This presentation outlines how BBC News Labs is currently working on entity extraction on BBC News content.
It looks at the challenge of how BBC News is going to leverage it's USP of Storytelling, and it's famous purpose - "to INFORM, EDUCATE & ENTERTAIN" globally.
With the millions and millions of "things" that are in our content, how do we discover and connect these things?..
It also mentions 2 key BBC News Labs projects "JUICER" and "#newsVANE", and promotes the http://newshack.co.uk/newshack-ii/ event on May 1st in Dublin & Glasgow.
Call Girls Hsr Layout Just Call 👗 7737669865 👗 Top Class Call Girl Service Ba...
The Newsroom of Things by BBC News Labs - for ISKOUK "Taming the News Beast"
1. Powered by BBC Connected Studio
The Newsroom of Things
ISKO UK “Taming the News Beast”
April 2014
matt shearer – innovation manager
@BBC_News_Labs
2. ABOUT US
“Driving Innovation in News”
NEW TECH AND
DATA
OPPORTUNITIES
NEW
JOURNALISM
FORMATS
EXPLORE
VIA
PROTOTYPING
Part of BBC Connected Studio.
The BBC‟s open innovation programme.
17. Stuff & Things (all 6 types = holy grail)
1.Verbatim transcript (to time) “…where she says „damnit!‟”
2.Contributors (face and voice) “who‟s in this segment?”
3.Objects (audio & image recog) “tank or a elephant?”
4.Scene geolocation “this looks like Bangor”
5.Topics mentioned (people, places, orgs,.. Storylines*)
6.Actions & Events (non verbal) “people laughing, kissing”
* Jeremy is telling you in a few mins…
21. What is it?
• News Content
• Tagged with Linked Data concepts
1
Get
Content
2
Extract
Concepts
3
Match to
DBpedia
4
Annotate
Content
5
Push to
Triplestore
The Juicer
6
Expose
via
API
27. We save a lot of manual tag time
tags in Juicer 5,700,000
seconds per tag guestimate 10
total seconds on tagging 57,000,000
Mins spent 950,000
Hours 15,833
working days tagging 1,979
working years tagging 9
NB – this is rough, and just for illustration.
28. Next :
Window on the Newsroom
(+AV with transcript generation and speaker recognition)