In this paper we describe rrecsys, an open source extension package in R for rapid prototyping and intuitive assessment of recommender system algorithms. Due to its wide variety of implemented packages and functionalities, R language represents a popular choice for many tasks in Data Analysis. This package replicates the most popular collaborative ltering algorithms for rating and binary data and we compare results with the Java-based LensKit implementation for the purpose of benchmarking the implementation. Therefore this work can also be seen as a contribution in the context of replication of algorithm implementations and reproduction of evaluation results. Users can easily tune available implementations or develop their own algorithms and assess them according to the standard methodology for o ine evaluation. Thus this package should represent an easily accessible environment for research and teaching purposes in the eld of recommender systems.
R represents a popular choice in Data Analytics and Machine Learning. The software has low setup cost and contains a large selection of packages and functionalities to enhance and prototype algorithms with compact code and good visualization tools. Thus R represents a suitable environment for exploring the eld of recommender systems. Therefore we present and contribute a novel R package that reproduces several of the most popular recommender algorithms for Likert scaled as well as binary rating values. The functionality of this framework is mainly focused on prototyping and educational purposes due to the compact code representation and the interactive way of invoking and visualizing results in R. We took advantage of the large selection of packages in R to implement the algorithms included in our package. We achieve competing performance due to highly vectorized and mixed R/C++ implementations.
We therefore introduce rrecsys[
rrecsys Package rrecsys has a modular structure as well as includes expansion capabilities. The core of the package includes the implementation of several popular algorithms, an evaluation component and a couple of auxiliary parts for data analysis and convergence detection. Next we concisely describe the included algorithms. 2.1
Baseline and Popularity: The included baseline predictors are the global mean rating (Global Average), item's mean rating (Item Average), user's mean ratings (User Average) as well as an unpersonalized Most Popular method that determines item popularity based on the total number of (positive) ratings.
Item Based K-nearest neighbors: Given a target user and her positively
rated items, the algorithm will identify the k-most similar items for each target a
and will rank them according to aggregated similarities with the di erent targets
as described by Sarwar et al. [
Items a and b are considered as rating vectors a and b in the user space. Cosine similarity measures the cosine angle between those vectors: sim(a; b) = cos(a; b) = jaaj bjbj (1) The adjusted cosine similarity is computed by o setting the user average on each co-rated pair on two item vectors. If Ia and Ib is the set of users that rated correspondingly item a and b the adjusted cosine similarity is measured as: sim(a; b) = r
P u2Ia\Ib
P u2Ia\Ib (Rua (Rua
Ru) (Rub
Ru) Ru)2
P u2Ia\Ib (Rub
Ru)2 Where Ru is the average rating of user u, computed as:
Ru =
X Rui i2Iu jIuj Where Iu is the set of items rated by user u.
Once similarities among all items are computed a neighborhood might be
formed by choosing the items with the highest similarity value. Prediction over
a target user u and item a are calculated as the weighted sum [
P (sa;k pu;a = all similar items;k
P all similar items;k
Ru;k) sa;k
User Based K-nearest neighbors: Herlocker et al. [
User u and v are considered as rating vectors u and v in the item space. Cosine similarity measures the cosine angle between those vectors using Formula 1
Instead the Pearson correlation is measured by o setting the user average on co-rated pairs among the user vectors: sim(u; v) = P earson(u; v) = r
P i2Iu\Iv
P i2Iu\Iv (Rui (Rui
Ru) (Rvi
Rv) Ru)2
P i2Iu\Iv (Rvi
Rv)2 Where Ru and Rv are correspondingly the average ratings of user u and user v, computed like in Formula 3.
Once similarities among all items are computed a neighborhood might be
formed by choosing the items with the highest similarity value. Prediction over
a target user u and item a are calculated as the weighted sum [
P (sa;k all similar items;k
P all similar items;k
Ru;k) sa;k
Weighted Slope One: proposed by Lemire et al [
P8ruj (devij + ruj)cij
P8ruj cij
: devij =
X rui cij ruj : Where cij is the number of co-rated items between items i and j and rui is an existing rating for user u on item i. The Weighted Slope One takes into account both, information from users who rated the same item and the number of observed ratings.
Simon Funk's SVD: Matrix factorization methods are used in
recommender systems to derive a set of latent factors, from the user item rating
matrix, to characterize both users and items by this vector of factors. The
useritem interaction are modeled as inner product of the latent factors space [
R^ui =
+ bi + bu + Uu ViT Where is the overall average rating and bu and bi indicate the deviation due to user u and item i from the mean rating.
The U(user) and V(item) factor matrices are cropped to k features and initialized at small values. Each feature is trained until convergence (where convergence specifying the number of updates to be computed on a feature before (5) (6) (7) (8) (9) considering it converged, it can be either chosen by the user or calculated automatically by the package). On each loop the algorithm predicts R^ui, calculates the error and the factors are updated as follows: The attribute ization term.
eui = Rui
R^ui Vik Uuk
Vik + Uuk + (eui Uuk (eui Vik
Vik) Uuk) (10) (11) (12) represents the learning rate, while corresponds to the regular
In addition, the following two algorithms address the One Class Collaborative Filtering problem (OCCF).
