What is slot filling?

Note: the slot filling article mentioned at the start of the video can be found here

(Autogenerated transcript)

[00:00:00]
 Alright, so if you go to my site, you will see this article where I talk about slot filling and you can see the URL is just miningbusinessdata.com slash dialogflow hyphen slot hyphen filling. And if you scroll down a little, you'll see this video here, which is about 15 minutes long. And I actually provide an overview of slot filling and I also explain sort of how it works under the hood. And then I make a case for why I avoid slot filling and remember that I'm still talking about dialogflow ES. And finally I talk about whether you really need slot filling for your bot. Now actually there are more things in that article, so I shouldn't have said finally. But for this particular course lesson, what I'm trying to say is that I also have a section in that same article which talks about whether you actually need slot filling that is required parameters. Because it turns out that a lot of people use slot filling even though they could have

[00:01:00]
 avoided slot filling altogether, they could have just used follow up intents or they could have constructed the whole flow just using input and output context, designing the conversation using just input and output context. So the reason to use slot filling is when there are a few things should happen. That is the user could come and say the same thing but they could provide multiple different parameters in the starting phase. And in that case, it actually makes a lot of sense to use slot filling. So I would say that the best like sort of canonical example for that is when user wants to book a flight. And you will see that a lot of people will start the first sentence or like the initial query will be very different. Sometimes they might say I would like to book a flight, sometimes they will say I would like to book a flight from place A to place B or they might say I want to book a flight for n number of people or they might say I would like to book a flight leaving on a certain

[00:02:00]
 date. These are all like the same intent that is they are representing the same intention to book a flight. But because of the way they finish the sentence where they could provide one or more parameters, all of which are actually necessary to book the flight remember that makes it a slot filling candidate. I mean this is a good reason, this is a good reason to use slot filling. So assuming that you are in need of a real slot filling bot that is a bot which really needs slot filling, I think that in CX you should go for slot filling. That is the only qualification as I am pointing out here is that it should have these you know I have a checklist here you know whether the can the opening sentence include any of the slots. Could the opening sentence include all the slots, could they be specified in any order,

[00:03:00]
 they all really required and so on. You know if you can go through this list and you are sure that it is a slot filling bot the one you are building then yes it makes sense to build it as a slot filling bot and it actually makes a lot of sense to use Dialogflow CX for this purpose. So what I am going to do next is I am going to explain how you can build a simple slot filling bot for the use case that I have presented over here and I will do that in the next lesson.

About this website

BotFlo1 was created by Aravind Mohanoor as a website which provided training and tools for non-programmers who were2 building Dialogflow chatbots.

This website has now expanded into other topics in Natural Language Processing, including the recent Large Language Models (GPT etc.) with a special focus on helping non-programmers identify and use the right tool for their specific NLP task. 

For example, when not to use GPT

1 BotFlo was previously called MiningBusinessData. That is why you see that name in many videos

2 And still are building Dialogflow chatbots. Dialogflow ES first evolved into Dialogflow CX, and Dialogflow CX itself evolved to add Generative AI features in mid-2023