Abstract: In one exemplary embodiment, a method of a computerized media-content recommender includes receiving a user-judgment score based on an historical user-listening data with respect to a media content. A first prediction score for a user with respect to the media content is calculated with a media-content recommender. The media-content recommender includes a first set of prediction parameters. A first prediction error including a difference between the user-judgment score and the first prediction score is determined. At least one parameter value of the first set of prediction parameters is modified with a machine-learning optimization technique to generate a second set of prediction parameters. A second prediction score for the user with respect to the media content is calculated with a media-content recommender. A second prediction error including a difference between the user-judgment score and the second prediction score is calculated.

Abstract: In one exemplary aspect, a sorted list of scored media content episodes is received with a computing device of a user. Each respective media content episode is scored by an iterative autotuning prediction algorithm, and wherein each element of the sorted list of scored media content episodes comprises a value that represents a likelihood of a user listening to the respective media content episode and a reference to a location of the respective media content episode. A number of bytes of a download iteration for each media content episode is determined based on value that represents a likelihood of the user listening to the respective media content episode and an index of the respective media content episode in the sorted list. It is detected that a mobile device is in the preferred network. The download iteration is implemented for each media content episode when it is detected that the mobile device is in the preferred network.

Abstract: In one exemplary embodiment, a method of a computerized media-content recommender includes receiving a user-judgment score based on an historical user-listening data with respect to a media content. A first prediction score for a user with respect to the media content is calculated with a media-content recommender. The media-content recommender includes a first set of prediction parameters. A first prediction error including a difference between the user-judgment score and the first prediction score is determined. At least one parameter value of the first set of prediction parameters is modified with a machine-learning optimization technique to generate a second set of prediction parameters. A second prediction score for the user with respect to the media content is calculated with a media-content recommender. A second prediction error including a difference between the user-judgment score and the second prediction score is calculated.

Abstract: In one exemplary embodiment, a system includes a social media tag-cloud generator. The social media tag-cloud generator obtains a user's social media feed and generates a topic tag cloud. The topic tag cloud includes a weighted key term representing a topic that occurs in the user's social media feed. A media-content source module obtains a first metadata about a first media-content episode. The media-content source module obtains a second metadata about a second media-content episode. The first metadata includes information to identify the first media content episode and to locate the first media content episode in a computer network. The second metadata includes information to identify the second media content episode and to locate the second media content episode in the computer network. A media-content scoring module determines a first score for the first media-content episode. The first score includes a first value judgment based on the weighted key term.

Abstract: A method for evaluating an input system interfacing a human user with an electronic device uses empirically determined bi-action times for users to perform a second input action (e.g., pressing a second key) immediately after performing a first input action (e.g., pressing a first key). The bi-action times (or, more generally, n-action times) are used together with a selected interface map which associates input actions (e.g., pressing keys) to corresponding signifiers (e.g., characters) to calculate a peak expert input rate for the input system. One or more optimized interface maps can be found by combining the evaluation method with any of various optimization strategies. For example, one method for optimizing the input system repeatedly changes the interface map and recalculates the peak expert input rate, while another optimization method calculates peak expert input rates for multiple interface maps in parallel.

Abstract: A method for evaluating an input system interfacing a human user with an electronic device uses empirically determined bi-action times for users to perform a second input action (e.g., pressing a second key) immediately after performing a first input action (e.g., pressing a first key). The bi-action times (or, more generally, n-action times) are used together with a selected interface map which associates input actions (e.g., pressing keys) to corresponding signifiers (e.g., characters) to calculate a peak expert input rate for the input system. One or more optimized interface maps can be found by combining the evaluation method with any of various optimization strategies. For example, one method for optimizing the input system repeatedly changes the interface map and recalculates the peak expert input rate, while another optimization method calculates peak expert input rates for multiple interface maps in parallel.