Francisco Iacobelli, Larry Birnbaum, Kristian J. Hammond
Intelligent User Interfaces
This article the authors are making a new internet news feed system called "Tell me more" a system that looks at new articles and then scans similar articles looking for more. Essentially, when the user is reading a news article the interface looks on the web for more information about the subject and then brings it in for the user. It does this through a system of internet searches that are formatted and then entered into the UI for the user. What the system does is find other related topics and runs them through a series of difference metrics that have been defined by the user, it then can determine if the article in question (the article that will tell the user more) is different enough from the story but still within the same vein so as to provide more information. The system is then run through a series of text analytics that run hueristic algorithms on the text to determine the amount of difference between the two articles and if it should be displayed or not. Upon release of the system the users found that the system was very useful but it was not completely finished. They said they were not satisfied with the results and wanted to make the system more robust so that they can get more kinds of articles searched and hopefully be able to choose from a large list of similar articles on the web for each news story. They also talked about generating a better hueristic that will be able to search articles better and more effectively. The system worked how they wanted it to but they want to make sure that it is more robust and has better searches in the future.
I think this is actually one of the better systems I have gotten to read about. While the system does not have a whole lot of fancy features the ones that they have implemented seem very solid and simple enough that the interface is not overwhelming the user. I would actually be interested in trying it myself to see if the articles that were found were indeed not along the vein of information I was looking for or they were actually relevant. My only problem with this interface is that in the picture of how the UI looks in the article, they make the related articles look like advertisements that are placed on websites and unless I specifically knew they were related articles I might just skip over them. The strange thing is despite rather good user responses from their test they still seemed to think that the system did not work entirely as intended but from my reading of the paper it seems like the system works really well .I would actually be interested in being one of the test subjects for this and seeing if it works and giving feedback. I hope there are more systems like this in the future and I hope that we will be able to test them soon.
This does seem like a good system if effectively implemented, as an individual could read a news story from several different perspectives. This would ideally give the reader several angles to a story, hopefully giving them all the information they need to form an accurate opinion. I do have to wonder, however, if this could lead to a Wikipedia-like reading binge, clicking link after link for hours.
ReplyDeletePatrick brings up a good point there at the end. Did the authors happen to mention anything about this "link binge" phenomenon, either from their own points of view or reported by the test subjects? I suppose ultimately it's up to the user to take responsibility for his addiction to links ;)
ReplyDeleteI agree that this was one of the more useful papers. I think this could work well with research papers if there was an option to only search through academic journals.
ReplyDelete