Autogenerated YouTube transcript[00:00] in the previous lesson we left off at this point. And the users asking what is the volume of Saturn you can see that the agent comes back with you asked for the radius of Saturn. Now this actually is a very I think of it as a really good example for explaining quite a few concepts in Dialogflow. Now the first thing is how intelligent is Dialogflow really. In fact if you ask for the [00:30] volume of Saturn it’s coming back with the radius of Saturn and you might conclude that you know that’s not really very smart at all and it happens that you can have some settings in this agent if you can go to this you can go to this tab over here. And you can actually like change this ml setting the classification threshold if you make it very high it will not make an error like that but let’s not let’s not worry about that at the moment for now what we are ⌛ [01:00] really concerned about is if the user is asking for the volume of Saturn it’s coming back with the value for the the response for the radius of Saturn question.
And this tells us that you can if you want to have a bot which can
Truly handle every possible question that the user can ask that can only be done by adding a lot of these intents and that’s probably not a good idea so [01:30] you still want to have a bot which is quite well defined in scope and those are the BOTS which are probably going to be the most practical. And the other thing that I can you can infer is that the Dialogflow can only work with what it already has which means when the user is asking for what is the volume of Saturn it’s going to take a look at these four intents which is pretty much all it has and it’s going to ask itself ⌛ [02:00] okay which of these is closest to what the user has asked here and it’s selected the radius of Saturn in there are some reasons for it but something that is good to remember here is dialogue flows intent mapping at the moment of this recording which is October 2018 is still a black box it’s not open source there isn’t any way for people to know what algorithm or [02:30] what happens under the hood when this intent mapping is taking place so there isn’t any way for me.
To explain exactly why this intent was chosen and let’s say not this one I can you can make some you know good educated guesses based on understanding how the natural language understanding field works but it’s still going to be sort of a black box at the end of the day now let’s come ⌛ [03:00] back to this question here which is when the user is asking for the volume of Saturn what can we do
Now one of the things that you can do is you can create another intent and call it user asks for volume of Saturn and we have the same think I’ll need the volume Saturn show our order Steve on the [03:30] oscillator let’s beam consistently.
And the response will be you asked for the volume of Saturn
And you can go ahead and save this and. While this is saving and doing its training it’s let’s let’s discuss something now now here we have you can see that we have these three attributes that we have declared for ⌛ [04:00] Saturn we took a single planet which is Saturn. In this case and we declared a set of attributes one of them being color the other being radius. And the third one being volume and this gave us a way to declare the users intention for three different attributes for a single planet now at the point you might be thinking wait a minute this stuff like whatever we [04:30] are having as the training phrase inside these intents they are.
Also similar that is the intents training phrase for radius of Saturn color of Saturn volume of Saturn at the end of the day these training phases are looking so similar so why do we need to keep on repeatedly typing out all these training phrases is there a way that we can make it a bit simpler to do the same thing and ⌛ [05:00] actually
Transcript auto-formatted using ATF