Do we want to do any additional filtering of text sectors to remove noisy ones?
Devise strategy to name sectors.
In the previous version of the taxonomy we used tfidf at the text sector level. We could alternatively do this inside SIC4s i.e. comparing a text sector only to those generated by the same SIC4.
We could also check alternative strategies e.g. use GPT3?
Do we want to do any additional filtering of text sectors to remove noisy ones?
Devise strategy to name sectors.
In the previous version of the taxonomy we used
tfidfat the text sector level. We could alternatively do thisinside SIC4si.e. comparing a text sector only to those generated by the same SIC4.We could also check alternative strategies e.g. use GPT3?