Similarity search

BusinessGPT offers two types of Similarity searches

  1. Invoking AWS similarity search using Amazon Bedrock on data ingested in AWS as a knowledge base

  2. Invoking BusinessGPT similarity search on data ingested in BusinesGPT

 

AWS Bedrock similarity search

After you enter the term, the gateway performs a search in the defined S3 knowledge bases. Currently, the knowledge base is defined by the support in the gateway configuration.

The use can limit the similarity configuration in the user settings.

amount_of_results_requested": (Default 500)

k_requested": Default 3 Toot tip: K most similar items to a query item,

Similarity score 0-100. Default 0. Tooltip - how similar the results should be to the search term.

Number of sample items to show - Default 10. Max 100.

 

The search API returns a Json file with all the links to the files and their name and the relevant chuck of text that was similar to the search term.

The Search page displays the top X items that are most similar . Each item contain a link to open and the chunk with the similarity score.

X is set by the user in the parameter of “Number of sample items to show.”

Once records are displayed, the user can download the full search results in Excel.

The Excel spreadsheet will also be saved in the content list with a name starting with the time, user, and term search.

For example “2024-09-19 yoav crombie, vegetables .xls”