Open mathematicalmichael opened 4 years ago
Metadata should be rich enough to do ML with it later on. Can help automate decisions.
Store as a separate file.
Currently probabilities are supplied relative to the dictionaries.
Would be nice to ALSO just show probabilities by summing up counts in total, so that they actually sum to 1.
This would help making decisions much faster
Can maybe even sort by probabilities, but I prefer the Known/Unknown breakdown. Maybe for a visual assist we add an asterix to the highest? Perhaps only to those that are above 1/N where N is total candidates. More asterixes for higher probabilities? Like, a visual bar chart in command line?
also be able to load in unknown dictionary with frequency and create a known from it.