real-world scenarios ranging from e-commerce to sport and demonstrate excellent performance in comparison to baseline approaches.

Alternative Abstract:
Alternative abstract Language

Empfehlungssysteme zielen darauf ab, die Interessen der Benutzer zu erfassen und mass geschneiderte Empfehlungen zu geben. Die Interessen der Benutzer sind oft einzigartig und hängen von vielen unbeobachtbaren Faktoren ab, z.B. Stimmungen oder externen Ereignissen. Dieses Phänomen schafft ein breites Aufgabenspektrum für Empfehlungssysteme. Alle Aufgabe zusammen sind schwer zu lösen. Durch die Analyse historischer Aktivitäten der Benutzer werden die charakteristischen Merkmale einzelner Verhaltensweisen gefunden, um qualifizierte Empfehlungen zu ermöglichen.

In dieser Arbeit beschäftigen wir uns mit dem Problem, diese Interessen zu verstehen, nach sachdienlich Empfehlungen zu suchen und sie zu ordnen, um den Benutzern die Relevantesten, in verschiedenen Kontexten und Situationen, zu empfehlen. Wir konzentrieren uns auf Empfehlungsprobleme in sequentiellen Szenarien, in denen der Verlauf vergangener Ereignisse die zukünftigen Entscheidungen der Benutzer beeinflusst. Diese Ereignisse sind entweder die entwickelten Vorlieben der Nutzer über einen längeren Zeitraum oder stark vom Zeitgeist und den gängigen Trends beeinflusst. Wir geh\"oren zu den Ersten, die Empfehlungssysteme sequentiell und unter Ausnutzung der kurzfristigen Umstände modellieren.

Wir nutzen Bestärkendes Lernen, um die zugrunde liegenden kurz- und langfristigen Benutzerinteressen ohne explizites Feedback zu erfassen und entwickeln Ansätze für sequentielle Empfehlungssysteme. Diese Ansätze sind darauf ausgelegt, Modelle für verschiedene Arten von Empfehlungsaufgaben effizient zu erlernen, und werden auf Stetig und multi-Agenten Probleme erweitert. Alle vorgeschlagenen Methoden werden empirisch von E-Commerce bis zu Sport untersucht und zeigen im Vergleich zu Baseline-Ansätzen eine hervorragende Leistung.

German
URN: urn:nbn:de:tuda-tuprints-86671
Classification DDC: 000 Generalities, computers, information > 004 Computer science
Divisions: 20 Department of Computer Science
20 Department of Computer Science > Knowl­edge En­gi­neer­ing
20 Department of Computer Science > Knowledge Mining and Assessment
Date Deposited: 09 Jun 2019 19:55
Last Modified: 09 Jun 2019 19:55
PPN:
Referees: Fürnkranz, Prof. Dr. Jo­hannes ; Brefeld, Prof. Dr. Ulf
Refereed / Verteidigung / mdl. Prüfung: 25 April 2019
Export:
Suche nach Titel in: TUfind oder in Google
Send an inquiry Send an inquiry

Options (only for editors)
Show editorial Details Show editorial Details