Conversational agents are increasingly present in the context of Industry 4.0, in particular for customer care applications. To be really useful, these agents must correctly recognize the users' intents in order to provide them with adequate and satisfactory answers. In this paper, we introduce an intent recognition approach that leverages Google's Bidirectional Encoder Representations from Transformers (BERT), a language representation model based on deep neural networks. The results of a comparative analysis performed on three publicly available datasets, including Kaggle, Alexa, and Converse datasets, show that the proposed approach can outperform other state-of-the-art models based on different techniques.

Roma, F., Sansonetti, G., D'Aniello, G., Micarelli, A. (2023). A BERT-Based Approach to Intent Recognition. In Proceedings of IEEE EUROCON 2023 - 20th International Conference on Smart Technologies (pp.568-572). IEEE [10.1109/EUROCON56442.2023.10198959].

A BERT-Based Approach to Intent Recognition

Roma, Federico;Sansonetti, Giuseppe
;
Micarelli, Alessandro
2023-01-01

Abstract

Conversational agents are increasingly present in the context of Industry 4.0, in particular for customer care applications. To be really useful, these agents must correctly recognize the users' intents in order to provide them with adequate and satisfactory answers. In this paper, we introduce an intent recognition approach that leverages Google's Bidirectional Encoder Representations from Transformers (BERT), a language representation model based on deep neural networks. The results of a comparative analysis performed on three publicly available datasets, including Kaggle, Alexa, and Converse datasets, show that the proposed approach can outperform other state-of-the-art models based on different techniques.
2023
978-1-6654-6397-3
Roma, F., Sansonetti, G., D'Aniello, G., Micarelli, A. (2023). A BERT-Based Approach to Intent Recognition. In Proceedings of IEEE EUROCON 2023 - 20th International Conference on Smart Technologies (pp.568-572). IEEE [10.1109/EUROCON56442.2023.10198959].
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11590/449947
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 0
  • ???jsp.display-item.citation.isi??? ND
social impact