Yes that’s my bot! Thanks
Actually my approach is similar to that but kind of the other way around.
The inputs are using templates (templates maybe aren’t the right word) as they are basically arranged by combinations of words based on the types of words or literals. Based on the template found the pronouns are translated (or implied pronouns that aren’t in the sentence). So if the “he” appears in the sentence it’s mapped to the last “he” in the conversation, but if somebody asks “how tall?” for example the “he” pronoun is implied and not in the question so it has to be added at the template level.
The output is actually done more like regular expressions. With expression types nested within others so like there can be many type of responses arranged by “type” or response like ways to say hello or goodbye but can be combined with other if it’s appropriate.
Hopefully that’s not confusing! Thanks for asking.
Sorry for the delay in my reply. Been real busy this week