Assignment5 & Final

[steeley21]

My Final Project Repo: https://github.com/steeley21/Backlogr

[EthanAGi]

Technical Implementation

Category	Criteria	Confirm/Deny
AI Integration	AI is used as a core part of the application's functionality (not just a novelty)	❌
Vector Search	Vector search is implemented and used in the application	❌
Back End	Uses classes, services, controllers, and DTOs appropriately	✅
Back End	Controllers are lean (logic lives in services)	✅
Front End	Uses classes and components appropriately	✅
Front End	Vuetify components used where appropriate	✅
Unit Tests	Unit tests cover core back-end functionality	✅
Unit Tests	Unit tests cover core front-end functionality	✅
Deployment	Application is deployed to Azure via a CI/CD pipeline	✅
Database	Application uses a database (back end connects to a persistent data store)	✅
UI/UX	UI looks polished, is a joy to use, and follows best practices	✅

Looks great so far!
I may be incorrect on this, but I believe that you're using a keyword-based search with token scoring instead of a vector search. Inside your StubSemanticSearchService.cs, the line (26 - 30):

var queryTokens = query
.Split(' ', StringSplitOptions.RemoveEmptyEntries | StringSplitOptions.TrimEntries)
.Select(token => token.ToLowerInvariant())
.Distinct()
.ToList();

Allows for exact keyword matching, instead of vector comparisons. I believe that this is fairly easy to fix! To do this, you could generate an embedding vector for the query and compare it against stored embeddings for each game instead of splitting the query into tokens. I think that if you just replace StubSemanticSearchService.cs with an implementation using AI to generate embeddings, you should be good! It looks great so far!