Slope One Recommender on Hadoop YONG ZHENG Center for Web Intelligence DePaul University Nov 15, 2012
Sep 03, 2014
Slope One Recommender on Hadoop
YONG ZHENGCenter for Web Intelligence
DePaul UniversityNov 15, 2012
Overview
• Introduction
• Recommender Systems & Slope One Recommender
• Distributed Slope One on Mahout and Hadoop
• Experimental Setup and Analyses
• Drive Mahout on Hadoop
• Interesting Communities
Center for Web Intelligence, DePaul University, USA
Introduction• About Me: a recommendation guy
• My Research: data mining and recommender systems
• Typical Experimental Research
1) Design or improve an algorithm;2) Run algorithms and baseline algs on datasets;3) Compare experimental results;4) Try different parameters, find reasons and even re-design
and improve algorithm itself;5) Run algorithms and baseline algs on datasets;6) Compare experimental results;7) Try different parameters, find reasons and even re-design
and improve algorithm itself;8) And so on… Until it approaches expected results.
Introduction• Sometimes, data is large-scale.
e.g. one algorithm may spend days to complete, how about experimental results are not as expected. Then improve algorithms and run it for days again, and again.
How can we do previously? (for tasks not that complicated)1). Paralleling but complicated synchronization and limited
resources, such as CPU, memory, etc;2). Take advantage of PC Labs, let’s do it with 10 PCs
• Nearly all research will ultimately face the large-scale problems , especially in the domain of data mining.
• But, we have Map-Reduce NOW!
Introduction
• Do not need to distribute data and tasks manually. Instead we just simply generate configurations.
• Do not need to care about more details, e.g. how data is distributed, when one specific task will be ran on which machine, or how they conduct tasks one by one.
• Instead, we can pre-define working flow. We can take advantage of the functional contributions from mappers and reducers.
• More benefits: replication, balancing, robustness, etc
Recommender Systems
• Collaborative Filtering
• Slope One and Simple Weighted Slope One
• Slope One in Mahout
• Distributed Slope One in Mahout
• Mappers and Reducers
Center for Web Intelligence, DePaul University, USA
Recommender Systems
Collaborative Filtering (CF)One of most popular recommendation algorithms. User-based: User-CF Item-based: Item-CF, Slope One
User
4 star
Rating?
4
5
5
5
4
Example: User-based Collaborative Filtering
Slope One RecommenderReference: Daniel Lemire, Anna Maclachlan, Slope One Predictors for Online Rating-Based Collaborative Filtering, In SIAM Data Mining (SDM'05), April 21-23, 2005. http://lemire.me/fr/abstracts/SDM2005.html
User Batman Spiderman
U1 3 4
U2 2 4
U3 2 ?
1). How different two movies were rated?U1 rated Spiderman higher by (4-3) = 1U2 rated Spiderman higher by (4-2) = 2On average, Spiderman is rated (1+2)/2 = 1.5 higher
2). Rating difference can tell predictionsIf we know U3 gave Batman a 2-star, probably he will rated Spiderman by (2+1.5) = 3.5 star
Simple Weighted Slope OneUsually user rated multiple items
User HarryPotter Batman Spiderman
U1 5 3 4
U2 ? 2 4
U3 4 2 ?
1). How different the two movies were rated?Diff(Batman, Spiderman) = [(4-3)+(4-2)]/2 = 1.5Diff(HarryPotter, Spiderman) = (4-5)/1 = -1“2” and “1” here we call them as “count”.
2). Weighted rating difference can tell predictionsWe use a simple weighted approachRefer to Batman only, rating = 2+1.5 = 3.5Refer to HarryPotter only, rating = 4-1 = 3Consider them all, predicted rating = (3.5*2 + 3*1])/ (2+1) = 3.33
Simple Weighted Slope OneUser HarryPotter Batman Spiderman
u1 5 3 4
u2 ? 2 4
u3 4 2 ?
To calculate the prediction ratings, we need 2 matrices:1).Difference Matrix
2). Count MatrixJust number of users co-rated on two items
Movie1 Movie2 Movie3 Movie4
Movie1
Movie2 -1.5
Movie3 2 1
Movie4 -1 0.5 -2
Question: Online or Offline?
