search

SpectData / MONAH

Multi-Modal Narratives for Humans
MIT License
3 stars 0 forks source link

Session level Inputs #7

Closed joshkyh closed 3 years ago

joshkyh commented 3 years ago

Marriane completed all coarse level information that she can complete. However, wee are pausing development on the harder ones because.

(1) We aim to also introduce the ability for user to select modules just like fine weave_session_transcript()

(2) We want to test (1) before proceeding.

mfmakahiya commented 3 years ago

Hi Joshua! I'm not sure if it's possible to test the session-level transcripts since in the original iemocap sessions, we made use of mu and standard deviation according to the other partitions, but our case here is just one session.

mfmakahiya commented 3 years ago

Oh I think we didn't do Narrative_Coarse in iemocap sessions, so I have nothing to compare the results to.

joshkyh commented 3 years ago

Hi Marriane,

Regarding the first comment, we need to use SQL to find out the mu and sd of the train partition and hard-code that into the configurations file in python.

As for Narrative_Coarse, the last time we done a coarse narrative is in EQClinic, not iemocap. Could you have a look in that schema (dbo) please? :)

Thank you.

Best Regards, Joshua

joshkyh commented 3 years ago

Coarse In Progress: Head Nodding, Forward Trunk Leaning, Sentiment (EMPATH)

Empath script: https://github.com/joshkyh/empath/blob/master/python_scripts/Data_Preprocessing/FeatureEngineering/session-level%20empath.py

We do EMPATH for session level, in other words, it is not available for fine.

joshkyh commented 3 years ago

Not doing Coarse. Close.