Joho the Blog
An Entry from the Archives

« Sentimental analysis || Back to Blog | Confession by tense »

January 06, 2007

Automated broadcdast news

The Wall Street Journal reports on NewsAtSeven.com, a beta site from Northwestern University's Intelligent Information Lab, that automatically aggregates a news broadcast, hosted by an avatar named "Alex." (Actually, it's Alix from Half-Life 2. In fact, the the whole presentation seems based around Half-Life machinima.)

As it stands, it's hard to tell how impressive the results are. The voice is clearly doing text-to-speech conversion, but I have no idea idea where the text is coming from. The editions I sampled are short and coherent, but for all I know, the system just picked a topic—e.g., Will Ferrell's new movie—located a news report on the Web, and read it. The graphics are appropriate, but, again, I don't know that they're any better than what you'd get if you took the top results from a search for "Will Ferrell" at YouTube, Google Images, etc. So, I don't know what's supposed to impress me. I'm not saying it's not impressive. I just don't know how.

(Yes, the project could use a good marketer.) [Tags: newsatseven news everything_is_miscellaneous ]

Posted by D. Weinberger at January 6, 2007 12:47 PM


Comments

Back in the first dot-bubble, some big company (HP, maybe?), came out with a news-reading avatar called Maya that always bubbled:

"Hiya! I'm Maya!"

Posted by: adamg | January 6, 2007 01:55 PM


hello, here
http://quinta.typepad.com/Temp/news10mag.mov
you can see a combination of an avatar (that's already used by Telecom Italia's customer care on massive numbers of users, realtime rendering) with a search engine specialized on news (www.presstoday.com)

the agent used a summarizing engine in order to reduce the length of the news read by the avatar

the background image could be authomatically changed based on images in the articles indexed and summarized.

the idea was to have to sliding controls in order to increase/decrease the number of news presented and the length of the summary (this part was not actually implemented in the user interface)

most relevant news were extracted based on auto-correlation of news in the corresponding sections of different newspapers (better, "sources").

the technology is all there; unfrotunately in Italy there were no customers...

Posted by: Stefano Quintarelli | January 6, 2007 05:18 PM


As Stefano has pointed out, the text manipulation is the only basic technology that's been developed here. The rest of it is impressive only as a mashup. But aren't mashups often impressive in their conception?

Posted by: Matt Norwood | January 8, 2007 01:33 PM


Post a comment

Guidelines for Commenting

Basically, you can say what you want. (Click here for the fine print.)

If you haven't left a comment here before, your comment may be put into a queue for me to approve. Sorry for the delay. Blame the damn spammers.