HI Nouran, 

That depends on what process you are using, and what validation technique you are using what discipline you are investigating. and what are your requirements. But as a rule of thumb, If you intend to use software visualization, Algorithm Animation might help you. This also depends on the scope of your research and how technicality you want to put in your research.

First you need to identify the criteria you want to study. For example, if you want to investigate the performance of your system, they might be some software out there you can use. But, if you want to validate the advice/recommendations that your system provides you usually need an expert for the validation.

So, your question is so general. First decide what aspects of your system needed to be evaluated and then find the relevant approach/technique/software for the evaluation.

I totally concur with Hossein. However, even the use of an expert to validate the advice/recommendations may not be viable. At least not for non-trivial cases (for trivial cases you might question the necessity of creating an expert system in the first place, of course). And here's why:

- Just assume that there are 10 different input criteria, with 4 possible values each. Simple math will show you that even validating 1% of the possible casses amounts to looking at 10.000 cases for validation. Something most experts will not exactly be happy to do.

But even if you were to get all the 1.048.576 cases validated, one might question whether

- a single expert would be enough (you know, experts tend to disagree on occasions)

- the human capacity to quantify probabilities (whic I expect are the outcomes of your system) is good enough to provide guidance

Since I don't know anything about the application area of the system, the following might be completely impossible, but you might consider validating the output on real, empiric datasets prospectively or retrospectively. For instance, if your system is designed to give a fault diagnosis for car engines, you might get hold of historic data and "gold standard" diagnoses from repair technicians' records.

Just a thought.