In equation (1) from Toscher, Jahrer, and Legenstein's Improved Neighborhood-Based Algorithms for Large-Scale Recommender Systems (2008)

$\hat{r}_u{_i} = \frac{\Sigma_j{_∈}{_N}{_{_k}}{_(}{_u}{_,}{_i}{_)} > c_i{_j} r_u{_v}} > {\Sigma_j{_∈}{_N}{_{_k}}{_(}{_u}{_,}{_i}{_)} c_i{_j}}$

where $N_k{_(}{_u}{_,}{_i}{_)}$ denotes the set of the k items most similar to item i that were rated by user u.

what is $r_u{_v}$? The user rating for item v, but what is v?


The meaning of the indicies of $r$ is spelled out on the first page of the paper

We define the set of users as the set of integers {1, . . . , m} and the set of items as the set of integers {1, . . . , n}. The $m \times n$ rating matrix $R = [r_{ui}] \ 1 \leq u \leq m;1 \leq i \leq n$ stores the rating sof users for items where $r_{ui}$ is the rating of user u for item i.

So $v$ is one of the items in the suite of objects you are attempting to recommend. Or, more pedantically, the index of one of the items after you have assigned unique indices to every item in your suite.

In the equation (1), I believe the $v$ is a type left over from a copy and paste. It should be a $j$, i.e. we are summing over all the similar items and weighting their rating by the similarity.

  • $\begingroup$ I don't think this answers the question. Which item is $v$ actually referring to? $\endgroup$ – RoryT Jul 25 '17 at 3:53
  • $\begingroup$ Oh, I see. Ok, I think that's a typo. $\endgroup$ – Matthew Drury Jul 25 '17 at 3:57
  • $\begingroup$ A possible reference for this is equation 8 in glaros.dtc.umn.edu/gkhome/fetch/papers/NbrRSsurvey2011.pdf $\endgroup$ – RoryT Jul 25 '17 at 4:04
  • $\begingroup$ I believe you're correct--it seems to be a typo. Thanks for the reference @RoryT! $\endgroup$ – johnklawlor Jul 27 '17 at 22:53

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.