Industrial Ecology Machine Learning Mapping

This Python module uses machine learning to map two classifications together, based on words similarity.

Uses in Industrial Ecology

Industrial Ecology is a field of study where a lot of disciplines co-exists and often interacts. Mapping is thus typically required in IE projects to connect two databases. This module can assist in mapping to

• Sector classifications (NAICS, NACE, ISIC, etc.)
• Product classifications (NAPCS, CPC, CPA, etc.)
• Life Cycle Impact Assessment (LCIA) classifications (ReCiPe, EF3.0, IMPACT World+, CML, etc.)

Installation

Install necessary module with pip

pip install -r requirements.txt

Getting started

Choose a reference classification, a machine learning model and the amount of guess the algorithm will display.

self = ML_mapping.Mapping(reference_classification='exiobase',
                          transformer_model='all-MiniLM-L6-v2',
                          number_of_guesses=5)

Then enter the inputs you want to match to the reference classification, calculate similarity scores and format the results

self.match_inputs(['ADPE System Configuration','Chocolate','Renting a film'])
self.calculate_scores()
self.format_results()

The module returns:

Classifications

The work for the following classifications has already been done in this module

• EIO/LCA databases
  • ecoinvent 3.8
  • ecoinvent 3.9
  • openIO-Canada
  • exiobase
  • USEEIO 2.0
  • GTAP 10
• Elementary flows classifications
  • ecoinvent 3.8
  • ecoinvent 3.9
  • IMPACT World+ 2.0
  • USEtox 2
  • EF 3.0
  • EF 3.1
• Product/Industry classifications
  • IOCC
  • NACE Rev.1.1
  • NACE Rev.2
  • CPA 2008
  • CPA 2.1
  • NAPCS 2017
  • NAPCS 2022
  • NAICS 2017
  • NAICS 2022
  • ISIC Rev.4
  • CPC 2.1
  • COICOP 2018

If the classification you want to match is not already covered, adding it is not complicated.

Credit

This module is simply applying the work of https://github.com/UKPLab/sentence-transformers