Sustainable Entrepreneurship

Identification of Startups’ ESG Properties from Text Data with Machine Learning

This repository implements the method described in the paper :

Mansouri, S. and Momtaz, P.P., 2021. Financing Sustainable Entrepreneurship: ESG Measurement, Valuation, and Performance in Token Offerings. Valuation, and Performance in Token Offerings

• Available on the SSRN
• Project website: http://sustainableentrepreneurship.org/

Requirement

The codes tested on Python 3.6 (Anaconda).

Word-lists generation

1. Training textual data gathered via “Link_white_paper” provided in the The Token Offerings Research Database (TORD)

2. The training procedure follows this GitHub repository. Clone the repository.

3. Replace the textual data in the /data/input/ of the Cloned GitHub repository with the documents in the additional files/documents.zip

4. Replace the python file global_options.py with the file presented in the additional files/global_options.py, and adjust it for your machine’s hardware.

5. Follow the training procedure:

    python parse_parallel.py
    python clean_and_train.py
    python create_dict.py

1. The word-lists will appear in the outputs/dict/expanded_dict.csv

   The paper’s latest version of the word-lists are in additional files/expanded_dict.csv as well as here.

ESG score calculator

• The notebook presented in Notebooks/ESG score calculation.ipynb implements the method described in the paper.

• An easy-to-use web app where you can paste text and immediately obtain ESG scores for the text is also accessible via: https://sustainableentrepreneurship.org/