Objective, handle false positives better.

When 2 entities are similar, the bot might get them mixed up, making a false positive.

This feature is here to make sure that for the second try, the bot will choose the good entity.

Example scenario

Context

We will ask a first name. There are 6 possible choices:

As you can see “Anton” & “Antony” are similar.

Let’s see what happen if we answer “Antoni”.

• Since Anton & Antony were close the bot made a false positive at first.
• But for the second try the bot knew the name was not Anton, so he proposed Antony.

Important remark

Please notice that both “Anton” & “Antony” entity-values were close enough to “Antoni”. That is why they were proposed by the bot.

If we had written something less similar, the bot would have not “understand” → cf the example below, answering “Tony”: