Post by ummefatihaayat22 on Feb 13, 2024 11:30:46 GMT 1
On the other hand, slots, also called slots or parameters, are a conversational strategy to collect user information and allow context variables to be put to use. Although the way they are implemented varies depending on which platform is used, the slots will always look for the information indicated in the user's response and store it in a context variable.
There are mandatory and optional slots . The latter will try to Germany Telemarketing Data detect in the user input the information to save in the context variable; if it is present, they store it, and if not, the dialogue continues. However, mandatory slots try to collect essential information, so this must be stored in the variable in order to continue with the conversation. Thus, when the information to be detected by the slot is not found in the user's response, they will be asked for said information until they provide it.
There are other strategies to process mandatory data that, depending on the type of chatbot and its purpose, may be more appropriate, but it is very interesting to know the option of slots because they help save time and make the assistant more condensed in less time. nodes.The following explains, using an example of a conversation, how to process information necessary for the dialogue through subnodes (explained in the previous post) and through slots.
How to collect essential information for dialogue. To explain with a practical example how the two strategies work to collect essential information for dialogue, we will return to the example used in the post How to create a chatbot with Machine Learning and NLP?: an assistant for a florist. To do this, we are going to consider a possible dialogue with this chatbot.
User: I would like to place an order. First of all, let's look at the information collection strategy using subnodes . As shown in Figure 1, this chatbot has four root nodes: one for each intention to be detected schedule , address and order and another for retrieving the dialogue. The node that jumps when #order is detected is the most complex, since, once the intention is activated, it enters through one of its five subnodes depending on the entities found in the user's input. Therefore, to exit through the fifth subnode (the one that confirms the order), the assistant always has to have collected certain essential information: the entities @producto and @flor , stored in the variables $producto and $flor.