Never be fully convinced of TM

As we all know TM (translation memory), it is very useful during the translation.
A translation memory is a collection of multilingual correspondences with optional control information, which stored with each correspondence. When we do translation, we can look for the words from the TM, This characterization abstracts away from the actual manner of storing the correspondences (one-one, one-many, or many-many).

When a translation memory is used to support a given direction of translation,
we can identify one segment of each correspondence as the source segment and another one as the target segment.

A given query with a current source segment may return a number of correspondences with matching stored source segments.

Though it is a powerful tool for translators, it just a tool, in which we should first import the translated sentences or articles we have finished. Later, when we do the same or similar field translation, we can look up the words in it or if the match rate is higher than what we have set up, the former full match or similar results will come out automatically. But the trouble is, if you import the wrong translated text, and you are fully convinced of TM, you will make the same mistakes over and over again, that is so terrible.

At the very start, I have no doubt, absolutely no doubt about the TM, one day some people reminded me that there were lots of mistakes in TM, I hug my half belief in it. But I begin to pay attention to this problem.

This week I was assigned a big task, the PM sent me a large TM, I was happy that lots of sentences matched with the TM, but when I read the fully matched sentences carefully, totally I found more than 6 mistakes in TM, I corrected them and told the PM, So the primary goal of this article is to tell you that make sure read the matched sentenced carefully from now on to avoid the mistakes.
Seeing is believing, never be fully convinced of TM.