Enhancing discoverability with Generative AI

[g4brielvs]

Starting this thread inspired by Data Lab's Bring Your Own Bot (BYOB) session today to create connections, foster the community and brainstorm how explore new capabilities to enhance discoverability of data opportunities available through the Data Lab.

[g4brielvs]

@avsolatorio @brunosan @Holly-Transport Your input and thoughts are very much welcome!

[brunosan]

Thanks @g4brielvs. Indeed it was great to see what everyone is working on. It was an important realization for me that we are all trying to do the same (new) thing. E.g. Semantic search. This is not unique to us, but also something our clients (and the wider community) is trying to make as simple and customizable as possible.

I must emphasize that it would fantastic if we all re-use as much as others have done. E.g. it would be great if we push for everyone to structure data-centric use cases to use your templating system [e.g. this instance].

For my work, as I've shared, I'm working to facilitate the ecosystem of using datasets and analytics for improving AgriFood outcomes by helping stakeholders discover, assess and adapt available documentation and use-cases of options. This has a core of semantic search, so I'm trying to see how to abstract out with minimal extra effort what can be re-used.

Let's keep the coordination going and cross-visibility. May be worth too another BYOB session in 2 months and ask what happened with the demos presented?

[avsolatorio]

Hello @g4brielvs,

[Sharing my reply here too! 😄]

Indeed, there’s a clear opportunity to collaborate!

As you may know, we have been working on the document and data discoverability agenda for a while. This includes the development of foundational elements such as the creation of metadata standards, metadata editor, and metadata augmentation techniques, up to the production of tools including NADA, NLP4DEV, and the recent DEC Chat.

We are actively working towards open-sourcing all our solutions. For example, we are in the process of making llm4data, and soon the chat app, public with the understanding that we cannot build everything and could benefit from other innovative contributions from people within and, more importantly, outside the Bank.

As you mentioned, I also see an opportunity for the Data Lab to support our broader effort to encourage wider adoption of metadata standards and help build communities around open-source solutions.

We have yet to address and build solutions on alternative data as well. This is an excellent space to discuss and share insights deeply!

Thank you so much!