Tags represent named entities that can be present in a user utterance. Let's say a user tells your chatbot: "Hi, my name is John Doe. I ordered your product x on November 12, and I have a question about my order's status. In this example utterance, all the bolded text segments represent tags. Notice that each tag falls in a category, or tag type. For example, the text "John Doe" is a name, "x" is a product, and "November 12" is the order date. Here, "name", "product", and "order date" are the tag types. When your chatbot has tag types associated with it, your Bavard chatbot can automatically extract relevant information from your users' speech, and store them as slots, which you can then use. For more information about what you can use extracted tag values for, see slots.
To begin leveraging tag extraction in our Bavard chatbot, go to the Tags page of the NLP section in the Bavard web app. Next, define your tag types, and give a few natural language examples of each. For example:
- "My name is John Doe"
- "They call me John Doe"
- "I'm John Doe"
Note that your tag types can be whatever you want, whatever fields you care about. In the example above, you could have separate tag types for first name and last name, or just a single tag type "name."