TRIRIGA Semantic Search

This is a quick proof of concept (POC) written that allows the user to pick a query and perform semantic search on the results. The top five results are returned (in order of similarity).

This project is inspired by SemanticFinder.

Read more about this UX application in my blog post.

Installation Instructions

1. Create a new UX Web View with type WEB_APP
2. Set the Root folder value to dist and index filename to index.html
3. Create a Model & View record with the newly created view and any model
4. Create a new application record withthe M&V from above and the exposed name semanticSearch and an instance ID of -1
5. Clone this repository
6. Run npm install to install dependencies
7. Run npm run build to create the final distribution files under the dist folder
8. Deploy the code using the command npx @tririga/tri-deploy -u <USER> -p <PASS> -v <VIEW_EXPOSED_NAME> -t <TRIRIGA> -v -m 3
9. Navigate to <TRIRIGA>/app/semanticSearch to view the application

POC Notes

1. I tested this code on my own sample data. Please open an issue if there are any issues with a TRIRIGA backend
2. Only queries with the tag Semantic should appear in the report list. This is hardcoded for the POC
3. The query MUST only contain one text field for semantic search. This is hardcoded to reduce POC complexity
4. Only the first 1000 rows of the query generate embeddings. This is to reduce the memory usage
5. Environments with context paths MAY require additional tweaks to the code. Send me a note if you are experiencing issues and we can debug this together
6. My sample data is provided in the file lease_clauses.md