Conversation Data Points
Our algorithm is inspired by DNA alignment algorithms from genetics. It measures correlation across textual documents using python data science libraries and the Natural Language Processing Toolkit. Our AI calculates a precise match, in milliseconds, analyzing a single phrase of free text to route each user to the best active peer support chat with other concurrent users who relate to their struggle. As the conversation flows, AI is at work again, selecting hyper-targeted resources from our knowledge base for our moderators to share with the group in real-time.
With our rich and diverse data points, we’ve used neural networks to create a custom mental health corpus that doesn’t exist elsewhere, empowering us to match users into the right peer group with precision. Our training data goes beyond text; we use a variety of signals to actively learn and improve the process. Users rate our chats and moderators as they experience various peer support groups and provide qualitative feedback, creating layered learnings which continuously fine tune our models.