Benjamin Bruecker said:

Generally speaking, can I reuse the translation memory? Or do format-specific information prevent reuse?

Yes... keeping in mind that formatting differences and segmentation differences could influence your leverage when using different filetypes.  But generally... yes.

Lets say I have a in different filetye a sentence, which contains a bold formatet part:

• Help- and Manual XML: <para styleclass="Normal"><text styleclass="Normal" translate="true">Activate the licence under the </text><text styleclass="Bold" translate="true">registry</text><text styleclass="Normal" translate="true"> tab.</text></para>
• HTML/Flare XML: <p>Activate the licence under the <strong>registry</strong> tab.</p>

Let's assume that I manage to create filters that are useful for Help and Manual XML. The translation memory is first created when the Help and Manual sentence is translated. Then the same translation memory is used for a flare document. Would I get a 101% hit for this example?

Probably not because you'd have penalties depending on how you prepared the files.  So like this for example:

I could turn off the formatting penalty and then see a 101% (or Context Match as we don't use the term 101%):

But this would not be the right thing to do because the tag formatting is incorrect.

You might find this article interesting:

https://multifarious.filkin.com/2013/05/02/a-clean-editing-environment/

Probably not because you'd have penalties depending on how you prepared the files.  So like this for example:

I could turn off the formatting penalty and then see a 101% (or Context Match as we don't use the term 101%):

But this would not be the right thing to do because the tag formatting is incorrect.

You might find this article interesting:

https://multifarious.filkin.com/2013/05/02/a-clean-editing-environment/