With more and more video shared on the Web, the practice of sharing a video object from a certain time point (deep-linking) has been implemented by many video sharing platforms. With so many media fragments created, annotated and shared, however, indexing video objects on a fine-grained level on the Web scale is still not implemented by major search engines. To solve this problem, this paper proposes Twitter Media Fragment Indexer, which monitors the Tweet text and uses the embedded URLs pointing to video fragments as the media to massively create index for media fragments. Some preliminary evaluation has shown that media fragments can be successfully indexed in large scale using this system.
This is a presentation from the LIME workshop at ESWC2014.