Event Alignment for Cross-Media Feature Extraction in the Football Domain

摘要

This paper describes an experiment in creating cross-media descriptors from football-related text and videos. We used video analysis results and combined them with several textual resources - both semi- structured (tabular match reports) and unstructured