Some questions about the Dataset about is13 HOT 5 CLOSED

For anyone looking as I was, you can find the intent labels for atis here (not sure if they are different to what you're talking about though):

https://github.com/hongyunnchen/JointSLU/tree/master/data

from is13.

1. set[1] corresponds to Name entities. so there is a lot of zeros. check the paper to see what Name Entity means.
   http://www.iro.umontreal.ca/%7Elisa/pointeurs/taslp_RNNSLU_final_doubleColumn.pdf
2. set[2] corresponds to labels and set[0] corresponds to word indexes
3. dicts[1] is the mapping to the Name Entity
4. inverting a dictionary can bed solved using a Google search. http://stackoverflow.com/questions/483666/python-reverse-inverse-a-mapping

from is13.

Hi Gregoire,

Do you happen to have the "intent" labels for the ATIS examples available somewhere? I'm interested in the joint utterance classification / slot-filling application. Thanks!

[I realize this isn't an "issue" per se, but it seemed the most appropriate place to ask. Also thanks for posting this code -- it's a great learning tool.]

from is13.

@cbcase Hi Carl, unfortunately Microsoft only let me release their version of the Slot-Filling dataset.

from is13.

Alright, thanks for the quick reply!

from is13.