Abstract: User behavior metrics may be inferred from user interactions with a client application through application programming interfaces (APIs) that perform backend services User behavior metrics may be inferred from user behavior that is correlated with data streams derived from traffic created by user interactions. Such user behavior metrics may be used to adjust client application offerings to increase the utility of the client application and function independently as a product to third parties.

Abstract: Generally disclosed herein is a mechanism to provide music and radio station recommendations based on observing radio and music listening activities of a user over time. In some examples, information about the radio stations receivable by a user device and programs broadcast by such radio stations are collected over time. Recommendations for different music and/or radio stations may be generated based on a single song or radio program that the user listens to. The recommended music and/or radio stations may be updated based on program information related to the radio stations receivable by the user device at a specific time at a specific geographic location.

Abstract: Generally disclosed herein is a mechanism to automatically recommend radio stations, audio content, or URLs for additional audio content based on analysis of the user's current and past listening activities and the program information of the radio stations received by the user device. The user's preference for particular topics or themes for particular additional audio content can be determined based on the user's historical or preference data. Such topics or themes can also be determined using machine learning models in real time or near real time while the user is listening to particular audio content. Based on the time and day when the user's listening activities occurred, time-specific recommendations for other audio content or radio stations can be output, such as through a display of the user's device.