Beyond Recommender Systems: Helping People Help Each Other |
---|
Payal Prabhu
Assignment 16
Due: Wed, April 3rd
As usual, I have a question: In their discussion of weblogs, I believe the authors did not address the issue of bias in recommendations when the weblogs are populated with annotations from a single or a few individual(s). Could you please go over this problem in class and how bias can be computationally reduced or eliminated (other than resorting to the obvious option of increase the number of recommenders).
The main message of the article is that people are faced with endless choices when searching for information on the Internet and the WWW, and in order to make thier lives easier, Recommender Systems evolved so that "people could share their opinions and benefit from each other's experience". By implicitly or explicitly collecting corpus, these systems enable people to help each other in making informed decisions.
For things like movies, I look up reviews online at IMDB BUT before I actually go to watch a movie, I ask some of my friends if they have seen the movie and what they think about it. I have different categories of "movie watching friends" (commercial, independent, folk, etc.) that I go to selectively based on the genre of the movie I am interested in watching.
Also, while deciding between graduate schools, I emailed/talked-to many of my friends, current students at those schools, and professionals so that I could get a more "personal" recommendation other than the regular NRA rankings on the WWW.
Yes, I have used Slashdot, CDNow, Amazon.com, and some other systems that were mentioned in the article. As indicated in the article, I too have to rely on the recommendations posted by other users to decide between scores of options that are presented to me when I conduct a query on the WWW.
I am a recommender as well as a seeker for Amazon.com system. I initially acted only as a user since I was not aware of the "level of expertise" required to become a recommender. After having read and successfully used recommendations about books from other "regular" buyers, I felt the need to input my thoughts on books that I had read; a responsibility to "give back" atleast some percentage of what I was getting out of the system.