fbpx

Code: EDI-2020-19-VRT_1

Domain: Internet & Media

Summary

One of the main challenges related to this is making that content easily retrievable, both for the media user and for our editing teams. In this context, chaptering of news bulletins, but also other programs, is often performed related to which topic was discussed.

Proposed by

 

Vlaamse Radio- en Televisieomroeporganisatie, known by its acronym VRT, is a public broadcasting company in Belgium.

Description

VRT produces a large amount of content on a daily basis, ranging from radio content to news articles to video content. One of the main challenges related to this is making that content easily retrievable, both for the media user and for our editing teams. In this context, chaptering of news bulletins, but also other programs, is often performed related to which topic was discussed. Thus, editing teams can easily retrieve parts of interest in a longer episode in order to reuse content for social media posts or making snippets available on video on demand services or podcast services. Typically, metadata representing which topics were discussed in a certain recording is used to achieve this.

 

Data

The challenge has the following sample datasets available for download

Data: initially, 31 videos of VRT’s evening news bulletin with time coded closed captioning in Dutch. This number of videos will be increased later on.

Goal of the challenge: automatic chaptering of episodes according to topic. This consists of automatic topic extraction out of a) time coded transcripts in Dutch, b) interviews in other languages than Dutch for which no transcripts are available. However, topic extraction should also be completely possible based on the audio information and visual information that is presented on the screen alone. Separate transcripts are not always provided for every program at VRT and the provided solution could then also be used to chapter radio content. In addition, a dashboard or query tool in which we can see how many minutes a certain topic was discussed in a certain period of time would be very valuable.

How do we apply?

Follow by Email
Facebook
Twitter
YouTube
SOCIALICON