search

aws-samples / amazon-bedrock-kendra-lex-chatbot

The repository offers a base implementation of incorporating an Amazon Bedrock based chatbot with Amazon Lex.
MIT No Attribution
42 stars 14 forks source link

Examples of context documents & questions #4

Open pundaripothini opened 8 months ago

pundaripothini commented 8 months ago

Hello, the blog works great out of box but the biggest issue is there are no examples on how to structure the context document in Kendra so my questions to the bot are much more contextual.

brettbs commented 8 months ago

Hi.. How many documents do you have indexed by Kendra?

The example uses the Retreive API, here is some documentation on how that works. https://docs.aws.amazon.com/kendra/latest/dg/searching-retrieve.html . Here is some additional information on Kendra relevance tuning. https://docs.aws.amazon.com/kendra/latest/dg/tuning.html#tuning-index

pundaripothini commented 8 months ago

I just have one. It would be great if you can provide an example document and the corresponding GenAi questions to ask so we know how to structure these documents. We're not Ai/ML specialists, trying to demo this in a contact center setup. Otherwise this blog will be very narrowly focused for people with a lot of time and interest to learn how kendra works.