Abstract: A computer-implemented method for training a machine learning system to classify content using multi-source human annotations. A pre-set of classification tags is predefined for classifying content according to content category, wherein the pre-set of classification tags is used for training the machine learning system. Content is received from a content poster with at least one poster-supplied tag. The machine learning system classifies the content and assigns at least one machine-learning-system-assigned tag. A content viewer provides at least one viewer-supplied tag in response to determining the content is not correctly classified. The content and associated tags are transmitted to a moderator interface where a human moderator reviews, approves, edits, or deletes tags from all sources.

Abstract: Disclosed embodiments provide techniques for computerized moderation, authorship recording, and distribution of social media content. Moderator-supplied tags are associated with content and supplied to a machine learning system as training data. Using blockchain, authorship is authenticated and can be converted from anonymous to non-anonymous. Collaboration among authors on content is supported with authorship lists that can contain a mix of anonymous and non-anonymous authors. Contribution limits are established to determine royalty payments for sale and rent of content.