Di Bratto, Martina (2024) Journey through Argumentation-based Dialogue: from theoretical to computational models for the implementation of Conversational Recommender Systems. [Tesi di dottorato]
Anteprima |
Testo
Di Bratto_Martina_36.pdf Download (4MB) | Anteprima |
| Tipologia del documento: | Tesi di dottorato |
|---|---|
| Lingua: | English |
| Titolo: | Journey through Argumentation-based Dialogue: from theoretical to computational models for the implementation of Conversational Recommender Systems |
| Autori: | Autore Email Di Bratto, Martina martina.dibratto@unina.it |
| Data: | 2024 |
| Numero di pagine: | 194 |
| Istituzione: | Università degli Studi di Napoli Federico II |
| Dipartimento: | Studi Umanistici |
| Dottorato: | Mind, gender and languages |
| Ciclo di dottorato: | 36 |
| Coordinatore del Corso di dottorato: | nome email Bacchini, Dario dario.bacchini@unina.it |
| Tutor: | nome email Cutugno, Francesco [non definito] |
| Data: | 2024 |
| Numero di pagine: | 194 |
| Parole chiave: | Argumentation Theory, Conversational Recommender Systems, pragmatics |
| Settori scientifico-disciplinari del MIUR: | Area 10 - Scienze dell'antichità, filologico-letterarie e storico-artistiche > L-LIN/01 - Glottologia e linguistica |
| Depositato il: | 10 Ott 2024 14:44 |
| Ultima modifica: | 12 Ago 2026 05:36 |
| URI: | https://www.fedoa.unina.it/id/eprint/15597 |
Abstract
This thesis deals with the application of theories concerning argumentation to support the building of a human-machine dialogue system involved in recom�mendation tasks. In particular, the work aims at the formalisation of a theoretical model for Argumentation-based Dialogue (ABD). According to Prakken [2018b], while Argumentation-based Inference has been long studied over the decades, ABD still lacks a solid reference framework. Starting with a literature review, a theo�retical framework for relevant argument selection has been selected. The frame�work focuses on the cognitive properties an argument should have to be considered plausible. Then, these cognitive features have been mapped onto mathematical measures, and the resulting model has been evaluated through its computational implementation in two steps. The first one was dedicated to the model validation through the generation of simulated dialogues and their human evaluation in terms of consistency, naturalness and quality of argument selection. Results show that the model-based simulated dialogues are recognised by the participants and well evalu�ated with respect the randomic simulated dialogues. The second step consisted of a real human-machine interaction, where the overall architecture has been tested and finally successfully evaluated by 10 users through a user experience questionnaire. The work concludes with a last experiment where the pragmatic value of asking questions using different forms has been analysed. More specifically, the study fo�cused on the informativeness of the elicited answer and the related cognitive load for the user. The study highlights how different pragmatic features can affect the usefulness of the answer for the resolution of the decision problem. In conclusion, the thesis presents a novel methodological approach to implement and evaluate a theoretical model through its computational realisation.
Downloads
Downloads per month over past year
Actions (login required)
 |
Modifica documento |
