I designed my own expert system language with built in tools for pool water chemistry calculations and then wrote a run time for it in C++. It is much easier to write the rules if the language is customized to make that process as natural as possible.
Don't even worry about the language at first. Start writing down rules in english, or any simpler language that seems natural to you. They can always be converted into whatever language you end up using easily enough. After you write a few hundred rules you can go back and develop a more formal system for how they will really work in a real system. By that point you should understand what kind of things the rules need to be able to do and what the simple ways of saying those things are. That knowledge will allow you to optimize the system to make writing rules easy. There are going to be many rules and they need to be as simple to write and to understand as possible. You need to understand the problem space before the simplest way to structure the rules will become obvious.
I started with a big dialog with lots of input fields, such as the one Isaac-1 described. However, right before I gave up I decided it was going to need to be more interactive that that, asking clarifying questions as it went along, if it was going to be really useful. As GreatCanadian mentioned, the big dialog is too intimidating because there are so many questions that need to be asked at least occasionally.
Don't even worry about the language at first. Start writing down rules in english, or any simpler language that seems natural to you. They can always be converted into whatever language you end up using easily enough. After you write a few hundred rules you can go back and develop a more formal system for how they will really work in a real system. By that point you should understand what kind of things the rules need to be able to do and what the simple ways of saying those things are. That knowledge will allow you to optimize the system to make writing rules easy. There are going to be many rules and they need to be as simple to write and to understand as possible. You need to understand the problem space before the simplest way to structure the rules will become obvious.
I started with a big dialog with lots of input fields, such as the one Isaac-1 described. However, right before I gave up I decided it was going to need to be more interactive that that, asking clarifying questions as it went along, if it was going to be really useful. As GreatCanadian mentioned, the big dialog is too intimidating because there are so many questions that need to be asked at least occasionally.