Overlooked errors

Of course LT does not detect all errors. Some because it is impossible, because it could also be correct, some because there is no rule covering it.

Apart from my proposed function to warn for word sequences that are unlikely or known to be wrong, we could ask the user ‘did LT overlook a mistake? Please supply us with some example sentences showing this mistake, and a corrected version. We can use that to improve the detection of mistakes. (Do not include any sensitive information!)’

In fact, the relevant range of words could immediately be added to a data file, to be used by LT for signaling this mistake. When this mistake is shown, LT could state ‘one of the LT users reported this as a mistake. Is this correct yes/no?’ Leading to more value.

The data could be shown in some LT web interface, for the community to see and maybe build a rule on.