For our data evaluation process, we need to efficiently manage and extract the data related to the user's current code implementation and the associated responses from the model (pseudo code, hint code, feedback). Currently this data is stored in a single table which could make the extraction process challenging.
To do:
Seperate Tables
create seperate tables for user code and model responses to maintain data seperation.
_LLM_usercode table: Stores the user code with metadata and a timestamp(ID's for exercise, course, assignment)
By doing this, we can write SQL queries that join user code and their model responses based on the ID's and the timestamps that exist in both tables.
For our data evaluation process, we need to efficiently manage and extract the data related to the user's current code implementation and the associated responses from the model (pseudo code, hint code, feedback). Currently this data is stored in a single table which could make the extraction process challenging.
To do:
By doing this, we can write SQL queries that join user code and their model responses based on the ID's and the timestamps that exist in both tables.