Title: Aspect-level News Browsing: Understanding News Events from Multiple Viewpoints
Authors: Souneil Park, SangJeong Lee, Junehwa Song
Presentation Venue: IUI’10, February 7–10, 2010, Hong Kong, China
Summary:
This paper looks to solve bias in news media by providing a classified view of news articles with different view points. This functionality is called Aspect Level News Browsing and will allow the reader to read different points of views of important news article so they obtain a better understanding away from the bias that comes with every news article.
The problem with this idea is classifying each article into categories for the reader. This is done computationally by reading through the articles and parsing phrases that occur commonly in certain categories. There are few different ways of clustering articles together to form a specific category.
This is done by a method called Framing cycle-aware clustering. This type of clustering allows articles to be sorted in a head-tail model. The head-tail idea is articles extremely relevant to the category are placed in the front and articles with a smaller match are placed towards the end.
By clustering many articles together from different news sources the user can mitigate media bias. The reader can formulate their own balanced viewpoint from reading the different contrasting aspects.
Discussion:
The idea of this is very interesting from a computational standpoint. I think this because it seems like it would be extremely difficult to accurately organize hundreds of news articles automatically. The ideas they use for sorting through the articles and clustering them together is pretty cool. I didn't go into a lot of detail about the parsing and sorting of the articles but if you want the technical details you can read the paper.
I like the idea of being able to read multiple news articles to try and get rid of bias because bias exists in everything we read. Even though bias cannot be completely removed, from reading multiple articles on the same news topic the authors hope the reader can realize different angles about the topic and have a full understanding. The only problem I have with reading multiple articles is I would have to be extremely interested in the topic to read about it more than once but those willing to read that much this kind of application would be useful for understanding news topics fully. I think its also worth noting a convience factor with this application, all of the articles are placed together so you do not have to go to different sources to view the same topics.
I think this is an interesting idea. I agree that it is nice to get rid of media bias. I think that if popular it may fluster many media organizations. I don't know that they would retaliate in any way. It would be interesting to see the different classifications of bias the design would group them in.
ReplyDeleteFor computation, I think with a big enough server farm the task shouldn't be beyond a system's capabilities; then again, the server farm size needed might be too expensive for an experiment like this.
ReplyDeleteI think that this would fail because people are lazy and won't spend the time reading multiple articles. For something like this to work, a program would have to interleave opinions into a single article.