Slope One in Mahout
Mahout, an open-source machine learning library.
1). Recommendation algorithmsUser-based CF, Item-based CF, Slope One, etc
2). ClusteringKMeans, Fuzzy KMeans, etc
3). ClassificationDecision Trees, Naive Bayes, SVM, etc
4). Latent Factor ModelsLDA, SVD, Matrix Factorization, etc
Slope One in Mahoutorg.apache.mahout.cf.taste.impl.recommender.slopeone.SlopeOneRecommenderPre-Processing Stage: (class MemoryDiffStorage with Map)for every item i
for every other item jfor every user u expressing preference for both i and jadd the difference in u’s preference for i and j to an average
Recommendation Stage:for every item i the user u expresses no preference for
for every item j that user u expresses a preference forfind the average preference difference between j and iadd this diff to u’s preference value for jadd this to a running average
return the top items, ranked by these averages
Simple weighting: as introduced previouslyStdDev weighting: item-item rating diffs with lower sd should be
weighted highly
Distributed Slope One in Mahout
Similar to our previous practice, e.g. the matrix factorizationProcess, what we need is the Difference Matrix.
Suppose there are M users rated N items, the matrix requires N(N-1)/2 cells. Also, the density is another aspect – how user rated items. If there are several items and the rating matrix is dense, the computational costs will increase accordingly.
Question again: Online or Offline?Depends on tasks & data.
Large-scale data. Let’s do it offline!
Distributed Slope One in Mahout
package org.apache.mahout.cf.taste.hadoop.slopeone;class SlopeOneAverageDiffsJobclass SlopeOnePrefsToDiffsReducerclass SlopeOneDiffsToAveragesReducer
package org.apache.mahout.cf.taste.hadoop;class ToItemPrefsMapperorg.apache.hadoop.mapreduce.Mapper
Two Mapper-Reducer Stages:1). Create DiffMatrix for each user2). Collect AvgDiff info, counts, StdDev
Let’s see how it works…
Mapper and Reducer - 1
Mapper1 (ToItemPrefsMapper) <UserID, Pair<ItemID, Rating>>Reducer1 (PrefsToDiffsReducer) <Pair<Item1,Item2>, Diff> (for all three users)
User HarryPotter Batman Spiderman
U1 5 3 4
U2 ? 2 4
U3 4 2 ?
<U1> Potter Bat Spider
Potter
Bat -2
Spider -1 1
<U2> Potter Bat Spider
Potter
Bat NULL
Spider NULL 2
Mapper and Reducer - 2
Mapper2 (org.apache.hadoop.mapreduce.Mapper)Reducer2 (DiffsToAveragesReducer) Average Diffs, Count, StedDev
<U1> Potter Bat Spider
Potter
Bat -2
Spider -1 1
<U2> Potter Bat Spider
Potter
Bat NULL
Spider NULL 2
<Aggregate> Potter Bat Spider
Potter
Bat -2, 1
Spider -1, 1 1.5, 2
Simply, <a,b> pair denotes a=averge diff, b=countNotice: we should use three matrices in practice, here I used 2.
Predictions
<Aggregate> Potter Bat Spider
Potter
Bat -2, 1
Spider -1, 1 1.5, 2
Simply, <a,b> pair denotes a=averge diff, b=countNotice: we should use three matrices in practice, here I used 2.
User HarryPotter Batman Spiderman
U1 5 3 4
U2 ? 2 4
U3 4 2 ?
