How NLP will accelerate EU Taxonomy reporting (and make way for a new era in sustainable finance)
Making the transition to a sustainable economy within just a few decades will require capital to be channeled into sustainable activities at an unprecedented scale and pace.
Europe is leading the way on this with flagship sustainable finance regulations like the EU Taxonomy which provide a clear framework to direct capital to truly sustainable projects.
However, the complexity of disclosing data for these regulations is currently creating a critical bottleneck to them being adopted and effective in time. Thankfully it’s also an area where AI and in particular Natural Language Processing (NLP) can really help. Today we’ll explore 5 ways NLP can accelerate sustainable finance:
- Activity classification: The EU Taxonomy provides a detailed definition of over 100 sustainable activities. However knowing which, if any, apply to your company, portfolio, supplier or customer list can be a tricky and time consuming process. Using NLP models to scan large volumes of unstructured documents and find evidence of these activities automatically can significantly increase accuracy compared to a manual approach, while reducing the time by orders of magnitude.
- Evidence extraction: Sustainable finance regulations such as the Taxonomy also include a large set of technical screening criteria to assess if an activity can be considered truly sustainable. NLP can help process millions of unstructured document sources, from sustainability reports to life cycle assessments on the hunt for evidence for these legal criteria, extracting the relevant passages and mapping them to the appropriate legal criteria automatically. When done manually, this process can otherwise take days or even weeks!
- Entity detection: Another issue with sustainable finance regulations is the large, interconnected and changing nature of associated regulations and legal requirements. For example it is common for sustainable finance regulations like the EU Taxonomy to refer to national legislation of member states which, in turn, points to further legal clauses in other documents. NLP can play an important role in being able to detect and extract these legal entities from the source text and create a knowledge graph mapping the interconnection of legal documents to help navigate this complex space faster, with the added bonus that it can be automatically updated and alert you as regulations change!
- Topic modeling: Sustainability is a notoriously complex topic as the scope of what needs to be included and measured constantly increases. For example, this year the scope of the EU Taxonomy will be increased from the original 2 objectives (climate mitigation and adaptation) to include 4 new areas (pollution, water, circular economy and biodiversity). NLP techniques like topic modeling can not only help track the emergence of new sustainability topics from news and publications, but can even show the developing sub-topics and concerns within these groups in order to help you stay ahead of the curve.
- Summarization: The progress of NLP abstractive summarization over the last years (with projects like Open.AI’s summarizer) has been astonishing. While there is still some way to go to commercialize this technology for legal text, it holds significant promise for supporting sustainable finance adoption with the prospect of rapidly ingesting and summarizing the deluge of legal documents surrounding regulations like the EU Taxonomy, allowing analysts and practitioners to understand the topic and implications of specific developments much faster.
At Briink, we are building the technology to make these applications possible and help move trillions of dollars into sustainable projects this decade. You can learn more about this on our website and stay updated on our journey by following us on Medium, LinkedIn and Twitter.