SUMMA Jetski
Multilingual Search on Live Feeds
Project Lead and Developers
SUMMA
Scalable Understanding of Multilingual Media
9 Project Partners
NLP Technologies
Media Processing
Machine Translation
Automatic Speech Recognition
Punctuation
Language detection
Relationship Extraction
Named-Entity Recognition
Deep semantic tagging
Knowledge base building
Fact checking
Sentiment Analysis
Summarisation
Clustering
Topic detection
QA
Language Understanding
Transcription with Arabic ASR
Translation with Arabic MT
Storyline Clustering: US Open
Multilingual (Spanish) Storyline Clustering
Named-Entity Recognition: Iraq
What can the BBC do
with these technologies?
SUMMA Jetski
Enable journalists to search and discover content
from multiple live broadcast sources,
by searching in vernacular transcripts.
Demo! The video playback only works from within corporate proxy…
With working video…
Learnings
Learnings
People have said...
“Very likely to use in production”
“Would save 30 minutes to an hour every day”
“Would save time when compiling reaction pieces”
Learnings
💪 💸💸💸💸
Architecture
Learnings
💪 💸💸💸💸
SUMMA ASR
Google Translate
"Russia will stop raping NASA astronauts on the ISS"
"Russia stops shipping American astronauts to the International Space Station (ISS)"
🙀!?
Learnings
💪 💸💸💸💸
SUMMA #newsHACK
Editorial Tools for Global News
Bonn, Germany
9-10 October 2018