NewsBlaster is an experimental news prototype authored by a team from Columbia University’s Computer Science Department which processes news stories and events using multiple sources. It then uses these to create summaries, externally rank them and AUTHOR lead storylines automatically:
The computer, a natural language processing algorithm, uses artificial intelligence techniques to cull through news reports published online, sort them, and summarize them. It does not simply lift sentences to use as the summary. Instead, Newsblaster uses natural language processing techniques to read what is written in published news reports.
It interprets the importance of different facts, based on its own news judgment, reflecting factors such as where a fact is mentioned in the published reports, how often it is repeated across reports dealing with the same event or subject, and the news value of those individual facts such as how many were killed or injured or how much damage to property occurred.
An article in the Online Journalism Review goes into detail on what implications may arise from using such a system in the future:
A more serious concern [...] deals with the basic nature of a machine writing leads. Newsblaster seems to make things somewhat generic or more conservative, especially when summarizing reports over several days.
This can take away the editorial edge or nuance that a reporter or editor might use to make a lead or report powerful. Summarizing news over several days in this approach results in a certain staleness.
on Friday, March 29th, 2002 @ 9:02 pm
§ Software