Hi Moveworks community!!
We are having a few use cases where we need to analyze large tabular datasets to get data based on similarity.
I was thinking maybe it would be interesting for Moveworks to have its own datalake to store tabular files. And a built in action to use a similarity model (k nearest neighbor or cosine similarity) to get items based on similarity.
I know Moveworks has its data ingestion connectors which works with knowledge and SharePoint data But also tabular data might also be interesting. Since API data analysis is limited and Moveworks cannot process one thousand json api items.
For example one of the use cases - we cannot use contain to lookup data, because data quality is not good and with similarity models this works very well. Also if we want to do recommendation systems within the chatbot this might also come in handy to recommend trainings and finance data.
I know most of this can be handled via middleware, but this is exact what we want to avoid and to everything on moveworks :)
just an idea!
Thank you!
