With automated QA rules in place, every linguist/project participant in each workflow step has to manually deal with all the flagged issues. At times there can be a number of false positives and that requires the user to ignore them in order to proceed. Once this has been dealt with, the next time a similar project is run where either the same content is leveraged from TM, or the same issues arise on another piece of content, the linguists again have to deal with each QA issue by reviewing and then ignoring the ones that are false positives. This also happens at every workflow stage. As you can imagine, this can take quite some time and when there are a lot of issues to resolve that are false positives, then it becomes a nuisance rather than a supportive feature.

We're looking to see if decisions on false positives could be stored and remembered, so if the same situation / context arises again, it is also ignored automatically and does not get flagged again. Much like you can train a dictionary new vocabulary / phrases.