search

elpaco-escience / scikit-talk

Scikit-talk is an open-source toolkit for processing collections of real-world conversational speech in Python. The toolkit aims to facilitate the exploration of large collections of transcriptions and annotations of conversational interaction.
Apache License 2.0
2 stars 0 forks source link

Document turn dynamics calculations, most importantly FTO #67

Open mdingemanse opened 2 months ago

mdingemanse commented 2 months ago

We have a fairly well-considered form of FTO calculations that is a substantial python improvement of a python port of my original R code. The code is in conversation.py.

However, we don't have good documentation yet of how FTO is calculated and why.

Here is a lot of useful material, including a great flowchart by @bvreede.

mdingemanse commented 2 months ago

BTW this is especially important because choices made early (e.g. which tiers to import) directly impact FTO calculations. E.g. if you import an IFADV file and it has tiers for speech and for gaze, scikit-talk won't know which is which and FTO calculations will simply not be carried out because it's detected as four participants and we (sensibly) only compute FTO for clearly dyadic data.

A vignette or workflow that explains this would be good.

Also, flexible ways to recalculate FTO using custom settings, which I'll make another issue for.