Get our free extension to see links to code for papers anywhere online!

Chrome logo Add to Chrome

Firefox logo Add to Firefox


Focus-Constrained Attention Mechanism for CVAE-based Response Generation

Sep 25, 2020
Zhi Cui, Yanran Li, Jiayi Zhang, Jianwei Cui, Chen Wei, Bin Wang



To model diverse responses for a given post, one promising way is to introduce a latent variable into Seq2Seq models. The latent variable is supposed to capture the discourse-level information and encourage the informativeness of target responses. However, such discourse-level information is often too coarse for the decoder to be utilized. To tackle it, our idea is to transform the coarse-grained discourse-level information into fine-grained word-level information. Specifically, we firstly measure the semantic concentration of corresponding target response on the post words by introducing a fine-grained focus signal. Then, we propose a focus-constrained attention mechanism to take full advantage of focus in well aligning the input to the target response. The experimental results demonstrate that by exploiting the fine-grained signal, our model can generate more diverse and informative responses compared with several state-of-the-art models.

* To appear in findings of EMNLP 2020 


Share this with someone who'll enjoy it:

   Access Paper Source



Share this with someone who'll enjoy it: