GithubHelp home page GithubHelp logo

alberto-paparella / musicrecommendation Goto Github PK

View Code? Open in Web Editor NEW
1.0 2.0 0.0 917 KB

Repository for the project of the Scalable and Cloud Computing course @unibo, a.y. 2022-23.

Scala 66.77% Jupyter Notebook 33.23%
cloud music music-recommendation scala scalable

musicrecommendation's Introduction

Music Recommendation

Repository for the project of the Scalable and Cloud Computing course @unibo, a.y. 2022-23.

Inspired by the Million Song Dataset challenge on Kaggle, a comparison of various solutions to the problem exploiting the MapReduce paradigm and providing an environment for sequential, parallel and distributed execution. Hence, we aim to provide not only a comparison between the different algorithms, but also an analysis of the behaviours of these algorithms across the different executions.

The different algorithms produced can be found in the src/main/scala/ directory as scala worksheet files. To execute the algorithms, one should download the Echo Nest Taste Profile Subset dataset from http://millionsongdataset.com/tasteprofile/ and place the train_triplets.txt file under the src/main/resources/ directory (we advise working on a subset of it; dataExtraction.ipynb can be a useful tool to do that). Once executed, the results of the training will be available through a text file with the same name of the algorithm which can be found under target/scala-2.12/classes/models/. Make sure to also have an empty file in src/main/resources/models/ with the same name of the model (in case not, create it, as it will be used to dynamically create the path for the output model, which will still be found under the target/scala-2.12/classes/models/ path).

We plan to implement the following algorithms:

  • UserBasedModel
  • ItemBasedModel
  • LinearCombinationModel
  • AggregationModel
  • StochasticAggregationModel

How to execute code on Google Cloud Platform

  1. Be sure that:
    • Every file path in code is in the local path (e.g., "train_100_50.txt" instead of "~/MusicRecommender/.../train_100_50.txt")
    • All files are loaded with Source.fromResource instead of Source.fromFile
    • You have already created a Google Cloud Project (here's a guide)
    • You have already installed Google Cloud CLI (here's a guide)
  2. Create an assembly .jar package using the following command (in the project directory):
    sbt clean assembly
    • (Optional, but strongly recommended) You can check if the code works: go to MusicReccomender/target/scala-2.12/ and run
        java -jar package.jar
  3. Create a new bucket inside the project in Google Cloud Platform (in this example it is called mr)
  4. Copy the package created inside MusicReccomender/target/scala-2.12/ to the bucket with the following command:
    gsutil cp package.jar gs://mr
  5. Create a Compute Engine Cluster on Google Cloud Platform
    • Be careful to choose the correct linux image, which is 2.1 (Debian 11, Hadoop 3.3, Spark 3.3) first release: 22/01/2021
    • You can plan the elimination of the cluster so that you won't waste credits
  6. Create and send a new job by:
    1. Selecting the created cluster
    2. Choosing Spark job as "Type of job"
    3. Typing gs://mr/package.jar in "Main class or jar"

Experiments

We designed three kind of experiments: the first two are meant to compare the scalability of the proposed solution and the efficiency of the models scaling on the number of test (1) or train (2) users, while the third one is meant to compare the models to find which one is the best, using as evaluation function the mean average precision, following the challenge spirit in which the problem has first been proposed (Kaggle); this is done only on the distributed version on Google Cloud Platform, making use of the proven scalability to compute the models on a much bigger dataset.

Local Experiments

The following results are obtained on an Intel® Core™ i5-8250U Processor, 6M Cache, up to 3.40 GHz, 4 cores, 8 threads. All times are in ms.

Incrementing the number of test users

  • Sequential:
TrainUsers TestUsers Songs ubm ibm
100 10 4798 42857 70839
100 50 5679 253379 516129
100 100 6620 739567 1403311
  • Parallel:
TrainUsers TestUsers Songs ubm ibm
100 10 4798 15507 25829
100 50 5679 105049 233113
100 100 6620 210372 402987

local_test

Incrementing the number of train users

  • Sequential:
TrainUsers TestUsers Songs ubm ibm
100 10 4798 42857 70839
200 10 8763 136434 166437
300 10 11660 277385 464244
400 10 14497 484250 415634
500 10 16785 1090257 823119
  • Parallel:
TrainUsers TestUsers Songs ubm ibm
100 10 4798 15507 25829
200 10 8763 46980 63112
300 10 11660 118278 185948
400 10 14497 201378 159863
500 10 16785 425118 330385

local_train

Google Cloud Platform Results

The following sequential and parallel results are obtained on a single node with a Google's N1 Processor with 4 virtual CPUs and 15GB of RAM, while distributed results are obtained on 4 worker nodes, each with a Google's N1 Processor with 4 virtual CPUs and 15GB of RAM. All times are in ms.

Regarding the distributed execution, we reported the results for both approaches (distributing over songs and locally parallelize over test users and vice versa), to study their behaviour over different numbers of users and songs.

Incrementing the number of Test Users

  • Sequential:
TrainUsers TestUsers Songs ubm ibm
100 10 4798 58175 95395
100 50 5679 655528 732643
100 100 6620 1453406 1803859
  • Parallel:
TrainUsers TestUsers Songs ubm ibm
100 10 4798 53851 52474
100 50 5679 359120 376764
100 100 6620 733755 1061279
  • Distributed:
TrainUsers TestUsers Songs ubm1 ubm2 ibm1 ibm2
100 10 4798 32020 42351 7790 7126
100 50 5679 95183 213722 98225 27701
100 100 6620 147121 424269 193660 68685

gcp_test

Incrementing the number of Train Users

  • Sequential:
TrainUsers TestUsers Songs ubm ibm
100 10 4798 58175 95395
200 10 8763 272104 249830
300 10 11660 513474 635022
400 10 14497 832376 625668
500 10 16785 1635081 1334183
  • Parallel:
TrainUsers TestUsers Songs ubm ibm
100 10 4798 53851 52474
200 10 8763 146988 113650
300 10 11660 220897 378365
400 10 14497 384275 301811
500 10 16785 799185 622275
  • Distributed
TrainUsers TestUsers Songs ubm1 ubm2 ibm1 ibm2
100 10 4798 32020 42351 7790 7126
200 10 8763 59725 95462 47734 18244
300 10 11660 162974 202068 109075 42347
400 10 14497 155877 125346 70841 66695
500 10 16785 313024 250710 138590 114828
1000 10 17748 1484891 1420594 574715 606498

gcp_train

Models comparison over 1000 train users and 100 test users

For each model, its mean average precision (mAP) value is reported.

Songs NewSongs ubm ibm lcm am scm
28720 2176 0.0385800219 0.142942848 0.1369886698 0.0950991402 0.101490377

Classification:

  1. Item based model
  2. Linear combination model
  3. Stochastic combination model
  4. Aggregation model
  5. User based model

map

Models comparison over 2000 train users and 100 test users

For each model, its mean average precision (mAP) value is reported.

Songs NewSongs ubm ibm lcm am scm
44451 1865 0.0296820445 0.1549078628 0.0923070961 0.0770281485 0.0666806146

Classification:

  1. Item based model
  2. Linear combination model
  3. Aggregation model
  4. Stochastic combination model
  5. User based model

map2000

Some notes about combination models and future experiments

All combination models have been realized such that a parameter alpha specifies how to weight the contributions from the user based and the item based model (as alpha and 1 - alpha, respectively). This means that one could eventually try all combinations of contributions from the two models, possibly finding a value for alpha such that at least the linear combination model gives better results than one singular model (in out case, the item based model), as it happens in the paper. For example, in our case one could think to weight the item based model more than the user based one (alpha=0.5 by default, giving the same weights to both models). It is also possible to calculate the two models separately and then importing them in a new execution to try these experiments on combinations without the need to recompute both models each time.

Experiment logs

Results with 100 train users and 10 test users
  • Songs: 4798
  • New songs: 262
  • Elapsed time for (Sequential) user-based model: 42857ms (42857364180ns)
  • Elapsed time for (Parallel) user-based model: 15507ms (15507377234ns)
  • Elapsed time for (Sequential) item-based model: 70839ms (70839461276ns)
  • Elapsed time for (Parallel) item-based model: 25829ms (25829145704ns)
  • Elapsed time for (Sequential) linear-combination model: 25ms (25468308ns)
  • Elapsed time for (Parallel) linear-combination model: 18ms (18808361ns)
  • Elapsed time for (Sequential) aggregation model: 25ms (25376443ns)
  • Elapsed time for (Parallel) aggregation model: 22ms (22824598ns)
  • Elapsed time for (Sequential) stochastic-combination model: 22ms (22814212ns)
  • Elapsed time for (Parallel) stochastic-combination model: 18ms (18587216ns)
  • Elapsed time for (Sequential) user-based model mAP: 441ms (441693398ns)
  • Elapsed time for (Parallel) user-based model mAP: 165ms (165774138ns)
  • Elapsed time for (Sequential) item-based model mAP: 461ms (461173721ns)
  • Elapsed time for (Parallel) item-based model mAP: 180ms (180187672ns)
  • Elapsed time for (Sequential) linear-combination model mAP: 398ms (398532015ns)
  • Elapsed time for (Parallel) linear-combination model mAP: 172ms (172372507ns)
  • Elapsed time for (Sequential) aggregation model mAP: 312ms (312546899ns)
  • Elapsed time for (Parallel) aggregation model mAP: 152ms (152993251ns)
  • Elapsed time for (Sequential) stochastic-combination model mAP: 242ms (242267654ns)
  • Elapsed time for (Parallel) stochastic-combination model mAP: 96ms (96231460ns)
  • (Sequential) user-based model mAP: 0.0618047983
  • (Parallel) user-based model mAP: 0.0618047983
  • (Sequential) item-based model mAP: 0.0990458015
  • (Parallel) item-based model mAP: 0.0990458015
  • (Sequential) linear-combination model mAP: 0.1025445293
  • (Parallel) linear-combination model mAP: 0.1025445293
  • (Sequential) aggregation model model mAP: 0.0637131952
  • (Parallel) aggregation model mAP: 0.0637131952
  • (Sequential) stochastic-combination model mAP: 0.0676572156
  • (Parallel) stochastic-combination model mAP: 0.0766630316
Results with 100 train users and 50 test users
  • Songs: 5679
  • New songs: 1394
  • Elapsed time for (Sequential) user-based model: 253379ms (253379538871ns)
  • Elapsed time for (Parallel) user-based model: 105049ms (105049963306ns)
  • Elapsed time for (Sequential) item-based model: 516129ms (516129726810ns)
  • Elapsed time for (Parallel) item-based model: 233113ms (233113107429ns)
  • Elapsed time for (Sequential) linear-combination model: 79ms (79054214ns)
  • Elapsed time for (Parallel) linear-combination model: 108ms (108582538ns)
  • Elapsed time for (Sequential) aggregation model: 168ms (168176485ns)
  • Elapsed time for (Parallel) aggregation model: 125ms (125442087ns)
  • Elapsed time for (Sequential) stochastic-combination model: 104ms (104458823ns)
  • Elapsed time for (Parallel) stochastic-combination model: 94ms (94851867ns)
  • Elapsed time for (Sequential) user-based model mAP: 4352ms (4352763743ns)
  • Elapsed time for (Parallel) user-based model mAP: 1970ms (1970362190ns)
  • Elapsed time for (Sequential) item-based model mAP: 5411ms (5411032642ns)
  • Elapsed time for (Parallel) item-based model mAP: 2138ms (2138962086ns)
  • Elapsed time for (Sequential) linear-combination model mAP: 5562ms (5562102258ns)
  • Elapsed time for (Parallel) linear-combination model mAP: 1624ms (1624478566ns)
  • Elapsed time for (Sequential) aggregation model mAP: 6203ms (6203478859ns)
  • Elapsed time for (Parallel) aggregation model mAP: 1806ms (1806769783ns)
  • Elapsed time for (Sequential) stochastic-combination model mAP: 4919ms (4919265860ns)
  • Elapsed time for (Parallel) stochastic-combination model mAP: 1396ms (1396721153ns)
  • (Sequential) user-based model mAP: 0.0320140683
  • (Parallel) user-based model mAP: 0.0320140683
  • (Sequential) item-based model mAP: 0.0730435327
  • (Parallel) item-based model mAP: 0.0730435327
  • (Sequential) linear-combination model mAP: 0.0732291573
  • (Parallel) linear-combination model mAP: 0.0732291573
  • (Sequential) aggregation model model mAP: 0.0461427795
  • (Parallel) aggregation model mAP: 0.0461427795
  • (Sequential) stochastic-combination model mAP: 0.054275165
  • (Parallel) stochastic-combination model mAP: 0.0504545796
Results with 100 train users and 100 test users
  • Songs: 6620
  • New songs: 2480
  • Elapsed time for (Sequential) user-based model: 739567ms (739567071965ns)
  • Elapsed time for (Parallel) user-based model: 210372ms (210372828326ns)
  • Elapsed time for (Sequential) item-based model: 1403311ms (1403311659284ns)
  • Elapsed time for (Parallel) item-based model: 402987ms (402987760631ns)
  • Elapsed time for (Sequential) linear-combination model: 194ms (194027395ns)
  • Elapsed time for (Parallel) linear-combination model: 146ms (146129808ns)
  • Elapsed time for (Sequential) aggregation model: 282ms (282815925ns)
  • Elapsed time for (Parallel) aggregation model: 242ms (242000556ns)
  • Elapsed time for (Sequential) stochastic-combination model: 195ms (195633915ns)
  • Elapsed time for (Parallel) stochastic-combination model: 121ms (121013083ns)
  • Elapsed time for (Sequential) user-based model mAP: 17711ms (17711445931ns)
  • Elapsed time for (Parallel) user-based model mAP: 5738ms (5738134130ns)
  • Elapsed time for (Sequential) item-based model mAP: 18828ms (18828945757ns)
  • Elapsed time for (Parallel) item-based model mAP: 6157ms (6157341845ns)
  • Elapsed time for (Sequential) linear-combination model mAP: 20731ms (20731335757ns)
  • Elapsed time for (Parallel) linear-combination model mAP: 5324ms (5324153841ns)
  • Elapsed time for (Sequential) aggregation model mAP: 17257ms (17257004635ns)
  • Elapsed time for (Parallel) aggregation model mAP: 6826ms (6826038612ns)
  • Elapsed time for (Sequential) stochastic-combination model mAP: 14665ms (14665562865ns)
  • Elapsed time for (Parallel) stochastic-combination model mAP: 6211ms (6211425487ns)
  • (Sequential) user-based model mAP: 0.0165354986
  • (Parallel) user-based model mAP: 0.0165354986
  • (Sequential) item-based model mAP: 0.0394516174
  • (Parallel) item-based model mAP: 0.0394516174
  • (Sequential) linear-combination model mAP: 0.0404551622
  • (Parallel) linear-combination model mAP: 0.0404551622
  • (Sequential) aggregation model model mAP: 0.026037198
  • (Parallel) aggregation model mAP: 0.026037198
  • (Sequential) stochastic-combination model mAP: 0.0272929769
  • (Parallel) stochastic-combination model mAP: 0.0232938355
Results with 200 train users and 10 test users
  • Songs: 8763
  • New songs: 180
  • Elapsed time for (Sequential) user-based model: 136434ms (136434211069ns)
  • Elapsed time for (Parallel) user-based model: 46980ms (46980268034ns)
  • Elapsed time for (Sequential) item-based model: 166437ms (166437036452ns)
  • Elapsed time for (Parallel) item-based model: 63112ms (63112854374ns)
  • Elapsed time for (Sequential) linear-combination model: 35ms (35896897ns)
  • Elapsed time for (Parallel) linear-combination model: 25ms (25774293ns)
  • Elapsed time for (Sequential) aggregation model: 50ms (50446720ns)
  • Elapsed time for (Parallel) aggregation model: 28ms (28951610ns)
  • Elapsed time for (Sequential) stochastic-combination model: 32ms (32615643ns)
  • Elapsed time for (Parallel) stochastic-combination model: 31ms (31329748ns)
  • Elapsed time for (Sequential) user-based model mAP: 430ms (430709814ns)
  • Elapsed time for (Parallel) user-based model mAP: 231ms (231304470ns)
  • Elapsed time for (Sequential) item-based model mAP: 607ms (607535794ns)
  • Elapsed time for (Parallel) item-based model mAP: 283ms (283896505ns)
  • Elapsed time for (Sequential) linear-combination model mAP: 564ms (564789426ns)
  • Elapsed time for (Parallel) linear-combination model mAP: 314ms (314617792ns)
  • Elapsed time for (Sequential) aggregation model mAP: 440ms (440703079ns)
  • Elapsed time for (Parallel) aggregation model mAP: 224ms (224796267ns)
  • Elapsed time for (Sequential) stochastic-combination model mAP: 472ms (472007082ns)
  • Elapsed time for (Parallel) stochastic-combination model mAP: 243ms (243478770ns)
  • (Sequential) user-based model mAP: 0.0674074074
  • (Parallel) user-based model mAP: 0.0674074074
  • (Sequential) item-based model mAP: 0.1117592593
  • (Parallel) item-based model mAP: 0.1117592593
  • (Sequential) linear-combination model mAP: 0.1085185185
  • (Parallel) linear-combination model mAP: 0.1085185185
  • (Sequential) aggregation model model mAP: 0.1066666667
  • (Parallel) aggregation model mAP: 0.1066666667
  • (Sequential) stochastic-combination model mAP: 0.0787962963
  • (Parallel) stochastic-combination model mAP: 0.0926851852
Results with 300 train users and 10 test users
  • Songs: 11660
  • New songs: 299
  • Elapsed time for (Sequential) user-based model: 277385ms (277385960888ns)
  • Elapsed time for (Parallel) user-based model: 118278ms (118278364427ns)
  • Elapsed time for (Sequential) item-based model: 464244ms (464244584608ns)
  • Elapsed time for (Parallel) item-based model: 185948ms (185948456337ns)
  • Elapsed time for (Sequential) linear-combination model: 44ms (44629761ns)
  • Elapsed time for (Parallel) linear-combination model: 33ms (33530636ns)
  • Elapsed time for (Sequential) aggregation model: 55ms (55272364ns)
  • Elapsed time for (Parallel) aggregation model: 67ms (67517837ns)
  • Elapsed time for (Sequential) stochastic-combination model: 42ms (42072013ns)
  • Elapsed time for (Parallel) stochastic-combination model: 19ms (19174736ns)
  • Elapsed time for (Sequential) user-based model mAP: 716ms (716026485ns)
  • Elapsed time for (Parallel) user-based model mAP: 407ms (407989568ns)
  • Elapsed time for (Sequential) item-based model mAP: 1136ms (1136933714ns)
  • Elapsed time for (Parallel) item-based model mAP: 457ms (457322012ns)
  • Elapsed time for (Sequential) linear-combination model mAP: 932ms (932210084ns)
  • Elapsed time for (Parallel) linear-combination model mAP: 432ms (432270977ns)
  • Elapsed time for (Sequential) aggregation model mAP: 737ms (737421888ns)
  • Elapsed time for (Parallel) aggregation model mAP: 281ms (281846488ns)
  • Elapsed time for (Sequential) stochastic-combination model mAP: 716ms (716559469ns)
  • Elapsed time for (Parallel) stochastic-combination model mAP: 275ms (275453178ns)
  • (Sequential) user-based model mAP: 0.1521579869
  • (Parallel) user-based model mAP: 0.1521579869
  • (Sequential) item-based model mAP: 0.2049370919
  • (Parallel) item-based model mAP: 0.2049370919
  • (Sequential) linear-combination model mAP: 0.2063306259
  • (Parallel) linear-combination model mAP: 0.2063306259
  • (Sequential) aggregation model model mAP: 0.1652810957
  • (Parallel) aggregation model mAP: 0.1652810957
  • (Sequential) stochastic-combination model mAP: 0.1807214525
  • (Parallel) stochastic-combination model mAP: 0.1717590381
Results with 400 train users and 10 test users
  • Songs: 14497
  • New songs: 155
  • Elapsed time for (Sequential) user-based model: 484250ms (484250895513ns)
  • Elapsed time for (Parallel) user-based model: 201378ms (201378619877ns)
  • Elapsed time for (Sequential) item-based model: 415634ms (415634469106ns)
  • Elapsed time for (Parallel) item-based model: 159863ms (159863170761ns)
  • Elapsed time for (Sequential) linear-combination model: 47ms (47734294ns)
  • Elapsed time for (Parallel) linear-combination model: 37ms (37730827ns)
  • Elapsed time for (Sequential) aggregation model: 55ms (55604232ns)
  • Elapsed time for (Parallel) aggregation model: 45ms (45141359ns)
  • Elapsed time for (Sequential) stochastic-combination model: 40ms (40383968ns)
  • Elapsed time for (Parallel) stochastic-combination model: 45ms (45322117ns)
  • Elapsed time for (Sequential) user-based model mAP: 704ms (704006043ns)
  • Elapsed time for (Parallel) user-based model mAP: 453ms (453804142ns)
  • Elapsed time for (Sequential) item-based model mAP: 725ms (725913616ns)
  • Elapsed time for (Parallel) item-based model mAP: 420ms (420842369ns)
  • Elapsed time for (Sequential) linear-combination model mAP: 736ms (736513671ns)
  • Elapsed time for (Parallel) linear-combination model mAP: 389ms (389822436ns)
  • Elapsed time for (Sequential) aggregation model mAP: 599ms (599806086ns)
  • Elapsed time for (Parallel) aggregation model mAP: 356ms (356975997ns)
  • Elapsed time for (Sequential) stochastic-combination model mAP: 589ms (589438205ns)
  • Elapsed time for (Parallel) stochastic-combination model mAP: 386ms (386097244ns)
  • (Sequential) user-based model mAP: 0.143609831
  • (Parallel) user-based model mAP: 0.143609831
  • (Sequential) item-based model mAP: 0.1968356375
  • (Parallel) item-based model mAP: 0.1968356375
  • (Sequential) linear-combination model mAP: 0.2011367127
  • (Parallel) linear-combination model mAP: 0.2011367127
  • (Sequential) aggregation model model mAP: 0.1312442396
  • (Parallel) aggregation model mAP: 0.1312442396
  • (Sequential) stochastic-combination model mAP: 0.1542089094
  • (Parallel) stochastic-combination model mAP: 0.1542089094
Results with 500 train users and 10 test users
  • Songs: 16785
  • New songs: 244
  • Elapsed time for (Sequential) user-based model: 1090257ms (1090257841293ns)
  • Elapsed time for (Parallel) user-based model: 425118ms (425118696144ns)
  • Elapsed time for (Sequential) item-based model: 823119ms (823119431191ns)
  • Elapsed time for (Parallel) item-based model: 330385ms (330385806410ns)
  • Elapsed time for (Sequential) linear-combination model: 61ms (61367983ns)
  • Elapsed time for (Parallel) linear-combination model: 36ms (36519326ns)
  • Elapsed time for (Sequential) aggregation model: 67ms (67567455ns)
  • Elapsed time for (Parallel) aggregation model: 44ms (44929172ns)
  • Elapsed time for (Sequential) stochastic-combination model: 110ms (110850397ns)
  • Elapsed time for (Parallel) stochastic-combination model: 78ms (78716238ns)
  • Elapsed time for (Sequential) user-based model mAP: 971ms (971745933ns)
  • Elapsed time for (Parallel) user-based model mAP: 614ms (614552047ns)
  • Elapsed time for (Sequential) item-based model mAP: 1060ms (1060226717ns)
  • Elapsed time for (Parallel) item-based model mAP: 501ms (501448199ns)
  • Elapsed time for (Sequential) linear-combination model mAP: 963ms (963367047ns)
  • Elapsed time for (Parallel) linear-combination model mAP: 467ms (467187741ns)
  • Elapsed time for (Sequential) aggregation model mAP: 1031ms (1031745672ns)
  • Elapsed time for (Parallel) aggregation model mAP: 410ms (410641312ns)
  • Elapsed time for (Sequential) stochastic-combination model mAP: 740ms (740897767ns)
  • Elapsed time for (Parallel) stochastic-combination model mAP: 290ms (290752657ns)
  • (Sequential) user-based model mAP: 0.1661316029
  • (Parallel) user-based model mAP: 0.1661316029
  • (Sequential) item-based model mAP: 0.280976776
  • (Parallel) item-based model mAP: 0.280976776
  • (Sequential) linear-combination model mAP: 0.2721116315
  • (Parallel) linear-combination model mAP: 0.2721116315
  • (Sequential) aggregation model model mAP: 0.2748162243
  • (Parallel) aggregation model mAP: 0.2748162243
  • (Sequential) stochastic-combination model mAP: 0.2099743039
  • (Parallel) stochastic-combination model mAP: 0.210208496
GCP with 100 train users and 10 test users
  • Songs: 4798
  • New songs: 262
  • Elapsed time for (Sequential) user-based model: 58175ms (58175637838ns)
  • Elapsed time for (Parallel) user-based model: 53851ms (53851335760ns)
  • Elapsed time for (Distributed) user-based1: 32020ms (32020236095ns)
  • Elapsed time for (Distributed) user-based2: 42351ms (42351342324ns)
  • Elapsed time for (Sequential) item-based model: 95395ms (95395813912ns)
  • Elapsed time for (Parallel) item-based model: 52474ms (52474346061ns)
  • Elapsed time for (Distributed) item-based1: 7790ms (7790517548ns)
  • Elapsed time for (Distributed) item-based2: 7126ms (7126385975ns)
  • Elapsed time for (Sequential) linear-combination model: 43ms (43001399ns)
  • Elapsed time for (Parallel) linear-combination model: 30ms (30947247ns)
  • Elapsed time for (Distributed) linear combination: 929ms (929160615ns)
  • Elapsed time for (Sequential) aggregation model: 37ms (37521982ns)
  • Elapsed time for (Parallel) aggregation model: 71ms (71014601ns)
  • Elapsed time for (Distributed) aggregation model: 658ms (658879738ns)
  • Elapsed time for (Sequential) stochastic-combination model: 29ms (29390764ns)
  • Elapsed time for (Parallel) stochastic-combination model: 21ms (21979575ns)
  • Elapsed time for (Distributed) stochastic combination model: 484ms (484372595ns)
  • Elapsed time for (Sequential) user-based model mAP: 452ms (452958276ns)
  • Elapsed time for (Parallel) user-based model mAP: 207ms (207369356ns)
  • Elapsed time for (Distributed) user-based model mAP: 635ms (635369429ns)
  • Elapsed time for (Sequential) item-based model mAP: 576ms (576920448ns)
  • Elapsed time for (Parallel) item-based model mAP: 355ms (355273819ns)
  • Elapsed time for (Distributed) item-based model mAP: 387ms (387956319ns)
  • Elapsed time for (Sequential) linear-combination model mAP: 510ms (510430481ns)
  • Elapsed time for (Parallel) linear-combination model mAP: 299ms (299033756ns)
  • Elapsed time for (Distributed) linear-combination model mAP: 377ms (377338776ns)
  • Elapsed time for (Sequential) aggregation model mAP: 395ms (395335099ns)
  • Elapsed time for (Parallel) aggregation model mAP: 216ms (216700655ns)
  • Elapsed time for (Distributed) aggregation model mAP: 425ms (425087107ns)
  • Elapsed time for (Sequential) stochastic-combination model mAP: 339ms (339214053ns)
  • Elapsed time for (Parallel) stochastic-combination model mAP: 195ms (195296050ns)
  • Elapsed time for (Distributed) stochastic-combination model mAP: 343ms (343252870ns)
  • (Sequential) user-based model mAP: 0.0618047983
  • (Parallel) user-based model mAP: 0.0618047983
  • (Distributed) user-based model mAP: 0.0618047983
  • (Sequential) item-based model mAP: 0.0990458015
  • (Parallel) item-based model mAP: 0.0990458015
  • (Distributed) item-based model mAP: 0.0990458015
  • (Sequential) linear-combination model mAP: 0.1025445293
  • (Parallel) linear-combination model mAP: 0.1025445293
  • (Distributed) linear-combination model mAP: 0.1025445293
  • (Sequential) aggregation model model mAP: 0.0637131952
  • (Parallel) aggregation model mAP: 0.0637131952
  • (Distributed) aggregation model model mAP: 0.0637131952
  • (Sequential) stochastic-combination model mAP: 0.0682569975
  • (Parallel) stochastic-combination model mAP: 0.0751272265
  • (Distributed) stochastic-combination model mAP: 0.069692839
GCP with 100 train users and 50 test users
  • Songs: 5679
  • New songs: 1394
  • Elapsed time for (Sequential) user-based model: 655528ms (655528740788ns)
  • Elapsed time for (Parallel) user-based model: 359120ms (359120014416ns)
  • Elapsed time for (Distributed) user-based1: 95183ms (95183929502ns)
  • Elapsed time for (Distributed) user-based2: 213722ms (213722236503ns)
  • Elapsed time for (Sequential) item-based model: 732643ms (732643430327ns)
  • Elapsed time for (Parallel) item-based model: 376764ms (376764789156ns)
  • Elapsed time for (Distributed) item-based1: 98225ms (98225442292ns)
  • Elapsed time for (Distributed) item-based2: 27701ms (27701237273ns)
  • Elapsed time for (Sequential) linear-combination model: 114ms (114233213ns)
  • Elapsed time for (Parallel) linear-combination model: 103ms (103899498ns)
  • Elapsed time for (Distributed) linear combination: 2560ms (2560212721ns)
  • Elapsed time for (Sequential) aggregation model: 283ms (283632211ns)
  • Elapsed time for (Parallel) aggregation model: 240ms (240741137ns)
  • Elapsed time for (Distributed) aggregation model: 2439ms (2439683732ns)
  • Elapsed time for (Sequential) stochastic-combination model: 148ms (148734243ns)
  • Elapsed time for (Parallel) stochastic-combination model: 98ms (98070158ns)
  • Elapsed time for (Distributed) stochastic combination model: 2060ms (2060960893ns)
  • Elapsed time for (Sequential) user-based model mAP: 6940ms (6940301366ns)
  • Elapsed time for (Parallel) user-based model mAP: 3809ms (3809971695ns)
  • Elapsed time for (Distributed) user-based model mAP: 4465ms (4465218766ns)
  • Elapsed time for (Sequential) item-based model mAP: 7571ms (7571172538ns)
  • Elapsed time for (Parallel) item-based model mAP: 3945ms (3945749021ns)
  • Elapsed time for (Distributed) item-based model mAP: 4964ms (4964142348ns)
  • Elapsed time for (Sequential) linear-combination model mAP: 7336ms (7336390079ns)
  • Elapsed time for (Parallel) linear-combination model mAP: 3966ms (3966774390ns)
  • Elapsed time for (Distributed) linear-combination model mAP: 4092ms (4092735442ns)
  • Elapsed time for (Sequential) aggregation model mAP: 8849ms (8849501325ns)
  • Elapsed time for (Parallel) aggregation model mAP: 4630ms (4630480849ns)
  • Elapsed time for (Distributed) aggregation model mAP: 5817ms (5817036211ns)
  • Elapsed time for (Sequential) stochastic-combination model mAP: 6581ms (6581419016ns)
  • Elapsed time for (Parallel) stochastic-combination model mAP: 3488ms (3488122302ns)
  • Elapsed time for (Distributed) stochastic-combination model mAP: 3103ms (3103679799ns)
  • (Sequential) user-based model mAP: 0.0320140683
  • (Parallel) user-based model mAP: 0.0320140683
  • (Distributed) user-based model mAP: 0.0320140683
  • (Sequential) item-based model mAP: 0.0730435327
  • (Parallel) item-based model mAP: 0.0730435327
  • (Distributed) item-based model mAP: 0.0730435327
  • (Sequential) linear-combination model mAP: 0.0732291573
  • (Parallel) linear-combination model mAP: 0.0732291573
  • (Distributed) linear-combination model mAP: 0.0732291573
  • (Sequential) aggregation model model mAP: 0.0461427795
  • (Parallel) aggregation model mAP: 0.0461427795
  • (Distributed) aggregation model model mAP: 0.0461427795
  • (Sequential) stochastic-combination model mAP: 0.0544634029
  • (Parallel) stochastic-combination model mAP: 0.0468094833
  • (Distributed) stochastic-combination model mAP: 0.0528467816
GCP with 100 train users and 100 test users
  • Songs: 6620
  • New songs: 2480
  • Elapsed time for (Sequential) user-based model: 1453406ms (1453406413461ns)
  • Elapsed time for (Parallel) user-based model: 733755ms (733755635985ns)
  • Elapsed time for (Distributed) user-based1: 147121ms (147121170586ns)
  • Elapsed time for (Distributed) user-based2: 424269ms (424269381726ns)
  • Elapsed time for (Sequential) item-based model: 1803859ms (1803859724567ns)
  • Elapsed time for (Parallel) item-based model: 1061279ms (1061279555871ns)
  • Elapsed time for (Distributed) item-based1: 193660ms (193660286090ns)
  • Elapsed time for (Distributed) item-based2: 68685ms (68685996627ns)
  • Elapsed time for (Sequential) linear-combination model: 305ms (305689384ns)
  • Elapsed time for (Parallel) linear-combination model: 403ms (403991220ns)
  • Elapsed time for (Distributed) linear combination: 4315ms (4315320767ns)
  • Elapsed time for (Sequential) aggregation model: 457ms (457062157ns)
  • Elapsed time for (Parallel) aggregation model: 386ms (386013535ns)
  • Elapsed time for (Distributed) aggregation model: 6069ms (6069394901ns)
  • Elapsed time for (Sequential) stochastic-combination model: 577ms (577378045ns)
  • Elapsed time for (Parallel) stochastic-combination model: 389ms (389767363ns)
  • Elapsed time for (Distributed) stochastic combination model: 3783ms (3783811305ns)
  • Elapsed time for (Sequential) user-based model mAP: 20357ms (20357906557ns)
  • Elapsed time for (Parallel) user-based model mAP: 10999ms (10999264390ns)
  • Elapsed time for (Distributed) user-based model mAP: 10193ms (10193798538ns)
  • Elapsed time for (Sequential) item-based model mAP: 23144ms (23144904340ns)
  • Elapsed time for (Parallel) item-based model mAP: 12996ms (12996608280ns)
  • Elapsed time for (Distributed) item-based model mAP: 11333ms (11333303544ns)
  • Elapsed time for (Sequential) linear-combination model mAP: 22430ms (22430335776ns)
  • Elapsed time for (Parallel) linear-combination model mAP: 11414ms (11414745953ns)
  • Elapsed time for (Distributed) linear-combination model mAP: 18913ms (18913232322ns)
  • Elapsed time for (Sequential) aggregation model mAP: 24708ms (24708000927ns)
  • Elapsed time for (Parallel) aggregation model mAP: 13230ms (13230178771ns)
  • Elapsed time for (Distributed) aggregation model mAP: 18505ms (18505648503ns)
  • Elapsed time for (Sequential) stochastic-combination model mAP: 20616ms (20616275484ns)
  • Elapsed time for (Parallel) stochastic-combination model mAP: 10746ms (10746474657ns)
  • Elapsed time for (Distributed) stochastic-combination model mAP: 11020ms (11020421021ns)
  • (Sequential) user-based model mAP: 0.0165354986
  • (Parallel) user-based model mAP: 0.0165354986
  • (Distributed) user-based model mAP: 0.0165354986
  • (Sequential) item-based model mAP: 0.0394516174
  • (Parallel) item-based model mAP: 0.0394516174
  • (Distributed) item-based model mAP: 0.0394516174
  • (Sequential) linear-combination model mAP: 0.0404551622
  • (Parallel) linear-combination model mAP: 0.0404551622
  • (Distributed) linear-combination model mAP: 0.0404551622
  • (Sequential) aggregation model model mAP: 0.026037198
  • (Parallel) aggregation model mAP: 0.026037198
  • (Distributed) aggregation model model mAP: 0.026037198
  • (Sequential) stochastic-combination model mAP: 0.0301067135
  • (Parallel) stochastic-combination model mAP: 0.0280217284
  • (Distributed) stochastic-combination model mAP: 0.0290859303
GCP with 200 train users and 10 test users
  • Songs: 8763
  • New songs: 180
  • Elapsed time for (Sequential) user-based model: 272104ms (272104848879ns)
  • Elapsed time for (Parallel) user-based model: 146988ms (146988939783ns)
  • Elapsed time for (Distributed) user-based1: 59725ms (59725581508ns)
  • Elapsed time for (Distributed) user-based2: 95462ms (95462669753ns)
  • Elapsed time for (Sequential) item-based model: 249830ms (249830617586ns)
  • Elapsed time for (Parallel) item-based model: 113650ms (113650211767ns)
  • Elapsed time for (Distributed) item-based1: 47734ms (47734644822ns)
