Paper Reading #17

Comments:
Comment #1
Comment #2
Reference:
Title: Aspect-level news browsing: understanding news events from multiple viewpoints
Author: Souneil Park, SangJeong Lee, Junehwa SongVenue: IUI 2010
Summary:
A large issue with news reporting is the bias of the author or the news station they work for. A result of this is that the general public does not get an accurate representation of an issue because most readers do not thoroughly search a subject. Instead readers go to one or maybe two providers and get an article that emphasizes certain aspects rather than the issue as a whole. If readers went to all sources and compared the results, they would get a much more accurate view of the issue at hand.

The purpose of this paper is to provide a solution to this problem. Google News, and other things similar to it, only provide event level browsing. This system would organize events based on the aspect that paper emphasizes. For example, each different aspect of the event and how it was covered would be represented in a different corner. Bias in reporters can be seen by the different topics or aspects that they choose to emphasize, as already said, so organizing items in this fashion is not unfeasible. Aspect-level classification is achieved in two steps: aspect extraction and article classification. Keywords play a prominent role in each of these things, but things such as meta-data also are important. The group had high success with their aspect classification system.
Discussion:
I think this is another good idea for news browsing. This one builds off the previous article on the list of readings, which I was actually assigned to read after this one. Reading the other one first made this article much easier to understand. Similar to the other article, this article had a lot of formulas and technical talk despite the somewhat easy concept. As a result I thought it was harder to read than it could (or should) have been. I believe that people, for the most part, are not interested in the technical "how" these things are made, but rather what it does and if it works. I think this system will work given the data that they collected and will be yet another improvement to Google's empire. I'm not sure if their idea of spacing, placing things in the corners, is going to work. People like to look in one area for all their information. Placing things all across the screen is not good management of the screen space, in my opinion. However, I do think that ultimately this will work, especially since it has Google backing. Rule of thumb for me: Never under estimate Google.3