You might have noticed that I instruct both GPT and PaLM to start the intent name with a verb.
This is actually an important step, because of these reasons:
- intents are natural verbs (user “intends” to do something)
- customer chat logs are very often based on the customer wanting to do some task and not being able complete it
- Large Language Models can do a much better job of inferring good verbs compared to humans simply because they usually have a much larger vocabulary
And using a verb as the first word in intents is also beneficial in another important way for this particular task.
The intent name format also helps the next step of the clustering task
Once you ask the LLM to do the initial clustering, you will usually have to do a next step of manually grouping the sentences.
By loading the output of the LLM into a spreadsheet (or even better, a tool like Airtable) you can go one step further and group the intent names into actual Dialogflow intents more easily.
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