Cage: A Hybrid Framework for Closed-Domain Conversational Agents

Abstract

Current conversational agents are primarily designed to answer user queries based on structured pre-defined utterance-response pairs. While question-answering (QA) systems extracts potential answers, to queries, from unstructured texts. However, in domain-specific settings, manual creation of query-response pairs is expensive, and domain adaptation of QA platforms is crucial. To this end, we propose Cage , a “hybrid” conversational framework seamlessly integrating structured and unstructured data to obtain precise answers for user queries – improving user experience and quality-of-service . We describe the different components combining query matching and extractive question answering , and demonstrate the multi-lingual chatbot interface provided to a user.

Cite

Text

Burgin et al. "Cage: A Hybrid Framework for Closed-Domain Conversational Agents." European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases, 2022. doi:10.1007/978-3-031-26422-1_46

Markdown

[Burgin et al. "Cage: A Hybrid Framework for Closed-Domain Conversational Agents." European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases, 2022.](https://mlanthology.org/ecmlpkdd/2022/burgin2022ecmlpkdd-cage/) doi:10.1007/978-3-031-26422-1_46

BibTeX

@inproceedings{burgin2022ecmlpkdd-cage,
  title     = {{Cage: A Hybrid Framework for Closed-Domain Conversational Agents}},
  author    = {Burgin, Edward and Dutta, Sourav and Assem, Haytham and Patel, Raj Nath},
  booktitle = {European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases},
  year      = {2022},
  pages     = {636-640},
  doi       = {10.1007/978-3-031-26422-1_46},
  url       = {https://mlanthology.org/ecmlpkdd/2022/burgin2022ecmlpkdd-cage/}
}