Interfaces for Reading News and
News Coverage of Events
Earl J. Wagner (ewagner@umiacs.umd.edu)
University of Maryland College of Information Studies (iSchool)
Associated Press Study of News Readers (2008)

Back Story


Big Picture



News is “here and now.
If you want background, it's up to you.”
"Breaking news... is kind of annoying sometimes. I don't like to
get bits and pieces of information."
I “[do] not want to be fed bits. I want to know all the details at
once”
Story Resolution

Sports and entertainment news preferred because they offer “a
beginning, middle and end, or clear next steps”
Associated Press Study Concluding Vision

The news “audience is looking to be transported to
relevant information no matter its source. Where online
consumers once surfed and bookmarked news sites, users
now wonder why a logical trail through the news
can't simply unfold, link by link, across a multitude of
sources.”
Brussell Demo
Brussell Features

Reads expecting to find
News Situations:




Stereotypical sequences of
related events in the news
Organizes news situation
information
Supports navigation to
background coverage
Spans various news
sources
How Brussell Works


Retrieves news articles daily via RSS feeds
Extracts events and facts from text with patterns:
e.g. “X raises its bid for Y to Z”





Notes news and article source
Creates record for event
Merges into set of all tracked situations and participants
Links with user’s news reading
Answers user questions
Information Extraction


Mature technology today
Reads news to answer questions



Oct 31 2004: “Oracle Raises Its Bid for PeopleSoft by $3,
to $24 a Share”
What happened?
Who was involved?
cnet News: http://news.cnet.com/8301-27076_3-20065720-248.html
Story Understanding


Still developing research area
Answer in-depth questions






What happened in between?
Why did that happen?
What else could have happened instead?
Why did he do that?
What did she expect might happen?
Learn story type patterns
Future Work


Learn motivations of people in stories
Discover types of answers, visualizations useful to news
readers



What questions can be answered automatically by software?
Supported by National Science Foundation grant IIS0325315/004.
Earl J. Wagner


Email: ewagner@umiacs.umd.edu
Website: www.umiacs.umd.edu/~ewagner