I would find it useful if I could select underlined text in the upper window of the Fragment Match window that resulted in showing only the Fragment Matches relevant to that part. In my field I'm usually presented with long sentences and quite often I'm presented with a long list of matches in the fragment window, which I find useful and which is the result of my settings and using a big TM. Narrowing this down a bit as I'm translating on a segment would make everything less cluttered and more efficient.
BTW, why does it not present the matches in the same order as they appear in the segment? The order seems to be a bit random. Probably the order in which they were found.