Hello recommender enthusiasts,
First of all, happy new year and shall your recommendations be useful all year long :).
Now to the topic, few weeks ago I noticed a linkedin post from my colleague (here is the link linkedin post ) which took a view from the pure user point of view on how inconvenient a filter bubble created by recommender system can be. Small interim, filter bubble is a term used in RS other name in public would be "echo chamber".
I liked the post because it came from someone who is not in RS academia and is mostly a user exposed to recommender system around the internet. The view, as you can read in the post, was rather negative about the effect of RS on filter bubble. The post even mentioned that it would be cool to present user with stuff they would deliberative dislike (his case was in news domain). It got me thinking that taken aside by bias to like recommender system I can see how users could find opposing news articles to their own views useful just to purely wanting to stay more objective with their opinions. Obviously not in every domain but in news articles I found it an interesting thought.
I liked the post and immediately decided to put together a reaction simply because at the RecSys 2017 I got into this discussion and I thought it would be good to put it together. I want to share it here with you too.
First surprising point I found out is that the recommender even helps to lessen the effect of filter bubble. There was research on movielens which showed that filter bubble was less of an issue for users who were exposed to recommendations comparing to users who were not (source Joe Konstant on quora https://www.quora.com/Do-recommender-systems-create-a-“filter-bubble” ) and additonally here is a link to Pierce's page which summarizes the research on "filter bubble" http://www.intotheminds.com/blog/en/recommendation-algorithms-the-myth-of-filter-bubbles-at-stake/ .
Interestingly enough Xaviar points out that even people running/controling the page's content are much worse in creating filter bubble than algorithms - source Xavier Amatriain on quora https://www.quora.com/Should-we-worry-about-the-filter-bubble-created-by-recommender-systems . And my last two cents. If you think about it from different perspective, RS are made to support business's goal. So even though there are techniques which can be used to deliberately aim to break the filter bubble, but if the business takes more benefit for enforcing filter bubble I don't think many managers would take the stance "Let's not enforce filter bubble because it is less ethical even though we make more money out of it....".
https://www.linkedin.com/feed/update/urn:li:activity:6344900674899185664https://www.linkedin.com/feed/update/urn:li:activity:6344900674899185664
