Abstract
Sequence-to-Sequence (Seq2Seq) models have witnessed a notable success in generating natural conversational exchanges. Notwithstanding the syntactically well-formed responses generated by these neural network models, they are prone to be acontextual, short and generic. In this work, we introduce a Topical Hierarchical Recurrent Encoder Decoder (THRED), a novel, fully data-driven, multi-turn response generation system intended to produce contextual and topic-aware responses. Our model is built upon the basic Seq2Seq model by augmenting it with a hierarchical joint attention mechanism that incorporates topical concepts and previous interactions into the response generation. To train our model, we provide a clean and high-quality conversational dataset mined from Reddit comments. We evaluate THRED on two novel automated metrics, dubbed Semantic Similarity and Response Echo Index, as well as with human evaluation. Our experiments demonstrate that the proposed model is able to generate more diverse and contextually relevant responses compared to the strong baselines.- Anthology ID:
- W19-4103
- Volume:
- Proceedings of the First Workshop on NLP for Conversational AI
- Month:
- August
- Year:
- 2019
- Address:
- Florence, Italy
- Editors:
- Yun-Nung Chen, Tania Bedrax-Weiss, Dilek Hakkani-Tur, Anuj Kumar, Mike Lewis, Thang-Minh Luong, Pei-Hao Su, Tsung-Hsien Wen
- Venue:
- ACL
- SIG:
- Publisher:
- Association for Computational Linguistics
- Note:
- Pages:
- 18–31
- Language:
- URL:
- https://aclanthology.org/W19-4103
- DOI:
- 10.18653/v1/W19-4103
- Cite (ACL):
- Nouha Dziri, Ehsan Kamalloo, Kory Mathewson, and Osmar Zaiane. 2019. Augmenting Neural Response Generation with Context-Aware Topical Attention. In Proceedings of the First Workshop on NLP for Conversational AI, pages 18–31, Florence, Italy. Association for Computational Linguistics.
- Cite (Informal):
- Augmenting Neural Response Generation with Context-Aware Topical Attention (Dziri et al., ACL 2019)
- PDF:
- https://preview.aclanthology.org/ingest-acl-2023-videos/W19-4103.pdf
- Code
- nouhadziri/THRED
- Data
- Reddit Conversation Corpus