In the evolving landscape of corporate reporting, processing large amounts of unstructured data is crucial to aligning companies' Environmental, Social, and Governance (ESG) strategies with the challenges brought by climate change and social injustice. The latest European sustainability reporting directive (CSRD) requires companies to disclose clearly their impacts, risks, and opportunities in terms of ESG.
Using a variety of NLP techniques, enhanced by state-of-the-art text generation models, we will show how to automatically scan through thousands of reports and extract relevant information, best practices, and potential risks or opportunities.
This session will illustrate a real-world case study on the ESG landscape using different Natural Language Processing (NLP) techniques. By leveraging zero-shot Text classification, sentiment analysis, Named Entity Recognition (NER), clustering on top of matryoshka embeddings and Topic Modelling, we can improve the insights and decision-making. In addition, we will present how to take advantage of the latest advances in Generative AI.
This presentation will be particularly valuable for data scientists, ESG analysts, and Python enthusiasts interested in the intersection of AI and sustainable business practices. The main takeaways that you will get from this presentation include:
- Gaining insights into how topic modelling with Bertopic can be used in corporate ESG initiatives.
- Discovering how clustering techniques help tailor insights about Impact, Risks and Opportunities from large ESG reports.
- Understanding the role of matryoshka embeddings for enhancing the latency when processing ESG-related texts.
- Understanding the role of Generative AI in extracting actionable insights from vast amounts of ESG data.