Project Background

  • Big data era
  • Tell the data as a news
  • Rapid news production
  • Short Video (easy to share)
  • Propose a novel way to arrange the story as a short video from its data-driven

General motivation

Is it possible to create a short data-driven video by automatically that can tell its story with its movement like the Hans video?

The specific problem

  • There are many events, which are suitable to select for representing?
  • When should the data be pointed with the highlight?
  • How to determine an action (effects) to the data movement in the short video
  • Moreover, How can the user arrange their story?

Proposed Topic

Propose a novel way to extract the movement behaviour of each data (hashtag) and then tell the story from the data (hashtags) behaviour and export as a short video that is easy to publish

Research Content

Label the level of the data cluster, the level can be used to determine the movement of the data (hashtag), the label can tell;

  • which data (hashtag) should be highlighted
  • when the video should be paused (based on highlighted data condition)
  • The transition of the data (hashtags) that can adjust the speed transition
  • The critical period (event) of the whole timeline

Our method proposed the way to get the level of data by using K-means clustering

From previous can sum-up as the Story Arrangement comes from;

  • Highlighted Data (Hashtags)
  • Pause
  • Convert Transition to Speed