🌍🤖 DataUnion Foundation proposal for OceanDAO round 9

Part 1 - Proposal Submission

Name of project: DataUnion Foundation

The proposal in one sentence: Creating local income by crowdsourcing data for responsible AI.

Description of the project and what problem is it solving:

Problem

Responsible AI requires high data quality in respect of bias and fairness. Yet the marginal costs to solve the long-tail problem in AI increase.
Over the past decade, the digital economy has become highly concentrated and prone to monopolization.
SDG 10 - Reduced inequalities
Target 10.1: By 2030, progressively achieve and sustain income growth of the bottom 40 percent of the population at a rate higher than the national average (Growth rates of household income).

Solution

Using blockchain technology to establish DataUnions that enable local income, quality data, and data co-ownership.

Grant Deliverables:
[ ] New pitch deck
[ ] Mobile App - annotation feature
[ ] Celo network mobile wallet integration
[ ] Data Portal improvements

Which category best describes your project? Pick one.
Build / improve applications or integrations to Ocean

Which Fundamental Metric best describes your project?
Total Value Locked

What is the final product (e.g. App, URL, Medium, etc)?

Our alpha is already online, please go ahead and check it out.. The mobile app is currently under review in the Google Play store. The Data Portal v1 is available here. There will also be other data unions in our DataUnion ecosystem.

How does this project drive value to the “fundamental metric” and the overall Ocean ecosystem?

Let’s do a ROI calculation - this round we are focusing on facts rather than speculation:

Currently there are 615,483 $OCEAN locked (TVL) in the DataUnion.app liquidity pool on the Ocean marketplace. Additionally we also have an active Twitter account, a Discord and a Telegram community to bring in additional users into the Ocean ecosystem. And we have 281 users contributing on our alpha website. The unleashing of data is happening continuously, check it out in our dashboard.

bang = 615,483 $OCEAN TVL

buck = 139350 $OCEAN - this is explained in more detail in the financial section of the proposal, we assume 20.000 $OCEAN for this proposal here

(% chance of success) = 100% as we are talking about the current situation and the TVL is there

ROI = 615483 $OCEAN / 139350 $OCEAN * 1.0 = 4.41 which shows that the ROI of the project after this grant is above the demanded ROI of 1.0

Funding Amount: 17,500.00 USD
Proposal Wallet Address: 0xF9ac73f30dBe52c10e3d5950db66357f9d0be44D
Have you previously received an OceanDAO Grant?: Yes

Team Website (if applicable): https://dataunion.app
Twitter Handle: @DataUnionA
Discord Handle: DataUnion Foundation#1870
Contact Email: info@dataunion.app
Current country of residence: Singapore

Part 2 - Team:

Core Team:

Robin Lehmann

Dr. Mark Siebert

  • Role: partnerships, business development
  • Relevant Credentials (e.g.):
  • Data publishing (10yrs)
  • Business Development for Data Markets
  • Owning and driving global executive engagements and partnerships, Data and Open Science
  • Web3 experience: < 1 Year
  • Positioning businesses in emerging markets or innovative fields with focus on data and AI-driven solutions.
  • https://www.linkedin.com/in/dsiebert/

Sarah Kay

Akshay Patel

  • Role: developer
  • Relevant Credentials (e.g.):
  • Background/Experience:
    • Backend developer in the finance industry
    • Ocean Protocol Ambassador
    • Ocean Protocol Bounty hunter

Okpo Ekpenyong

Part 3 - Proposal Details

What problem is your project solving?

This is the final app humanity needs - it gives us the ability to contribute to AI and robotics and enable humanity to profit from the data this technology needs. We want to give everyone the ability to use their data for a better future and their own profit.

Here is our new pitch deck for DataUnion services.

In our pitch deck we are describing how new data unions can be created using the DataUnion tools and methodology on top of Ocean Protocol. This enables more and more data unions to be created and join forces to develop a common stack of tools and help each other to be successful. This reduces the overhead costs for the individual data union and allows them to focus on their domain specific topics and their community instead of having to develop the same set of tools and processes again.

