This repository contains notebooks and code used in the GenAI Mini Simulator.

It is published mainly for illustrative purposes, not intended for any production use.

There are five notebooks that illustrate the steps of review analysis pipeline.

We try to keep them simple and do not use any frameworks

• Run pip install -r requirements.txt to install necessary libraries
• Input your OpenAI token in run_ctx = RunCtx("<YOUR_TOKEN>") cells
• Notebooks use gpt-3.5-turbo by default to avoid excessive costs
• Notebooks use small samples by default to avoid excessive costs
• Intermediate data is saved and can be inspected
• Run notebooks sequentially for 0 to 4
• All prompts can be found in src/prompts/prompts.py, with the same versioning as in the course

Notebook: 0_fetch_reviews.ipynb

Here we use a slightly patched app_store_scraper library to download a sample of app reviews

Notebook: 1_extract_topics_and_sentiments.ipynb

Here we execute a pipeline for topic and sentiment extraction

Notebook: 2_categorize_topics.ipynb

Here we execute a pipeline for finding main topic categories

Notebook: 3_map_topics_to_categories.ipynb

Here we execute a pipeline to map topic into category

Notebook: 4_analyze_results.ipynb

Here we visualize the results of review analysis

This project is licensed under the terms of the MIT License.