Multi-modal and multi-lingual child safety application

儿童对社交媒体的广泛使用最近引发了与网络欺凌和其他对他们福祉的威胁有关的问题。该项目的主要目标是在图像、视频、短文本和来自在线交流事件的其他可用元数据中确定有关童年行为和生活环境的详细信息，当情况发生时，这些元数据可以用作成人干预的一部分。这个项目建立在之前NSERC CRD拨款的基础上进行的研究合作，这对公司真的很有用：我们开发的分类算法被集成到其产品中，并创造了9个新的工作岗位。我们建议增强我们的预测儿童行为和心理状况的模型，除了文本交流之外，还包括可能有害的图像和视频，以及从儿童设备中提取的元数据，这些元数据与他们的在线交流相关。我们还提出了一种新的研究路线，为每个孩子实施个性化的模型，而不是“一刀切”的模型。我们将开发一套算法程序，从而对儿童有发育威胁的经历或表现出有问题的心理状况进行可能性估计。我们的算法将使用我们的合作者的指导，包括医疗保健和技术方面的专家，以及专门研究儿童行为问题的心理学家和精神病学家

The widespread usage of social media by children has recently raised problems related to cyberbullying and other threats to their well-being. The main objective of this project is to identify details about childhood behaviours and life-circumstances within images, videos, short pieces of text, and other available metadata from online communication events that can be used as part of a reason for adult intervention when a situation arises. This project builds on a research collaboration conducted on a previous NSERC CRD grant that was really useful to the company: the classification algorithms that we developed were integrated into its product and 9 new jobs were created.We propose to enhance our models for forecasting behavioural and psychological conditions for children to include, in addition to textual communication, images and videos that could be harmful, as well as metadata drawn from children's devices relevant to their online communication. We also propose a new line of research to implement personalised models for each child, rather than a "one size fits all" model.We will develop a set of algorithmic procedures resulting in likelihood estimates that a child is having a developmentally threatening experience or manifesting a problematic psychological condition. Our algorithms will use guidance from our collaborators, including experts in healthcare and technology, and psychologists and psychiatrists specialised in child behaviour issues