The Connectivity Hub

for the Stockholm Environment Institute

Neural Alpha is working with the Stockholm Environment Institute to build out a knowledge-graph-based connectivity tool that leverages its expertise in large language models (LLMs) and data engineering to add value to heterogeneous data sets of research and other scholarly articles.

What is the Connectivity Hub?

Try the Connectivity Hub for yourself here: https://connectivity-hub.weadapt.org/

‍

‍

Originally launched in 2019, the Connectivity Hub is a "search and discovery" platform designed to help planners, decision-makers, researchers, policymakers, students, and citizens find knowledge and organizations working on climate change adaptation (CCA) and disaster risk reduction (DRR). The hub aims to close the knowledge gap between CCA and DRR communities by integrating 5 primary data sources into a single search application. The platform uses annotation and graph visualization to show the interconnectedness of CCA and DRR topics and concepts, and has catalyzed feedback and ideas for future advancements, including leveraging AI technologies.

‍

“The increasing ability to harness data and to repurpose it in innovative, visually powerful ways provides new avenues for learning from existing initiatives, and for accelerating the bridging of science to action”, said SEI Senior Research Fellow Sukaina Bharwani, who leads the project for the SEI.

Stockholm Environment Institute

The Stockholm Environment Institute (SEI), founded in 1989 by the Swedish Parliament, is an international non-profit research organization. Its creation and name were inspired by the 1972 United Nations Conference on the Human Environment in Stockholm, which resulted in the Stockholm Declaration. This Declaration's principles, still relevant today, encompass human rights, wellbeing, natural resource preservation, renewable resource maintenance, conservation, pollution, disaster risk, financial and technological transfer, urbanization, scientific research, international cooperation, and integrated development policy and planning.

Neural Alpha’s work on the Connectivity Hub

‍

The datasets underpinning the Connectivity Hub are being continually updated as new research is published and new scholarly articles are produced. Neural Alpha has solved three problems:

1. Where the categorization of new content is missing or incomplete, Neural Alpha uses LLM inference to accurately tag incoming articles with appropriate keywords. This enables links to related articles from other content sources to be identified, so users can see all related material in a single place.
2. The underlying taxonomy is also evolving. Neural Alpha has created a partially automated workflow to identify new keywords and add them to the taxonomy as new ideas and concepts are introduced to the research areas under consideration.
3. The existing content was rapidly growing stale. Neural Alpha has automated the collection of new content from all the content sources, as well as adding new ones, so the Connectivity Hub is always up-to-date with the latest material.

Without Neural Alpha’s work, none of these would be possible without an army of researchers to read all new articles and both link them appropriately to related research and identify new concepts for the taxonomy.

The taxonomy itself is closely managed by subject matter experts using workflow tools provided by Neural Alpha, but the majority of the text processing is done by LLM-based automated tools.

This allows the Connectivity Hub to be continually up-to-date and accurately categorised. With new content and rich links being added daily, its value grows geometrically.

‍Delivering value

Neural Alpha was able to deliver this iteration of the Connectivity Hub very rapidly and with a small team. We are subject matter experts in the relevant technologies and were able to hit the ground running when the brief was agreed.

We took the decision to concentrate on the areas where we could add clear value to the application: LLM-based content classification, LLM-based taxonomy development and automated ETL processes. We were able to work with the legacy code from a previous iteration and so avoided the need to redevelop features that had already been implemented.

Overall the project was delivered in weeks rather than months, and the client is already preparing for future enhancements with us. We look forward to continuing to develop this exciting application in partnership with them.

‍Related applications

‍Neural Alpha’s expertise in LLM and AI technologies comes from experience with other applications. In particular, its own product Responsible Capital uses these technologies to great effect to rapidly and accurately process corporate disclosure documents and expose the data using an efficient and productive interface, including Q&A and knowledge graph tools.

The AI-powered ESG data platform

Responsible Capital provides capabilities to enhance your ESG data collection and analysis, transform research and process efficiency, and drive sustainable, values-based decision making.

Leveraging fine-tuned generative AI and a substantive ESG knowledge base, Responsible Capital is creating a transformative impact across benchmarking, scorecards, due diligence, screening, risk management and more.

Sign up for a free trial here: https://responsiblecapital.io/