I noticed there are many problems of understanding the Bot for a language other than the default one (English).
For example, in Italian, articles and apostrophes are interpreted incorrectly.
A word that contains an article and an apostrophe is considered as a single word creating problems for end users when they have to search for a specific FAQ within the KG.
For example, the word “un’eccedenza” contains the article “un” with the apostrophe " ’ ". It is considered by your algorithm as a single word and therefore “un’eccedenza”. In this case, when the end user searches for that same word without the article, the algorithm will have problems identifying the relevant FAQ.
Thank you for your reply.
That is what I’m currently trying to do. But having a lot of FAQs makes it difficult to handle every possible situation.
Another problem with your workaround is the possibility of spaces between the article, the apostrophe and the word: “un’ eccedenza”, “un’eccedenza”, “un ’ eccedenza”, “un ’ eccedenza”.
In this case, for each word there would be a multiplication factor x4.
I do not see a problem in the latest version of the Platform for the combinations you have mentioned. If possible please test it as per the settings I suggested. But I do agree that in general, the solution may need some work and handling.
We will try our best to make things better and more seamless for all languages supported. There are some technical challenges but as technology leaders, we will need to overcome them. Your support case is open. I am trying to get feedback from the dev too. Will keep you updated.