OK, not that serious... but... here is proof that the AI needs to update it's "AMAZON.TVSeries" built in slot Labels :)
She immediately thinks I'm referencing the second Captain America Movie, when I am, literally, asking for the new TV series. Most likely because the 'weight' (in the NN) is on the "Winter Solider".
I think someone needs to tell Alexa about "The Falcon, and the Winter Solider" being a TV show... LOL!
I realize I could circumvent this with custom slot types.
Also, my skill looks at the "Selected intent" and ignores the "considered intents".
Note:
She did the same thing with WandaVision.
It could be a phoneme issue, but she thought I was saying "1 Division", or "One Division".
This also leads me to believe she doesn't necessarily know that show either.
Because "Winter Soldier" fits a slot, I can't just catch misrecognition and act on it.
"One Division" isn't a cataloged database entry, so I can catch the misrecognition and persuade the response to show the proper item by keeping a list of misrecognitions in the back end code.
//My misrecognition list
switch(slot.value)
{
case "One Division" return "WandaVision";
case "1 Division" return "WandaVision";
case //<-- Can't put "Winter Solider" here without messing up the Movie slots.
}
side note:
(Thanks Jeff Blankenburg for the idea of keeping track of misrecognitions in my skill... all hail Office Hours!)
