Lecture 9

Homework

For next Tuesday:

Start to build a Yelp reputation system, on the basis of the Yelp data that was crawled. The idea is to store, for each restaurant, a "star distribution", and for each user, a "bias distribution", as we mentioned in class.
Call a user a n-user if a user has performed at least n reviews, and plot the number of n-users vs. n. Also, call a restaurant a k,n-restaurant if it has at least k reviews by n-users, and plot the number of k,n-restaurants vs. k for n=1, 2, 3, 4. These two plots (actually, five plots, if you do them all separate) will give us at least some idea of how connected the graph of restaurants and users is, and thus, of how useful it is to transfer information we have on users to restaurants, and vice versa.

Here is some material on content-driven reputation systems:

Main:

Other:

B.T. Adler, L. de Alfaro, S.M. Mola-Velasco, P. Rosso, A.G. West. Wikipedia Vandalism Detection: Combining Natural Language, Metadata, and Reputation Features. In CICLING 2011: Proceedings of the 12th International Conference on Intelligent Text Processing and Computational Linguistics, LNCS 6609, pages 277-288, Springer Verlag, 2011. PDF
B.T. Adler, L. de Alfaro, I. Pye. Detecting Wikipedia Vandalism Using WikiTrust. PAN lab report, CLEF (Conference on Multilingual and Multimodal Information Access Evaluation), 2010. PDF

Page updated

Report abuse