I assume you're asking about automating a conversion process?
As has been demonstrated, just abuot anything can be automated, but not everything should be automated.
Additionally, there are so many styles and levels of complexity (not to mention multiple definitions) of use cases, that every one is going to be different.
You're going to want to study the use case provided for ideas on how the software is going to be used upon implementation. From them, you can (hopefully) extract some reasonable test objectives. And from the use cases you may be able to tell what the priority of the feature usage is.
Has someone given you a sample that you're wondering about? It always requires a lot of thought to determine what kind and how many test cases can be derived from use cases.