Basic AIML cleaning #31
Conversation
…tool. Probably essentially just a demo/toy.
…le data. Also some overlapping knowledge-of-self. We can extract just a little, respectful portion and place it into another, more central file: ai.aiml.
…ment), there is no category content whatsoever.
… pattern was the only real entry point, which was not found anywhere else in remaining aiml files.
…k when the aiml processor is sort of subordinate to another processor. Additionally, it has some typed-text oriented assumptions.
|
After all that effort, a classic case of "right under my nose". The very aiml library utilized by the skill has in its repository a set of "standard" aiml files, which does appear to be of the same general lineage, but a bit more recent and cleaned up. 23,000 categories or so, vs 50,000 or so in the current list, and nothing about pandorabots: I think this is the clear best way forward, so I am going to create a new PR that entirely replaces existing aiml set with the "standard" and then later on we can cherry pick the clever stuff (that will actually work in a mycroft context) from https://github.com/pandorabots/Free-AIML (which appears to be the origin of the current set). |
For now, just removing more stuff that is too-specific to original sources (pandorabots, mitsuku, etc) or is cannot be triggered. Just a wee little bit of Mycroft-ification, but only what is peripheral and obvious. Later PRs will address direct knowledge of Mycroft by Mycroft and consolidation and reorganization.
This would further address #9