Engagingness in Open-Domain Dialogue Systems: A Systematic Survey of Datasets, Metrics and Methods
AI, Computer Science and Robotics Technology · 09.04.2026 · 5
DOI: 10.5772/acrt.20250081 ↗Kurz gefasst
Fachartikel in der Zeitschrift AI, Computer Science and Robotics Technology, online erschienen am 09.04.2026. Verfasst von Ekata Mitra und 2 weiteren Autor:innen. Die Arbeit nennt 141 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
As conversational artificial intelligence becomes more deeply integrated into our daily lives, understanding what makes a dialogue engaging is essential for building effective open-domain dialogue (ODD) systems. The concept of engagingness remains inherently subjective and is therefore difficult to formalize and operationalize. This survey provides a systematic overview of datasets, evaluation metrics, and methods developed to measure and improve engagingness in ODD systems. We analyze existing datasets annotated for engagement, including their annotation protocols and coverage of languages. In this survey, we emphasize the dynamic evolution of dialogue datasets, which is continuously enriched with human-annotated labels for an additional parameter relevant to conversational quality. Our primary focus is to help future research on the creation of a thoroughly curated dataset, ensuring both purity and reliability for non-tasked dialogue system. We then review evaluation methodologies and discuss recent methods for improving engagingness in dialogue systems. The survey concludes by identifying critical challenges and open research directions, including the need for multilingual benchmarks, multimodal engagement modeling, and hybrid evaluation frameworks.
Zitieren
APA
Mitra, E., Ferron, A., & Agrawal, A. (2026). Engagingness in Open-Domain Dialogue Systems: A Systematic Survey of Datasets, Metrics and Methods. AI, Computer Science and Robotics Technology, 5. https://doi.org/10.5772/acrt.20250081
Vancouver
Mitra E, Ferron A, Agrawal A. Engagingness in Open-Domain Dialogue Systems: A Systematic Survey of Datasets, Metrics and Methods. AI, Computer Science and Robotics Technology. 2026;5. doi:10.5772/acrt.20250081
BibTeX
@article{Mitra2026,
author = {Mitra, Ekata and Ferron, Amila and Agrawal, Ameeeta},
title = {Engagingness in Open-Domain Dialogue Systems: A Systematic Survey of Datasets, Metrics and Methods},
journal = {AI, Computer Science and Robotics Technology},
year = {2026},
volume = {5},
doi = {10.5772/acrt.20250081}
}
EUVIMED