Assigning NAICS codes to Kickstarter projects, using the ChatGPT API from OpenAI

Introduction

The goal of this project is to assign NAICS codes to Kickstarter projects.

Kickstarter is a crowdfunding website to raise money for creative projects. The North American Industry Classification System (NAICS) is a standard for classifying businesses on the basis of their type of economic activity.

We started with already-scraped data about Kickstarter projects from 2014 to 2023. We used the ChatGPT API (model gpt-3.5-turbo) to assign NAICS codes to these projects, on the basis of these four fields available from the data:

• Name (Title of the project)
• Blurb (A short description of the project)
• Kickstarter category
• Kickstarter subcategory

Prerequisites

• Open an OpenAI account and generate an OpenAI API key.
• Install the requirements by typing the following on the terminal: pip install -r requirements.txt

Usage

• Create input data file: Read the raw data (which is in JSON format) into an input CSV file containing, at a minimum, the project id and the fields you want to use as your input to ChatGPT. Here the fields name, blurb, category and subcategory have been used.

• Obtain OpenAI API key: Set the OpenAI key as an environment variable.

  • On Linux, type in the terminal:

  export OPENAI_API_KEY=<your openai api key>

  • On Windows, type in the command prompt:

  setx OPENAI_API_KEY <your openai api key>

• Create configuration file: Specify user-defined variables in the config.json file.

• Run program: Run the program from the terminal:

  python src/assign_naics_code.py

Input

The input is a CSV file with information about Kickstarter projects. Here is an example with a few rows and the relevant columns shown:

name	blurb	category	subcategory
Herbal Teas	All your herbal tea remedies here. From colds and sore throats to cramps and stress relaxation, you can find it here.	Food	Drinks
The Meat Candy Experience	The BEST beef sticks, beef jerky and signature sauces you will find anywhere...PERIOD.	Food	Small batch
WHILE THE TREES SLEEP	CalArts thesis film inspired by true events surrounding the 1965 murder of Civil Rights activist Viola Liuzzo.	Film & video	Shorts
Orage: Interactive Lightning Experience at Burning Man 2015	A Musical Lightning Storm. Get struck by lightning while collaboratively playing music with a 18-foot-tall musical Tesla Coil.	Art	Installations
Graphic design tools for award-winning designs	Professional graphic designers need professional tools. These are the best.	Design	Graphic design
The Gettysburg Story	America’s greatest battle as you have never seen it before: 'The Gettysburg Story' on public television.	Film & video	Television

Output

The output is a CSV file which is the input file with additional columns containing the NAICS codes as well as the number of input tokens used and the output tokens produced by ChatGPT. Here is the output corresponding to the example input above, with the NAICS codes being from the year 2017:

name	blurb	category	subcategory	naics code	input tokens	output tokens
Herbal Teas	All your herbal tea remedies here. From colds and sore throats to cramps and stress relaxation, you can find it here.	Food	Drinks	7223	108	2
The Meat Candy Experience	The BEST beef sticks, beef jerky and signature sauces you will find anywhere...PERIOD.	Food	Small batch	3116	101	2
WHILE THE TREES SLEEP	CalArts thesis film inspired by true events surrounding the 1965 murder of Civil Rights activist Viola Liuzzo.	Film & video	Shorts	5121	111	2
Orage: Interactive Lightning Experience at Burning Man 2015	A Musical Lightning Storm. Get struck by lightning while collaboratively playing music with a 18-foot-tall musical Tesla Coil.	Art	Installations	7113	115	2
Graphic design tools for award-winning designs	Professional graphic designers need professional tools. These are the best.	Design	Graphic design	5414	97	2
The Gettysburg Story	America’s greatest battle as you have never seen it before: 'The Gettysburg Story' on public television.	Film & video	Television	5121	110	2

License

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

About the project

Date: July - September 2023

Researcher:

• Nicola Cortinovis (n.cortinovis@uu.nl)

Research Assistant:

• Marte Vroom (m.m.vroom@uu.nl)

Research Engineers:

• Modhurita Mitra (m.mitra@uu.nl)
• Shiva Nadi Najafabadi (s.nadinajafabadi@uu.nl)
• Parisa Zahedi (p.zahedi@uu.nl)