I've been told that during translation, once a segment has been translated once, it will after that be recognized as a 100% match instead of as a repetition.
That's correct. That's the fundamental idea, concept and use of every translation memory.
I'd like to prevent this, uote]
Why? I don't see any reason why this could be useful? And what is the desired alternative? A "no match"? And if so, why would anyone like to get a no match for an already translated sentence? I suggest that you work without a Translation Memory if you do not want to get the advantage from it.
d was told there was a way of doing this in an html.ini file.[/quote]
This is by design of the translation memory concept and has nothing to do with the html.ini or any other configuration file.
However, you can create your own html and configure the html ini to exclude specific parts of the html file.
A typical use case are multilingual files like this html code:
<p lang="de">Das ist der erste Satz.</p>
<p lang="en">This is another sentence.</p>
You can configure the ini to prevent any "p" elements that have a attribute "lang" with the value "de" from analysis and translation.
But this has nothing to do with your requirement