Prediction(U3, Spiderman) = [(4-1)*1 + (2+1.5)*2] / (1+2)= 3.33333333333333333333
Experiments
• Data
• Hadoop Setup
• Running Performances
Center for Web Intelligence, DePaul University, USA
Experiment SetupData: MovieLens-1M ratings
# of users: 6,040# of movies: 3,900# of ratings: 1,000,209
Density of the ratings: each user has at least 20 ratingsobviously, some users have many more ratings
Rating format: UserID, ItemID, Rating (scale 1-5)
Data Split: 80% training, 20% testing
Experiment Setup
Hadoop Cluster Setup IBM SmartCloud 1 master node, 7 slave nodes Each node is as SUSE Linux Enterprise Server v11 SP1 Server Configuration:
64 bit (vCPU: 2, RAM: 4 GiB, Disk: 60 GiB) Hadoop v.0.20.205.0 Mahout distribution-0.6
The environment setup follows the typical workflow as:http://irecsys.blogspot.com/2012/11/configurate-map-reduce-environment-on.html
Thanks Scott Young, neat writeup!!
Experimental AnalysesStage-1: SlopeOneAverageDiffsJob by Map-Reduce
Goal: Build DiffStorageOutput: DiffStorage txt file, 1.45GBRunning Time: real 13m 34.228s user 0m 5.136s sys 0m 1.028s
Stage-2: Java evaluator to measure MAE on testing setRunning Time: Load Testing Set (21K records), 299ms Load Training Set (79K records), 1,771ms Load DiffStorage, 176,352ms = 2.9m Prediction (21K records), 18,182ms = 0.3m MAE = 0.71330756
Item1 Item2 Diff Count StdDev
221 223 -1.02 197 0.5
Experimental Experiences1. Why not MovieLens 10M data?
Map-Reduce on 10M data may cost several hrs;Running time depends on cluster and configuration;Also, DiffStorage file will be too large.
2. Java EvaluatorLoad full DiffStorage file is time-consuming.Also, incur Java heap space and GCOverlimit errors;Those errors can not be fixed by –Xmx or other solutions;Two solutions:1). Just use simple weighting, discard StdDev weighting.2). Simple Mapper and Reducer, run it on clusters.
For MovieLens 1M, it is not that efficient compared with the live SlopeOne recommendation; 10M data may be better, will try MovieLens-10M data later; Slope One is simple but memory-expensive.
More …
• Drive Mahout on Hadoop
• Interesting Communities
Center for Web Intelligence, DePaul University, USA
Mahout + HadoopHow to put more Mahout algorithms to Hadoop?
1. Pre-set Command in MahoutLet’s see bin/mahout – help, then it provides a list of available programs such as svd, fkmeans, etc.
Some are basic functions, such as splitDatasetSome can be executed as Hadoop tasks
e.g. Run and evaluate Matrix Factorization on rating dataset
bin/mahout parallelALS --input inputSource --output outputSource--tempDir tmpFolder --numFeatures 20 --numIterations 10
bin/mahout evaluateFactorization --input inputSource --output outputSource --userFeatures als/out/U/ --itemFeatures als/out/M/ --tempDir tmpFolder
Mahout + Hadoop2. More Algorithms on Hadoop
Mahout provides a way to run more Mahout algorithms. Simply,
$HADOOP_HOME/bin/hadoop jar $MAHOUT_HOME/core/target/mahout-core-<version>.jar <Job Class> --recommenderClassName Class <OPTIONS>
Which kinds of Jobs it supports? Mahout implemented some versions.
Some popular ones:1).org.apache.mahout.cf.taste.hadoop.pseudo.RecommenderJob
--recommenderClassName ClassName2).org.apache.mahout.cf.taste.hadoop.item.RecommenderJob3).org.apache.mahout.cf.taste.hadoop.als.ParallelALSFactorizationJob4).org.apache.mahout.cf.taste.hadoop.slopeone.SlopeOneAverageDiffsJob
Interesting CommunitiesBeyond Hadoop and Mahout official sites
1. Data MiningKDnuggets, http://www.kdnuggets.comPopular community for Data Mining & Analytics. Lots of usefulinformation, such as news, materials, datasets, jobs, etc.
2. Big DataSmartData Collective, http://smartdatacollective.com/Smarter Computing, http://www.smartercomputingblog.com/Big Data Meetup, http://big-data.meetup.com/
3. Recommender SystemsACM Official Site, http://recsys.acm.org/RecSys Wiki, http://recsyswiki.com/
Thank You!
Center for Web Intelligence, DePaul University, USA