How are the matches data in dictionary.txt created?

[ningyuv]

I noticed in the readme:

dictionary.txt is derived from data from Unihan and CJKlib.

I tried my best to find these data but could only find cdb-strokeorder-rules.

so I want to know how matches data created? Is it by rules or manually?
I want to update and correct some of the content to adapt to the latest Unihan Database.