一 Datasets

1. ubuntu chat corpus: http://daviduthus.org/UCC/

2. dialogue  datasets (circa 2005)

二 Tasks

task1: next utterance classfication

the dual encoder model

task2: large copurs next-utterance retrievl 

task3: narural language generation

hierachical encoder-decoder architecture(bengio AAAI 2015)

三 Interesting topics 

(Interesting! ) exploring ethical and social  challenge:

bias:

 bias in image classification, debiasing word embeddings(bolukbasi et al 2016), hate and offensive language detection  (Davidson at al 2017)

privacy

safety

value alignment

adversarial examples

 四 Challenges

Challenge: how to evaluation ? 

evaluation for generative models:

perplexity

word error rate

word overlap metrics, BLEU

camparing response with human

 五 adversarial examples