SBIR Phase II: Social Platform with Machine Learning Moderation

SBIR第二期项目旨在通过创建一个新的无广告、反滥用的社交网络平台来解决社交网站上的滥用和骚扰问题。在现有的社交网站上，网络虐待和骚扰十分猖獗，对女性、有色人种和LGBTQ群体来说尤为严重。然而，可以遏制这种滥用的技术和设计方法无法在基于广告的商业模式中实际实施，因为这种反滥用方法通常会限制广告印象。这个新的社交平台拥有创新的反滥用技术和无广告的新商业模式。在这个SBIR二期项目中，创新使用了新颖的深度学习技术来提供一个新的创新的混合人机调节系统。这个审核系统能够从审核决策中学习，并以在线社区的价值观为基础。此外，这项创新运用了同意的应用理论概念，在内容如何显示和用户如何交互的标准方面提供了开创性的设计。本研究的目标和范围是提高算法效率，开发更多的适度模型，完善适度API，开发用户驱动的模型再培训，与目标终端用户进行协同设计，开发产品前端。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

This SBIR Phase II project aims to address the problem of abuse and harassment on social network sites by creating a new ad-free, anti-abuse social network platform. Online abuse and harassment are rampant on existing social networks sites, and is especially severe for women, people of color, and the LGBTQ community. However, technical and design approaches that could curb such abuse cannot be realistically implemented in advertisement-based business models, as such anti-abuse approaches often limit advertisement impressions. The new social platform has innovative anti-abuse technologies and a novel business model with no advertisements. In this SBIR Phase II project, the innovation uses novel deep learning techniques to provide a new and innovative hybrid human-machine moderation system. This moderation system is able to learn from moderation decisions and is grounded in the values of the online community. In addition, the innovation uses applied theoretical concepts of consent to provide groundbreaking design in how content is displayed and standards for how users interact. The goal and scope of this research is to improve upon the algorithmic efficiency, develop more moderation models, refine the moderation API, develop user-driven retraining of the models, conduct co-design with target end users, and develop the front-end of the product.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.