Random Sample Of a Population


I have a file with more than 25K Support Cases that spam across many different categories. In the other hand I have 5 categories that are labeled as the top 5 categories my to solve. I want to check if those allegedly top 5 categories are the top 5 categories that i have in my population. However is not possible to read 25K support cases, therefore I want to take a random sample. I need to be sure, within a confidence level of at leas 95% that the distribution of categories in the sample represents the one in the population.

Te problem is that i cannot get to think how to calculate the confidence interval. I don't have a Mean for what I want to measure, or at least I don't see how to determine it.

Any help or guidance will be much appreciated.