@inproceedings{ma-etal-2019-triplenet,
    title = "{T}riple{N}et: Triple Attention Network for Multi-Turn Response Selection in Retrieval-Based Chatbots",
    author = "Ma, Wentao  and
      Cui, Yiming  and
      Shao, Nan  and
      He, Su  and
      Zhang, Wei-Nan  and
      Liu, Ting  and
      Wang, Shijin  and
      Hu, Guoping",
    booktitle = "Proceedings of the 23rd Conference on Computational Natural Language Learning (CoNLL)",
    month = nov,
    year = "2019",
    address = "Hong Kong, China",
    publisher = "Association for Computational Linguistics",
    url = "https://www.aclweb.org/anthology/K19-1069",
    pages = "737--746",
    abstract = "We consider the importance of different utterances in the context for selecting the response usually depends on the current query. In this paper, we propose the model TripleNet to fully model the task with the triple {\textless}context, query, response{\textgreater} instead of {\textless}context, response {\textgreater} in previous works. The heart of TripleNet is a novel attention mechanism named triple attention to model the relationships within the triple at four levels. The new mechanism updates the representation of each element based on the attention with the other two concurrently and symmetrically.We match the triple {\textless}C, Q, R{\textgreater} centered on the response from char to context level for prediction.Experimental results on two large-scale multi-turn response selection datasets show that the proposed model can significantly outperform the state-of-the-art methods.",
}