My brother sent me this interesting article about some of the sampling bias and representational issues present in online rating systems. I had noticed the new Amazon graphical representation which shows the distribution of ratings rather than just the mean. I wonder if clustering and showing the clusters in ratings interfaces would also be an effective way of representing the means and modes?
Nick Diakopoulos, Ph.D.
Associate Professor
School of Communication, Northwestern UniversityDirector
Computational Journalism Lab
http://cjlab.northwestern.edu/Contact
Email
nad[at]northwestern[dot]eduTwitter
@ndiakopoulosOffice Location
2240 Campus Rd., Evanston, IL
Frances Searle Building, room 2-254Areas of Expertise
- Computational & Data Journalism
- Algorithmic Accountability & Transparency
- Automated & Algorithmic News Production
- Social Media in News Contexts
- Storytelling with Data Visualization
New Book!
Home