grev

The README opens with

Unix filters that ask questions instead of matching patterns.

The tools run on TypeSafe's Jev models, which answer typed questions with calibrated probabilities instead of generating text. That's why these tools behave like grep, sort or cut: output is always your input. Labels and scores appear only as explicit columns. A few thousand lines cost fractions of a cent and take seconds.

tool like does --------- grev grep print the records the model says yes to isv test answer a yes/no question about the whole input with the exit status oneof case print which label fits the whole input tagv awk label every record; --split DIR routes records into files sortv sort sort in an order described in words, by pairwise comparison rank sort sort records by...