| 摘要 |
Method and system for processing video data and text-based media items, with ingesting (300) video data, segmenting (310) the video data into segments (280); extracting a first feature vector representation (315, 510) of the segments; annotating (235, 320, 520) each of the segments with metadata instances (273), annotating (235, 520) the segments with metadata to an annotated event (530); ingesting (300) text-based media items (610) from social media sources; extracting (620) a second feature vector representation (625) by converting input of the annotated event (530) and text-based media items (610), the second feature vector representation (625) including at least one content feature (620c), generating scores that indicates whether the text-based media item (620) refers to the annotated event, generating for the content feature (620c), using a feature specific sub-function, with each of the sub-function's score being based only on the information extracted in that particular feature. |
