In my personal opinion, some of these matches may be genuine and some may not, and based solely on this kind of analysis, it is not possible to know definitively. I feel that this method of comparing facial features and personality traits, as impressive as it is, doesn't constitute scientific proof in the sense that Dr. Stevenson's work does. I am also personally skeptical that reincarnations can overlap, with the same soul living as an old man and as a child prodigy simultaneously. I think it's far more likely that these proposed matches are simply wrong guesses.* That being said, I think it's quite possible that some percentage, perhaps as high as 75-80%, of the matches proposed this way are accurate, and it's fascinating to view them side-by-side! (see also the recommended book section and the description of "Return of the Revolutionaries".) In the case of Dr. Walter Semkiw's work in particular, reliance on confirmation by one medium channeling a spirit entity should not be considered the "bottom line" for a case! (The medium's reading is not used as additional input--time and again in Dr. Semkiw's book it is relied upon as the final authority as to whether to accept or to reject a proposed past-life match.) At the very least, to be considered confirmation, several different mediums should be found to have come to the same conclusions about a case independently, under controlled conditions which ensure that they had no communication with each other nor any prior knowledge of the case. Otherwise, the entire method, and by inference the respect the public has for reincarnation research itself, hangs on the credibility of that one medium. This simply isn't science, and in my opinion it does as much damage as good to the credibility of reincarnation studies to put it forth as science.
One way that this method could provide stronger proof would be if the current-life personality had documented memories of the previous life which were not a part of the general public record. Suppose, for example, a person writes letters describing memories of having been a historical personality, or better yet publishes them, and that these facts about the historical person are not part of the general public record. Then, suppose that the facts are unearthed from an obscure or previously unavailable source, and that there is no reasonable way they could have been known by the subject previously. If the memories are specific enough and extensive enough, this, taken together with the other comparisons of facial features, likes and dislikes, handwriting anaylsis and so-on, would in my opinion constitute much stronger proof.
There are people doing comparison astrological charts on reincarnation matches, which again would be enlightening but wouldn't constitute proof. DNA comparisons have been proposed, but my guess is that there will be found no similarity in DNA. That finding would not disprove the matches at all, because apparently it is the mind of the incoming person which uses the DNA as raw material to construct the new body. I think the current materialistic model of genetics assigns too strong a role to physical DNA.