Bayesian Personalized Ranking: The algorithm has been introduced by
[
mation matrix R^ = (R^ij )m n = U V T , where U and V are the usual feature
matrix cropped to k features as introduced by [
L(R^) = L(U; V ) = X Wij (Rij ij
Ui VjT )2 + (kUik2F + kVik2F ) (13) The regularization term weighted by is added to prevent over- tting. The expression k:kF denotes the Frobenius norm. The alternated least square algorithms optimization process solves partial derivatives of L with respect to each entry U and V , @L@(UU;iV ) = 0 with xed V and @L@(UVj;V ) = 0 with xed U , to compute Ui and Vi. Then U and V are initialized with random Gaussian numbers with mean zero and small standard deviation and are updated until convergence. The matrix W = (Wij )m n 2 R+m n is a non-negative weight matrix that assigns con dence values to observations (hence the name weighted ALS). 2.2
The library is currently distributing two di erent methodologies for computing
the top-N recommendation. The rst is the Highest Predicted Ratings (HPR),
which proposes a sorted list based on the highest computed rating values by an
algorithm. The second method is the Most Frequent (MF), that determines the
top-N list based on the most frequent items available in the neighborhood of an
user or item. This methodology is known to produce better performance than
HPR [
The evaluation module is based on the k-fold cross-validation method. A stratied random selection procedure is applied when dividing the rated items of each user into k folds such that each user is uniformly represented in each fold, i.e. the number of ratings of each user in any fold di ers at most by one. For k-fold cross validation each of the k disjunct fractions of the ratings are used k 1 times for training (i.e. Rtrain) and once for testing (i.e. Rtest). Practically, ratings in Rtest are set as missing in the original dataset and predictions/recommendations are compared to Rtest to compute the performance measures.
We included the most popular performance metrics according to the survey
in [
In this Section, we compare our rrecsys library with the popular Lenskit [
In Figure 1, we compare both libraries in terms of RMSE and MAE, using
5-fold cross validation. Lenskit and rrecsys were con gured both with the same
algorithms and evaluation methodology. We used the MovieLens100K dataset
[
For the experiments in Figure 1 on the FunkSVD, we have tuned its parameters as follows, the latent space is set to 100 features, the learning rate is set to 0.001, the regularization term is set to 0.015. In the case of the item based k-nearest neighbor algorithm, we have set the number of nearest neighbors to 100, and adjusted cosine similarity is the similarity measure. In the case of the user-based k-nearest neighbor algorithm, we have set the number of nearest neighbors to 100, and used Pearson as similarity measure.
As shown in Figure 1, all the reported results demonstrate our ability to
clearly reproduce and replicate the same results with those of Lenskit in terms of
both RMSE and MAE while other libraries failed to do so [
In Table 1 we show the performance of rrecsys based on the latest implementations with R/C++ code on the MovieLens100K dataset running on the same machine. It is noticeable that rrecsys performs similarly to LensKit. We compared only optimized algorithms. In future we will provide optimized implementation of more state of the art algorithms.
1:5
1 0:5 glob
RMSE rrecsys
MAE rrecsys In this Section, we introduce an executable script in R for running some of the functionalities of rrecsys in order to demonstrate its intuitive use. Please notice that due to space limitations in this paper we do not describe all commands in detail. The library is distributed with a full range of vignettes and a manual describing all available functionalities1.
The package is loaded on the Comprehensive R Archive Network (CRAN), therefore download, installation and loading of the package requires the execution of the following two functions: install.packages("rrecsys");
library(rrecsys)
Once the package is loaded the dataset MovieLens Latest2 will be available within the environment. A setup of the data is required to de ne possible limits and the structure of the dataset. Users can explore the dataset by checking the 1 https://cran.r-project.org/package=rrecsys 2 We redistribute the MovieLens Latest datasets for demonstration purposes only.
Please notice that these datasets change over time and are not appropriate for reporting experimental results. number of ratings, its sparsity or even by modifying it to contain a speci c number of ratings for each itemnuser. data("mlLatest100k") m <- defineData(mlLatest100k, minimum = 1, maximum = 5, halfStar = TRUE) sparsity(m); numRatings(m); rowRatings(m); colRatings(m) #Crop the dataset to contain at least 200 ratings on each user and 10 ratings on each item. smallmlLatest <- m[rowRatings(m) >= 200, colRatings(m) > 10]
The following code shows how to train a model (e.g., ub10) on an algorithm (e.g., UBKNN), which can be used for either rating prediction (e.g., p) or item recommendation (e.g., rHPR and rMF). ub10 <- rrecsys(smallmlLatest, "UBKNN", neigh = 10, simFunct = 1) p <- predict(ub10) rHPR <- recommendHPR(ub10, topN = 10) #pt is the positive threshold for recommending an item. rMF <- recommendMF(ub10, topN = 10, pt = 3)
The following code shows how we generate the k-folds. Same fold distribution can be used to evaluate di erent algorithms. folds <- evalModel(smallmlLatest, folds = 2) #Recommendation evaluation. evalRec(folds, "UBKNN", topN = 10, goodRating = 3,
simFunct = 2, recAlg = 1)
The output of the evaluation function looks like as follows: #Prediction evaluation. > evalPred(folds, "funksvd", k = 10) #Output:
MAE RMSE globalMAE globalRMSE Time 1-fold 0.8565404 1.056737 0.9175207 1.161164 2.419723 2-fold 0.8473669 1.043400 0.8959667 1.131292 2.669417 Average 0.8519536 1.050069 0.9067437 1.146228 2.544570 5
This paper contributed a recently released package for prototyping and interactively demonstrating recommendation algorithms in R. It comes with a nice range of implemented standard algorithms for Likert scaled and binary ratings. Reported results demonstrate that it reproduces results of the Java-based Lenskit toolkit. Thus it remains to hope that this e ort will be of use for the eld of recommender systems and the large R user community.