As an example data union we are building an image dataset that is created and owned by the people that contribute to it. These contributions include uploading the data, annotating it as well as verifying that the data and the annotations are correct. By doing this the contributors are motivated to contribute highest quality content as they will be rewarded with shares of the dataset they contributed to. We are baking this slowly and release the features separately to have the maximal amount of feedback from our community.

So this covers the creation part of the project. But to refinance this process we have to find customers. The customers can create Data Bounties to increase the incentives for the creation of specific machine learning ready images. Customers can also bring their own datasets and ask only for annotation and/or verification by the contributors.

In the most important option for customers they can train algorithms on parts of the data via Ocean Protocol’s compute-to-data technology. In this way they can train object recognition algorithms without the risk of the data to leak thus the project remains in full ownership of the data. This is done via a marketplace on top of Ocean that we create and contains previews and evaluation of parts of the data (via a search engine on tags and annotations) as well as algorithms. The sale of these assets will be done via the liquidity pool of the whole dataset.

So the contributors not just benefit from their contributions but also from the usage of their contributions. This creates long term incentives to maintain control over the created tokens. This project is a showcase of ownership economy based on the economy creation capabilities of Ocean Protocol.

As a next step on our journey we are now starting to simulate the ecosystem and individual pools. There are already VisionTherapy and SenseNation in the Ocean ecosystem that are using our advice and expertise to create their own data unions. Here we are also taking the approach to work bottom up with them and help them where they need us.

Project Deliverables - Category

Data asset statistics:

Start of round 8:


Start of round 9:

Project Deliverables - Roadmap

Any prior work completed thus far?

Check our alpha web app, alpha data portal and [alpha mobile app](in review for the Google Playstore).

A lot of the work done in the project so far was recruitment of the team, organisational work, looking for funding and organising the work. We also published our alpha now and have made significant progress with our mobile application. The team now has 8 team members with different strengths to move the project forward. The work is organised via Trello and we have a communication infrastructure via Discord. We also have a social media presence and a website.

What is the project roadmap? That is: what are key milestones, and the target date for each milestone.

  • New annotation tools and mechanisms (ongoing)
  • Onboard new data unions to our tech and concepts (ongoing)
  • Release Swipe-AI (Q42021)
  • Simulate the token value flow using simulation tools (Q4 2021)
  • Increase decentralisation of our solution by moving the control to smart contracts (Q4 2021)
  • Algorithm training and sales via our Data Portal - have the data providers become co-owners of that as well (Q4 2021)
  • Allow addition of data while it resides on different storage (2022)
  • Include NLP to translate annotations and cater to a worldwide audience (2022)
  • Recruit more people and facilitate development via OceanDAO grants (ongoing)
  • Potentially launch a governance token (2022)

Please include the milestone: publish an article/tutorial explaining your project as part of the grant (eg medium, etc).

Our web app and Data Portal include tutorial videos explaining the project.

Any maintenance?

These prototypes are the beginning of the journey for the project. There will be a lot more steps after that but we want to first validate our idea via the crypto enthusiasts of the Ocean community (via the web based version) and after that for other target audiences in Nigeria, South America, India and Thailand (via the mobile version).

For us it is important to proceed in an agile manner and to proceed from working prototype to next working prototype to get validation and feedback from our community.

Foreseen or possible additions?

Big additions in the future will be an independent rating system for the quality of the data via state of the art machine learning algorithms. Another validation system for the quality will be algorithm creation competitions a la Kaggle where the winning algorithms will become part of our marketplace.

The contributors that worked on the data as well as the liquidity provider of our pool will get a share in these algorithms’ liquidity pools to benefit from the results of their hard work.

These algorithms will also enable pre labeling of freshly uploaded data to reduce annotation times (basically annotation becomes quality control) and the sale of pretrained models for transfer learning.

After this is setup and working we will move to the next data type - either text or audio data.

Are there any mockups or designs to date?

