asappresearch/dialog-intent-induction

Code and data for paper "Dialog Intent Induction with Deep Multi-View Clustering", Hugh Perkins and Yi Yang, 2019, EMNLP 2019

dialog-intent-induction's Issues

About ask_ubuntu dataset

Hi,

Sorry it's me again 😝. Two questions regarding the ask_ubuntu dataset,

1. The label column are all UNKs, is this expected?
2. I checked the original ubuntu dataset here https://drive.google.com/drive/u/0/folders/0B-btHzfJjPnobXZ0MndjSkxkRkk. Are you using both askubuntu_2014_posts.xml.gz and askubuntu_2015_posts.xml.gz (maybe merge them) as input to the preprocessing code?

Thanks!

duplicate entries in airlines_processed.csv

Hi,

I am trying to run the data construction code and I notice something weird. Specifically, in the file airlines_processed.csv, there are about 700 entries with duplicate tweet_id. To name a few, 638092, 827255, 394166...etc. Is this expected?

Thanks.

context of the data

Hi,

Thanks for sharing the code! One question, is it possible to separate the context into different utterances? Since currently the context is constructed by merging user and system utterances if I understand correctly.

Thanks.