|
I'm not sure what you mean. All automated content editing is difficult to do well, and religious words are the hardest because they have "good" and "bad" meanings.
The filter has two modes for the word "God", and the modes are guessed at by the general vulgarity of the entire book. When a vulgar book is detected, all instances of "God" are replaced assuming no one is going to use the word well. Otherwise a more subtle scheme is used.
It's all very easy to change if you don't like the behavior. Instructions are included on the github page.
|