Environmental, Social, and Governance (ESG) compliance has emerged as a crucial benchmark in contemporary corporate assessment. The present work develops and evaluates a software tool called ESG-Consultant to facilitate ESG compliance for companies. We also discuss two tasks related to the application of ESG: determining the Applicability of ESG Law to companies and ESG Law Question Answering. To the best of our knowledge, no previous works tackle these tasks specifically on ESG regulations. For both tasks, we create datasets and develop pipelines based on Retrieval Augmented Generation using LLMs. The results are quite promising, proving that our system can be effectively used to support, but under no circumstances replace practitioners. We hope this work contributes to more effective corporate practices for a sustainable future. The code of our tool and demonstration system are available online ( https://github.com/angelonti/ai4esg_experiments , https://ai4esg-app.azurewebsites.net/ (pass1: e2c306nt for the annotations), https://www.youtube.com/watch?v=JKKGD3lFiUA ).

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

ESG-Consultant: Developing of an ESG Compliance Consulting Tool for Companies Using RAG

  • Angel Ontiveros,
  • Irina Nikishina,
  • Moritz Gomm,
  • Christopher Schmitt,
  • Chris Biemann

摘要

Environmental, Social, and Governance (ESG) compliance has emerged as a crucial benchmark in contemporary corporate assessment. The present work develops and evaluates a software tool called ESG-Consultant to facilitate ESG compliance for companies. We also discuss two tasks related to the application of ESG: determining the Applicability of ESG Law to companies and ESG Law Question Answering. To the best of our knowledge, no previous works tackle these tasks specifically on ESG regulations. For both tasks, we create datasets and develop pipelines based on Retrieval Augmented Generation using LLMs. The results are quite promising, proving that our system can be effectively used to support, but under no circumstances replace practitioners. We hope this work contributes to more effective corporate practices for a sustainable future. The code of our tool and demonstration system are available online ( https://github.com/angelonti/ai4esg_experiments , https://ai4esg-app.azurewebsites.net/ (pass1: e2c306nt for the annotations), https://www.youtube.com/watch?v=JKKGD3lFiUA ).