Welcome to the Movie & Book Recommendation System! This application was developed for the 2024 Hackathon: Assistants-api-llamaindex-mongodb-battle. The primary function of this app is to provide movie and/or book recommendations based solely on the user's description of the movie.
The development of the Movie & Book Recommendation System is divided into the following stages:
-
Data Filtering and Preprocessing:
- Collection of raw data from reliable sources.
- Filtering of data to remove any irrelevant information.
- Cleaning and preprocessing of data to ensure consistency and accuracy.
- Transformation of data into a format suitable for our machine learning models.
-
RAG Pipeline with LLamaindex, GPT-4, and MongoDB Vector Database:
- Development of a MongoDB vector database for storing and querying vector representations of data.
- Implementation of the Retrieval-Augmented Generation (RAG) pipeline.
- Integration of LLamaindex for efficient data indexing and retrieval.
- Use of GPT-4 for generating high-quality text based on user input.
-
LLM Testing with TruLens:
- Evaluation of the Language Learning Model (LLM) using TruLens to ensure its performance and reliability.
- Identification and rectification of any issues or discrepancies in the LLM.
- Continuous testing and improvement of the LLM based on user feedback and performance metrics.
-
Data Filtering: #################--: 90%
-
DataSet construction: #################: 100%
Run app_ERROR.py.
app.py only contains a description of the app.