Assignment 16

Jon Marbach

1. What did you find interesting about the article?
   
   

   
   What I find interesting is that this article has very broad, immediate, practical relevance: everyone relies on other people to find new books, movies, music, places to hike, ways to get somewhere.... I hadn't considered that recommendation was such an integral part of everyday life, especially implicit recommendation. The discussion of serendipity was important in bringing balance to the approach and keeping computational recommendation organic - even with an automated system, you would still want to get that feeling of "stumbling across" something.
   

   
   




2. What did you find not interesting about the article?
   
   

   Nothing really - good article!

   
   

   Ok, on second thought. There's a "trust" issue that the article doesn't address. I was just thinking through a little "what-if" scenario: Amazon recommended me the movie "Ghost World". I'm not sure whether I'll like it or not, but I'm leaning toward checking out the movie. What happens to my trust level in that recommender if I don't like it? Will I/Should I ever take a recommendation from Amazon again? So in a collaborative filtering scenario, there might be a need to track the results of particular recommenders' recommendations.
   




3. What do you consider the main message of the article?
   
   

   
   We face an ever increasing number of choices and information about those choices. Computational recommendation and recommendation support systems that deliver personalized results based on multiple factors and sources while maintaining privacy will help people manage these choices.
   

   
   




4. Describe how you rely on non-computational recommendation systems in your daily life?
   
   

   
   I constantly rely on recommendations. Besides the common ones (music, movies, books) that I'm sure everyone mentioned, I use recommendations from restaurant workers quite often: "Which do YOU prefer, the Pad Kra-Prao or the Pad Woon Sen?" I also get recommendations from people on where to take visitors from out of town.
   

   
   




5. Have you used any of the systems mentioned in the article (e.g., slashdot)? What motivates you to use them?
   
   

   No. I'm info overloaded as it is!

   
   




6. Choose one of the systems mentioned (see URLs in the article and in the slides for the lecture) and briefly describe your experience with the system?
   
   

   I breifly checked out ReferralWeb - an expert and research paper finder/linker for the computer science field that lets you visualize connections between experts as a graph of interconnected nodes. I looked up Dr. Fischer and he came up linked with Jonathan Ostwald but the applet showed a stronger linking (represented by a thicker line) with Andreas Lemke and Kumiyo Nakakoji.

   
   

   I also decided to check out Amazon.com and see what it would recommend as good matches for some of my favorite movies. The first thing I saw when I went to Amazon is a link that says "sign in for personalized recommendations", and I realized how important recommendation has become on Amazon. I didn't sign in but just went to check out what it would say for "Other people who bought this also bought...".

   
   

   Nothing really ground breaking came up, although I was surprised to see that searching for somewhat "indie" movies as "Pi", "Buffalo 66", and "Amelie", Amazon recommended "Charlies Angels" and was offering $25 off on a Bissel Steamer Carpet Cleaner! Although, I do have to admit that Amazon is suggesting "Ghost World" which doesn't really sound like a movie I'd like but I might take Amazons word for it and check it out. [That brings up a thought that I'll add to the What You Didn't Find Interesting section.]

   
   

   The really interesting section was "the page that you made". Although I wasn't signed in, and without cookies, Amazon had tracked everything I searched for (I would guess by IP) and can present it as a dynamic page to "help you find what you want and discover related items".