Event Recommendation Through Language-Specific User Behaviour in Clickstreams

verfasst von

Sara Abdollahi, Elena Demidova, Simon Gottschalk

Abstract

The relevance and perception of events with global and local impact, such as national elections and terrorist attacks, can vary significantly among different language communities. This chapter discusses recent user access models for event-centric multilingual information, focusing on assisting users, including social scientists and digital humanities researchers, who analyse such events and their impacts. These models aim to facilitate information exploration by emphasising cultural and linguistic differences, a dimension often overlooked by existing entity recommendation methods. Developing recommendation models supporting cross-lingual and cross-cultural analysis of event-related information is particularly challenging due to language barriers and the lack of established datasets. To address these challenges, our prior work involved the creation of the EventKG+Click dataset, which contains event-centric user interaction traces extracted from the EventKG knowledge graph and Wikipedia clickstream data. Additionally, we intro-duced LaSER-a language-specific event recommendation model that considers the user's linguistic and cultural preferences. To improve recommendations, LaSER in-corporates language-specific click data from EventKG+Click. Furthermore, LaSER integrates language-specific embeddings of entities and events, along with their spatio-temporal features, into a learning-to-rank model. This chapter provides an overview of these methods, datasets and evaluation results.

Organisationseinheit(en)

Forschungszentrum L3S

Externe Organisation(en)

Rheinische Friedrich-Wilhelms-Universität Bonn
Lamarr-Institut für Maschinelles Lernen und Künstliche Intelligenz

Typ

Beitrag in Buch/Sammelwerk

Seiten

149-168

Anzahl der Seiten

20

Publikationsdatum

2025

Publikationsstatus

Veröffentlicht

Peer-reviewed

Ja

ASJC Scopus Sachgebiete

Allgemeine Computerwissenschaft, Allgemeine Sozialwissenschaften

Ziele für nachhaltige Entwicklung

SDG 16 – Frieden, Gerechtigkeit und starke Institutionen

Elektronische Version(en)

https://doi.org/10.1007/978-3-031-64451-1_8 (Zugang: Offen)