Proposal: Community Powered Discord-to-Knowledge Base Curation Module

[Sha]

Name of Project: Curate and Chat POC - Community Powered Discord-to-Knowledge Base Curation Module

Proposal in one sentence: Community-powered Discord curation module that can feed content into databases like Notion and Airtable, enabling interfacing with existing Algovera AI modules, such as Q&A chatbots or chatbots over knowledge bases.

Description of the project and what problem it is solving: A community communication channel is a hard thing to manage and navigate. Especially in Discord, it is very hard to follow all the activity across different channels, threads, and forums. Also, it is hard to navigate a community’s knowledge base. We have a running Beta of an open-source curation module in usage on the CSm (Common Sensemakers) Discord server and on the TEC (Token Engineering Commons) Discord server. The curation module connects Discord to Airtable, it enables members of the community to mark and vote on Discord content, which is routed to Airtable views serving various applications - add to a community knowledge garden, add an event to a shared calendar, manage a shared Twitter account and more. Making the content available and discoverable by the community is a desirable next step. With the funds of this grant, we will integrate our product with Notion and incorporate it with the Algovera “Notion to GPT-3” workflow to:

• Allow the community to decide on relevant content to be stored in the knowledge base. This will enable the creation of a chatbot more representative of the community’s interests.
• Allow the community to curate frequently asked Q&A from Discord channels, to improve the chatbot’s Q&A ability.

In this way, we will give tools and empower the community to mark important content, make it accessible, increase discoverability, and improve wayfinding. Additionally, we will add precision to the chatbot by focusing on content that is important to the community.

Grant Deliverables:

• A module that feeds curated Discord messages into a Notion database:
  • Define a threshold on a pre-defined reaction (emojis on Discord message).
  • Monitor Discord messages where the defined reaction count passes the threshold.
  • Add these to a Notion database in a predefined format:
    • Title
    • Message content
    • Message author
    • Message score
    • Reactions on message
    • URL
    • Scraped URL
    • Conversation thread
• Running POC on the CSm server and potentially on the TEC server.
• Producing a POC with documentation and infrastructure to enable any community to onboard the service.
• Create an infrastructure for adding the workflow into Algovera’s AI workflows in the next phases.

Squad:

Squad Lead:

• Shahar Oriel (Sha#6179, @ShaRefOh)

Squad Members:

• Rochdi Khalid (rochdikhalid#7903)
• Ronen Tamari (ronent#2267, @rtk254)

[pfedprog]

This proposal looks very promising and is a great way to empower communities to mark and curate important content to improve discoverability and wayfinding. The proposed deliverables are comprehensive and the squad looks well-equipped to complete the project. One suggestion would be to add more detail on how the Notion database will be formatted and how the integration with Algovera’s AI workflow will be accomplished. Additionally, it would be helpful to have a timeline of when the project is expected to be completed. Overall, this is a great proposal and I look forward to seeing the results of this project.

[Sha]

Hey @pfedprog thanks for the feedback! As for your suggestions, perhaps I could elaborate a little over here:
We consider right new two use-cases,

1. Community curates it’s knowledge base in a certain topic, say Token Enginering (as the case for our design partners, the TEC). Community members or other who are interested in the topic and value the community’s knowledge would be able to have conversations with a chatbot on these certain topics, and the bot will be prompt by the knowledge base of the community.
2. Is more related to the Algovera’s Q&A chatbot workflow for learning about the community. Here community members would be able to curate good questions and good answers that appear in the discord server, and send them to a Notion database to feed the chatbot.

The two usecases require a different Notion formatting. The first will include creation of Notion Data base, or require from the community manager an id of a Notion database which include the next fields: Author of the discord message, reactions, URL (for attached URLs in the discord message), Jump URL (a URL of the discord message), Scraped URL in the page content (for example, if it is a blog post, it will scrape the post and display it on the page.) We need the scraped data so that the chatbot will have access to it.

We expect to have a running POC of this use case in our server and in the TEC’s by 20.3.

The second use-case will require appending the hole conversation thread, the question and the answers. We need to farther understand how the bot works to see if we need to get the whole conversation or just the questions and the marked answers. These will appear in the page content. and except of that I believe that the rest is similar to the last use case (without the scraping). We expect to achieve this by the 1.4.
Down the line we would like to add these flows to the Algovera AI workflows attached to the Notion+GPT-3 flow. And to have Algovera’s workflow appear in our UI. Also the next thing we would like to work on is connection the chatbot back into discord and close the loop That should be rather quick once we have the other use cases running.

I hope it brought more clarity.

Looking forward for realizing this!

[ijac]

This is a obvious user needs. Looking forward to see your POC.

[Sha]

I agree! We already have the POC, and I am working towards publishing it in the discord.

We did run into some complications, as one of our team members decided to leave (DevOps), so it took more work on my side to deploy it to the cloud.