Discovery, especially non-text discovery, is challenging. When looking for new music or guided meditation tracks, users might not know exactly what they want, but can identify a general preference for a particular style or theme. Clean, user-generated tags can improve discoverability on websites like SoundCloud. However, free-form tags may degrade discoverability due to misspelled words or personal tags. The ConceptNet5 dataset provides a solution by helping users select and search by relevant tags in multiple languages. This dataset is used in conjunction with the SoundCloud API to recommend tags for user-generated items, improving discoverability and user experience. By leveraging graph databases like Neo4j, developers can create powerful recommendation systems that utilize relationships between concepts and tags to suggest relevant content to users.