During a diagnosis, the AI uses data collected from a machine or from the user, and suggests the best step. A trained AI can learn to do this efficiently, but in the beginning it won't know very much.
Just like you don't hire a new tech without providing training, you can help the AI learn with some hints.
A dependency just says that a step depends on some condition. Since you don't want to ask for observations or suggest solutions that make no sense, dependencies can narrow down where they are applicable. In addition, sometimes an expert knows one step should always come before another. You don't want the AI to learn that by trial and error, you can simply tell it the rule.
Say a step depends on error_code=MF_3214, that step will not be suggested unless that is true.
Updated over 2 years ago