Systems and methods for improved coverage of input media in content summarization

摘要

The disclosed technology includes techniques for improved content coverage in automatically-generated content summaries. The technique may include clustering a set of input content, determining diffusion for each cluster, and selecting representatives of each cluster to optimize other secondary metrics. Various types of input content may be used, including groups of images, video clips, or other multimedia content. Contiguous content may be manually or programmatically divided into discrete portions before clustering, for example, a lengthy video divided into a number of short clips. In some implementations, the disclosed technique may be implemented effectively on a mobile device. In other words, the processing required may be computationally feasible for execution on a smartphone or similar device.

主权项

1. A method comprising:
receiving, at a computing device, content to be summarized; defining, by the computing device, and based on the content, a plurality of content segments; defining, by the computing device, and based on a determined similarity between content segments from the plurality of content segments, a plurality of content segment clusters, each content segment cluster from the plurality of content segment clusters associated with one or more content segments from the plurality of content segments; determining, by the computing device, a plurality of representative content segments including a representative content segment for each content segment cluster from the plurality of content segment clusters, wherein, for a first content segment cluster from the plurality of content segment clusters, a first representative content segment is determined by:
calculating a diffusion value for each content segment associated with the first content segment cluster, the diffusion value calculated based on a quality value associated with the respective content segment and a proximity of the respective content segment with other content segments associated with the first content segment cluster; andresponsive to calculating the diffusion value for each content segment, selecting, as the first representative content segment for the first content segment cluster, a content segment with a highest calculated diffusion value; selecting, by the computing device, and to be included in a summary of the content, one or more representative content segments from the plurality of representative content segments, wherein a magnitude of the one or more representative content segments exceeds a predetermined threshold; and outputting, for display, the summary of the content wherein the summary of the content comprises the one or more selected representative content segments.