Sberbank: ESGify NLP model

The main articles are:

• Speech Recognition (Technology, Market)
• Speech technology: On the path from recognition to understanding

ESGify is an open-source NLP model for classifying ESG risks that analyzes text data built on the BERT architecture.

2023: Introduction of ESG Risk Assessment Tool for Companies

Specialists from the Laboratory of Artificial Intelligence and the ESG Directorate of Sberbank have developed a convenient tool for assessing the ESG risks of companies ― an open ESGify NLP model. Alexander Vedyakhin, First Deputy Chairman of the Management Board of Sberbank, spoke about this on November 22, 2023.

A model based on NLP (Natural Language Processing) technology allows you to process and structure text, information supports and Russian. English A datacet (set) of data 2,500 unique texts on ESG risks was used to train the model. To improve the quality, a two-stage verification of the results of marking the datacet was carried out.

The classifier is developed on the basis of a risk-oriented approach. The model is able to correlate any text from open sources with a certain type of realized risk or indicate its absence. Thus, the service helps to solve the issue of analyzing ESG risks based on public information in the absence of standardized data by companies.

The service is embedded in any ESG risk assessment products or mechanisms. ESG risk classification helps the development of ESG practices in Data Science and business communities. Sberbank's AI model can help assess the investment attractiveness of a business taking into account ESG risks, as well as help in checking suppliers, counterparties and customers for compliance with ESG principles. The service will also be useful to consulting companies and analytics agencies.

Шаблон:Quote 'author = said Alexander Vedyakhin, First Deputy Chairman of the Management Board of Sberbank.