Welcome to the "Development of Generative AI Moderator for Trusted Web3 Community” challenge, sponsored by Fujitsu and TC3!

This challenge aims to create a safe and trusted society and community by developing a communication support function that protects social media posters and improves the community. Building a trustworthy community on social media is an important factor, but online communication has various challenges. There is potential to solve these issues with the support of AI technology. To this end, we seek to harness the capabilities of the rapidly developing Large Language Models (LLM) to power an AI moderator that analyzes social media content. If any problems are detected, the system will proactively warn or advise the poster, thus addressing these issues. As a first step, we will conduct an ideation and development challenge to design new perspectives for analyzing social media posts and develop the program necessary to perform this analysis.

This challenge is the third installment of the Web3 × AI Innovation Challenge series, hosted by Fujitsu.

Important Notes:
Please note that participation may be subject to certain geographic restrictions, detailed further below in this document.

Background

2.1 Fujitsu stands with Web3 and AI

The World Economic Forum's Annual Meeting 2023 theme "Cooperation in a Fragmented World" drew attention as a sign of the transition to a decentralized structure. In response to these changes, Fujitsu has developed multiple Web3 technology based systems.
Furthermore, we believe that the evolution of Web3 and AI can help to address the negative aspects that have been brought by ICT technology and play a key role in the creation of a new social economy.

Fujitsu is a strong advocate for open innovation with a history of commitment to the concept.
In the Web3 and AI field, Fujitsu provides 2 platforms to accelerate open innovation: the “Fujitsu Research Portal” and “Fujitsu Web3 Acceleration Platform”.

2.2 Fujitsu and Federated Social Networking Service(Federated SNS)

Fujitsu's Vision

• Fujitsu is promoting activities around the Web3 Acceleration Platform (AP) to accelerate the societal implementation of Web3 by allowing for various app developments to be tried and tested. However, a new challenge identified through activities on the Web3 AP is how to invigorate collaboration among users and organizations.
• To address this issue, we are exploring a method that combines decentralized social networking services (SNS) that allow for easy communication between self-sovereign organizations with AI agents that can foster community activities aligned with participants' goals.
• Through this approach, it becomes possible for agents and humans to collaborate on social networks, and we aim to realize a world where humans and AI agents autonomously interact, nurturing human networks, and promoting effective activities in cooperation with others according to individual objectives.

Federated SNS

• Decentralized social networking services like Bluesky are characterized by their non-centralized nature. Unlike platforms such as X or Facebook that are controlled by a single organizational entity responsible for data management and order maintenance, these are managed and operated by the community as a whole.
• As an implementation of decentralized SNS, there is the AT Protocol provided by Bluesky as open-source software (OSS), and Fujitsu is developing an SNS that extends this protocol. Fujitsu is utilizing its own technology to innovate and improve upon its unique Federated SNS.

The Need for AI Moderators

• Online communities, including social networking services (SNS), come with several communication challenges. For example, unpleasant or harmful posts can negatively affect other users, or unintended leakage of personal information may occur due to inferences drawn from post content, victimizing the poster.
• One approach to solving these challenges is through content moderation, which involves monitoring and managing post content to maintain order within the community. This could mean deleting offensive content or alerting users about the risk of personal information leakage before they post.
• Content moderation often requires nuanced judgment that has traditionally been performed by humans. However, the task of continuously monitoring for offensive content can cause significant mental strain for content moderators. For instance, there have been reports of Facebook moderators developing PTSD-like symptoms due to their work conditions. (Reference: https://www.theverge.com/2019/2/25/18229714/cognizant-facebook-content-moderator-interviews-trauma-working-conditions-arizona)
• Therefore, reducing the burden on content moderators is an essential issue that necessitates the development of automatic systems capable of making highly accurate determinations.

2.3 Introduction of Fujitsu Research Portal

The Fujitsu Research Portal is an environment that gives users the ability to quickly test Fujitsu’s technologies by providing APIs and web applications of tech components for free.
We're delighted to invite Topcoder members to take part in the Fujitsu Research Portal — an innovative platform that grants access to Fujitsu's cutting-edge Web3 and AI technologies.

For this Challenge, we will be using Fujitsu's Federated SNS to acquire data. The Federated SNS can be accessed through the Fujitsu Research Portal.
In addition to this challenge, we hope you will find the technologies and resources available on the Fujitsu Research Portal to be valuable for exploring your ideas.
We also welcome your feedback on the Fujitsu Research Portal so that we can continue to improve it.
Let's start having fun with the Fujitsu Research Portal!