It depends on what the purpose of this is and who is the recipient of such a result (i.e., who is going to make any useful decisions based on the result of this). That is, you can, of course, combine the numbers, but what for? Can you provide us with more details?

I want for summarizing the results. That is want to give one output from Two dataset by using naive bayes combiner. So, that it becomes easy for user as they dont have to refer both the datadets.

please tell me the method.

Ah, the problem is that by combining you may hide important details. It may be better to consider various approaches to combine and present apart from ONLY the naïve Bayes combination. It depends on what the data sets are, for example, if one is before a treatment and the other is after, then it may be more interesting to check the changes in accuracy etc. and, if you have multiple sets, perform a sign test.

Thank You @Jonas Mellin