Generating Narratives from Personal Digital Data: Using Sentiment, Themes, and Named Entities to Construct Stories

摘要

As the quantity and variety of personal digital data shared on social media continues to grow, how can users make sense of it? There is growing interest among HCI researchers in using narrative techniques to support interpretation and understanding. This work describes our prototype application, ReelOut, which uses narrative techniques to allow users to understand their data as more than just a database. The online service extracts data from multiple social media sources and augments it with semantic information such as sentiment, themes, and named entities. The interactive editor automatically constructs a story by using unit selection to fit data units to a simple narrative structure. It allows the user to change the story interactively by rejecting certain units or selecting a new narrative target. Finally, images from the story can be exported as a video clip or a collage.