Roger Ebert

The final report:
Look here for a web page version or download it in one of the following formats:
[adams_project.ps.gz] [adams_project.pdf] [adams_project.doc]

The source tree:
Top level

• K means clustring (kmean)
  • K means clusting with a genetic algorithm for search (cluster_with_GA)
• Naive bayes classification (nb)
• Linear regression (lin)

The data:
Top level (All code was created for Matlab, version 6, Release 12)

• eb.mat: Titles, star ratings, reviews, union set of words and corresponding frequency
• ffvec.mat: All reviews evaluated along the entire feature vector
• Xphi_f005.mat: The binary version of ffvec, filtered for all features with high mutual information with the "thumbs-up/thumbs-down" vector (see paper for details).
• Code to generate above files (processing_scripts)