Textual Analysis: comparing textual descriptions and make classifications

I have two sets of company descriptions and keywords of their businesses, and I would like to develop a measure to capture how different they are based on these textual information. Descriptive figures such as histogram displaying the distribution of keywords is a simple example that I can easily think of. However, I would like to get some numerical measures using the "text" to capture the differences in the business scopes of the two groups of companies.