Since January 2023, I've been working with a startup in the UC Berkeley incubator program. The company is called DaanMatch. It is the brain child of my friend Cara Arellano. Cara had an idea that was inspired by her work with the Human Rights Center at UC Berkeley. The center was documenting examples of human rights abuses using video, images and other media together with tools like reverse image lookup in Google. Cara is writing her thesis about the Indian government's Corporate Social Responsibility (CSR) program. India has a very ambitious program where large corporations are required to donate back a small percentage of their profit via CSR programs. CSR programs fund local charities (aka Non-Governmental Organizations or NGOs). One of the problems she has identified is that the NGOs that often provide the best Return on Investment are very small and lack basic business skills. As a result much of the CSR money goes to large NGOs that spend much more on salaries, audits, consultants, etc. compared to the small NGOs who are closer to the community (especially in rural areas) and spend much more of the donated money on helping those in need.
The "elevator pitch" for DaanMatch is to enable small NGOs to use similar technologies as the Human Rights center only instead of using them to document human rights abuses, they will use them to document work done on projects funded by CSR projects. From this seed sprouted a number of other ideas of ways to use technology to make it easier for small NGOs to acquire and execute CSR projects without requiring significant business skills. E.g., chatbots can be used to simplify the process of filling out forms to request CSR funds.
Originally, the architecture for DaanMatch was a traditional relational database with a mobile GUI. This is required for the NGOs because most of the small NGOs only use their phones to access the Internet. However, as we began to collect data we had another idea, a system geared more toward business users that would utilize a knowledge graph. That's what I've been working on this year along with a very bright intern from the UCB Data Science program named Tejas Jyothi who has done most of the Python programming, especially the GUI which we developed in QT for Python.
One of the goals for the knowledge graph sub-system is to correlate NGOs and CSR programs with the UN Sustainable Development Goals. That was the reason I created a UN SDG ontology a while back; We also leveraged the Free Text Index (FTI) feature in AllegroGraph to match up the descriptions of each SDG Goal and Task to the mission and vision of NGOs and CSR programs. The FTI is a very powerful way to match up text properties on different knowledge graph objects using basic NLP tools such as bag of words, N-grams, and stemming. So far, we have published two papers on our work.
Most recently we published The DaanMatch System: Matching NGOs with CSRs using the UN Sustainable Development Goals in the IEEE Global Humanitarian Technology Conference (GHTC). This paper described the entire system, both the mobile system and the knowledge graph. Previous to that we published the paper: DaanKG: An Ontology model of the UN Sustainable Development Goals to Facilitate and Improve Corporate Social Responsibility in the International Workshop on Ontologies for Services and Society (OSS2023) that was part of the FOIS 2023 conference.
The following figures are just two examples of how we leverage the Gruff graph visualization tool in AllegroGraph to provide various visualizations of the knowledge graph. The first figure shows some of the individuals that make up the graph for the model of the UN SDGs. The purple nodes are the high-level goals such as to eliminate poverty. Each goal has several targets (shown as green nodes) that are more concrete and measurable. Each target also has one or more indicators (not shown in this graph) which measure how well the target is being met.
The second graph shows some of the NGOs and some of the objects in the graph that they connect to. This includes UN SDG Goals that the NLP analysis using the AllegroGraph FTI feature discovered were relevant to each NGO. The GitHub site for the project can be found here.
Comments