Document-based Chatbot AI

from Vagelis H.

We have discussed in the previous article how a chatbot can use a collection of documents to answer user questions. A simpler method is to return the part of the document that matches the question, while a more difficult method is Question-Answer (QA), which tries to generate the actual answer to the question.

SmartBot360 is at the forefront of document-based chatbot-like capabilities that enable the creation of effective customer service chatbots. In this article, we’ll discuss the ways SmartBot360 offers to enable document-based AI in a chatbot.

Check for a live demo https://smartbot360.com/demo/docaiwhich shows the different ways to query documents in SmartBot360.

The first two paths return the relevant segment, while the third tries to answer the question directly.

1) Document searchThis method allows the chatbot to search the pages of one or more websites. The bot administrator first specifies a set of URLs (eg smartbot360.com/documentation) to be indexed, then places a Documentation search box in the bot’s flowchart. The bot then asks the user to enter a question and returns the most relevant snippets with links to documents containing them, as shown below.



Source link