Check our alpha web app, alpha data portal and [alpha mobile app](in review for the Google Playstore).

An overview of the technology stack?

Frontend: React.JS (web), React Native (mobile)
Backend: Flask + Python libraries, CouchDB (+ PouchDB)

If the project includes community engagement:

Running the campaign on social media for how many weeks?

We will actively promote each phase with datatoken incentives for participants.

For the webapp we will focus on the Ocean community.

Other?

At the moment we are including more and more community members in the creation of content and the project itself. We work with an agile work model where tasks are distributed to team members and these are then rewarded. This opens our team up to a large number of participants.

Additional Information

Market situation?

Currently there are no end-to-end solutions for the capturing, annotation and machine learning training available on the market. Especially not on this scale and with the involvement of a worldwide workforce.

Any grants or fundraising to date?

We got 106.915 $OCEAN in the funding rounds so far. 7435 $OCEAN from the Ocean Shipyard program and won 5000 $OCEAN in the Ocean datatoken hackathon.

Other costs are bootstrapped by Robin, approximately 180.000 $OCEAN as a personal investment into the project. Most of this sits in the datatoken pool but we also invested a good amount to buy 70 QUICRA-0 from the pool to reward the contributors in our initial challenges.

Time and tokens spending

At the moment we do not reward anyone in the management team for their hours. Over the course of the project, we have been working on this for 11 month now, around 4000 hours were spent by the management team.

We do reward the engineering team and there have been around 10000 hours spent by the engineers on the product so far. This means that we are very cost efficient with an average hourly rate of ~10 $OCEAN for our engineers.

To see what the tokens were spent for, please check our wallet here - all transactions have been made from there. And except the buying of the $QUICRA-0 as rewards and some money for Token Engineering courses (~600$) every other token has been spent on the engineering team.

Social implications of the solution?

If this project succeeds it will change the life of many people in dire situations on our planet. As soon as they get access to a mobile phone and internet they can start making a living. In Venezuela 200$ can feed a family for one month, we expect to be able to reward much, much more than that for a full month of contributions. So it is a global game changer that will enable these contributors to take care of their families right now but also retain datatokens for the future to create a passive income for retirement. Something that was not available to them ever before.
We connected to potential users in Caracas, Venezuela in May 2021 to get a first impression of their needs, limitations and willingness to help. The feedback is going to shape the project to fit to their needs.

What are the competitive advantages of your product over, say, Swash?

Hey @inKin,

great question.

Swash so far creates a big stack of data that is coming out of users browsers.
Our data is annotated and the annotations are verified. So it becomes machine learning ready. This is an important differentiation as we are convinced that the quality and usability of the data is very important.

On the other hand we also enable other data unions to be created based on our software and concept stack. So a data union for browser data based on our software would also be a theoretical option.

Swash is now moving in the direction to create more products around their data. We do that also for our stack by adding more features to it, these then become available to all the data unions in our ecosystem.

And one last difference is that for us the AI algorithms created from the data are very important. We offer the ability to use our Data Portal to train AI algorithms on subsets of our dataset, publish them, or apply published ones to data. These AI algorithms also are represented by datatokens, so they can also be co-owned by a community. Especially the one that has contributed data that is used by the algorithm. This creates an end-to-end cycle from the production, cleaning & annotation to the use in algorithm creation until the final sales of the algorithms. As all of these steps are available it allows us to optimize the alignment of the incentives to make this process and the cycles in it attractive for contributors as well as buyers.

[Deliverable Checklist]
[x] New pitch deck -> https://docs.google.com/presentation/d/18_3V87U0D1f3-yvZ-cFy1Rq0GRjpTR5rm1FDoi1I7YE/edit#slide=id.geff767e975_2_0
[x] Mobile App - annotation feature -> new version released in the Google Play store
[x] Celo network mobile wallet integration -> Celo wallet and Valora wallet integrated
[x] Data Portal improvements -> Metamask handling improvements, pagination issues fixed, multi select improved, worldmap image details fixed

@AlexN