I have a client who uses a lot of videos in her chatbot and it mixes them up more often than not.
Is there a way of using the descriptions of the videos to minimize this, what are your suggestions on how to structure it?
Next I’m trying:
Description:
Keywords:
Negative Keywords:
Will see if results improve. Will the bot understand the “negative keywords”?
Hello Paradisian,
We use embedding & semantic search technologies to determine the optimal dataset internally.
We’re always willing to explore and refine our systems. However, I would require some data from your setup to proceed.
Would it be possible for you to send some information to hi@answerly.io for us to investigate?
We would need data such as your client’s address and the queries that are not returning the expected results.
Thank you