Scalable fake news detection DApp

Elevator pitch
NuNet is building the scale-able fake news detection application with browser-extension UI, which will enable to dynamically combine best community created software and execute it on community owned hardware -- with the internal tokenomic mechanism powered by Cardano.
Overview
In order to offer a sustainable way to separate the fake from reliable news, the solution has to be evolving and adapting to new sources of information, new methods to detect fake content, user preferences, feedback and the changing media landscape. The basic principle behind the architectural design of the DApp is that it has to allow most, if not all, components sourced from community now and adapted by community in the future. This includes information sources, methods, code, immediate user feedback and hardware for executing the computational workloads.
The proposed NuNet's fake-news-detection DApp integrates and builds upon three ecosystems, which themselves are in heavy development: NuNet (sourcing hardware resources from decentralized community), SingularityNET (decentralized AI and algorithms) and Cardano Blockchain (workflow composition, verification, telemetry information logging on metadata, native token mining based on computational work done).
Goals / differentiation
The concrete goal of this project is to create a scalable fake-news-detection app, which is user-facing and responds to the immediate needs of the community now and in the future. For that reason we aim to experiment and demonstrate tools for harnessing collective intelligence of developer community (OSS code), researchers as well as collectively owned private computing resources. These tools, however will be demonstrated within a usecase of a proposed DApp.
An important goal is to prototype and experiment the implementation of tokenomy of decentralized AI agents (SingularityNET) and computing resources (NuNet) on the basis of Cardano Blockchain, namely KEVM and Cardano Native Tokens Pre-Production Environment. The latter is of most interest and potential, but we are aware that it is being developed right now, therefore will involve a lot of experimentation, hacking and hopefully interaction with IOHK developers. The former means porting current Solidity code and experimenting on running tokenomics on different blockhain.
Besides, but notwithstanding the immediate goals of fake-news-detection DApp, we aim at enabling other (arbitrary simple or complex computational workflows) to benefit from decentralized AI, community sourced hardware and the tokenomic incentives of collaboration. The non-trivial challenge with we embark on is to balance long term perspective of platform(s) development and immediate goals of a concrete DApp. (Therefore, this project is much related to the other challenge of Fund3 -- Developer Ecosystem; however, since one cannot apply to both and it involves DApp creation, I decided to submit here).
Idea in more detail and breath
The idea is to develop a Chrome/Brave extension which will display the probability that a web page contains fake news. The probability displayed to a user will be a combined measure of fake-news-detection algorithms, running on the back-end of the app. The back-end will be build on top of NuNet architecture, which runs arbitrary algorithms (currently containerized as SingularityNET containers) on any hardware resources onboarded on NuNet platform. Currently these resources are devices of team members dedicated for testing and experimenting.
Both NuNet and SingularityNET are built with economic logic / tokenomic architecture incentivizing software developers and hardware owners to provide their expertise and latent computing power for executing computational logic and workflows that have economic and social value. We use blockchain tech to power the transactional logic and interaction between different software components (as in service mesh) and interaction between software and hardware. We want to prototype the Cardano integration due to its novel and highly relevant features, such as metadata on blockchain, multi-assets and others.
Most broadly and in the long term, NuNet aims to create a self-organizing, global economy of computations and hardware resources. In some sense the idea is reminiscent of Ethereum's 'world computer', but much more decentralized -- i.e. not having a single eventually consistent global state. It is also related to the Cardano research efforts named 'Tartagalia', which deals with private, ephemeral computational events with the help of special purpose multiparty computation protocols. (however, there are no mentions of updates of this research direction besides mention in Cardano white paper (https://why.cardano.org/en/introduction/cardano-computation-layer/).
Architecture
Concretely, the DApp consists of:
* Chrome/Brave browser extension for UI/UX interacting with the backend;
* The back-end consisting of a decentralized service mesh of OSS fake news detection algorithms running on different machines and providing inputs for combination of the fake news probability of the content displayed to a user;
* Telemetry API, collecting actual compute usage by each call to the app and distributing computing resources;
* Tokenomics API, providing payments to compute owners, AI algorithm providers, native token mining and possibility of economic transactions originating from a user;
(see https://nunet.gitlab.io/fake-news-detection/catalyst-fund3-proposal/architecture/ for more details and preliminary diagrams)
Estimated tasks / workload
(see https://nunet.gitlab.io/fake-news-detection/catalyst-fund3-proposal/planning/workload for more details):
* Network protocol engineering and communication of private devices behind NATs;
* Development/adaptation of the NuNet adaptor which is a device that exposes its resources for running services;
* Development/adaptataion of the NuNet network organizer which is a device that accesses and coordinates other devices running services;
* Rehashing Telemetry and Tokenomics APIs for making it compatible with the DApp and Cardano blockchain;
* Adaptation of individual fake news detection algorithms (until the tokenomic incentives are in place and the platform working as intended we have to do this ourselves);
* Porting existing Solidity code to KEVM and deploying tokenomics logic on the its devnet;
* Research and experimenting with porting tokenomic logic to Cardano Native Tokens Pre-production Environment);
* Technical coordination of DApp development in line with the availabilities and plans of Cardano, SingularityNET and NuNet ecosystems.
Budget / resources
(1) Two developers with C, python, linux admin skills service mesh engineering skills -- $11,200, combined total 14 weeks;
(2) Database / data structures engineer -- $ 3,200, 4 weeks;
(3) Blockchain / Solidity / Cardano integration -- $ 19,200, combined 12 weeks;
(4) Hardware hacking, technical advice and research - $4,800, combined 8 weeks;
(5) Strategic lead/advice, ecosystem integration and coordination -- $8,000, combined 5 weeks;
Total $46,400
Time to completion: 12 weeks;
IP policy
All platform code developed by the team is open source and will be made available with most permissive OSS licenses (MIT/ Apache);
Success / KPIs
* Achieved technical goals of integrating Cardano to NuNet / SingularityNET;
* Working telemetry, tokenomics API and dynamic service mesh in the background;
* 5-10 individual devices on NuNet network running the app;
* 2-4 OSS algorithms running on decentralized hardware;
* Publishing extension to Chrome Web store;
* Sub-second fake-news-detection of digital content;
* 100 - 250 independent downloads with subsequent backend calls;
MVP
The previously created Minimal Viable Product which will be turned into the full app by the proposed project:
* short video: https://youtu.be/eJGrPORbuRw
* longer description: http://bit.ly/devupdate_2021-1

46400