  • Elapsed time for (Distributed) item-based2: 18244ms (18244475993ns)
  • Elapsed time for (Sequential) linear-combination model: 47ms (47410182ns)
  • Elapsed time for (Parallel) linear-combination model: 32ms (32178193ns)
  • Elapsed time for (Distributed) linear combination: 1604ms (1604338317ns)
  • Elapsed time for (Sequential) aggregation model: 69ms (69162881ns)
  • Elapsed time for (Parallel) aggregation model: 49ms (49962981ns)
  • Elapsed time for (Distributed) aggregation model: 1231ms (1231156378ns)
  • Elapsed time for (Sequential) stochastic-combination model: 42ms (42390397ns)
  • Elapsed time for (Parallel) stochastic-combination model: 45ms (45704704ns)
  • Elapsed time for (Distributed) stochastic combination model: 1137ms (1137404684ns)
  • Elapsed time for (Sequential) user-based model mAP: 467ms (467654374ns)
  • Elapsed time for (Parallel) user-based model mAP: 371ms (371685224ns)
  • Elapsed time for (Distributed) user-based model mAP: 1290ms (1290212320ns)
  • Elapsed time for (Sequential) item-based model mAP: 862ms (862356942ns)
  • Elapsed time for (Parallel) item-based model mAP: 520ms (520080699ns)
  • Elapsed time for (Distributed) item-based model mAP: 1090ms (1090271325ns)
  • Elapsed time for (Sequential) linear-combination model mAP: 709ms (709940618ns)
  • Elapsed time for (Parallel) linear-combination model mAP: 469ms (469383492ns)
  • Elapsed time for (Distributed) linear-combination model mAP: 741ms (741782964ns)
  • Elapsed time for (Sequential) aggregation model mAP: 578ms (578254604ns)
  • Elapsed time for (Parallel) aggregation model mAP: 390ms (390699269ns)
  • Elapsed time for (Distributed) aggregation model mAP: 550ms (550742238ns)
  • Elapsed time for (Sequential) stochastic-combination model mAP: 574ms (574419748ns)
  • Elapsed time for (Parallel) stochastic-combination model mAP: 272ms (272123627ns)
  • Elapsed time for (Distributed) stochastic-combination model mAP: 617ms (617888150ns)
  • (Sequential) user-based model mAP: 0.0674074074
  • (Parallel) user-based model mAP: 0.0674074074
  • (Distributed) user-based model mAP: 0.0674074074
  • (Sequential) item-based model mAP: 0.1117592593
  • (Parallel) item-based model mAP: 0.1117592593
  • (Distributed) item-based model mAP: 0.1117592593
  • (Sequential) linear-combination model mAP: 0.1085185185
  • (Parallel) linear-combination model mAP: 0.1085185185
  • (Distributed) linear-combination model mAP: 0.1085185185
  • (Sequential) aggregation model model mAP: 0.1066666667
  • (Parallel) aggregation model mAP: 0.1066666667
  • (Distributed) aggregation model model mAP: 0.1066666667
  • (Sequential) stochastic-combination model mAP: 0.1040740741
  • (Parallel) stochastic-combination model mAP: 0.0989814815
  • (Distributed) stochastic-combination model mAP: 0.0757407407
GCP with 300 train users and 10 test users
  • Songs: 11660
  • New songs: 299
  • Elapsed time for (Sequential) user-based model: 513474ms (513474835482ns)
  • Elapsed time for (Parallel) user-based model: 220897ms (220897243579ns)
  • Elapsed time for (Distributed) user-based1: 162974ms (162974656940ns)
  • Elapsed time for (Distributed) user-based2: 202068ms (202068498129ns)
  • Elapsed time for (Sequential) item-based model: 635022ms (635022705931ns)
  • Elapsed time for (Parallel) item-based model: 378365ms (378365846228ns)
  • Elapsed time for (Distributed) item-based1: 109075ms (109075509142ns)
  • Elapsed time for (Distributed) item-based2: 42347ms (42347346322ns)
  • Elapsed time for (Sequential) linear-combination model: 60ms (60638515ns)
  • Elapsed time for (Parallel) linear-combination model: 49ms (49544830ns)
  • Elapsed time for (Distributed) linear combination: 1704ms (1704897767ns)
  • Elapsed time for (Sequential) aggregation model: 76ms (76549110ns)
  • Elapsed time for (Distributed) aggregation model: 1223ms (1223364907ns)
  • Elapsed time for (Parallel) aggregation model: 46ms (46226635ns)
  • Elapsed time for (Sequential) stochastic-combination model: 61ms (61203819ns)
  • Elapsed time for (Parallel) stochastic-combination model: 43ms (43150582ns)
  • Elapsed time for (Distributed) stochastic combination model: 1176ms (1176701812ns)
  • Elapsed time for (Sequential) user-based model mAP: 1003ms (1003983244ns)
  • Elapsed time for (Parallel) user-based model mAP: 524ms (524008462ns)
  • Elapsed time for (Distributed) user-based model mAP: 1136ms (1136203413ns)
  • Elapsed time for (Sequential) item-based model mAP: 1385ms (1385536381ns)
  • Elapsed time for (Parallel) item-based model mAP: 756ms (756806573ns)
  • Elapsed time for (Distributed) item-based model mAP: 1164ms (1164101859ns)
  • Elapsed time for (Sequential) linear-combination model mAP: 1122ms (1122234784ns)
  • Elapsed time for (Parallel) linear-combination model mAP: 706ms (706187033ns)
  • Elapsed time for (Distributed) linear-combination model mAP: 1074ms (1074936398ns)
  • Elapsed time for (Sequential) aggregation model mAP: 958ms (958655855ns)
  • Elapsed time for (Parallel) aggregation model mAP: 510ms (510040426ns)
  • Elapsed time for (Distributed) aggregation model mAP: 985ms (985196120ns)
  • Elapsed time for (Sequential) stochastic-combination model mAP: 929ms (929581272ns)
  • Elapsed time for (Parallel) stochastic-combination model mAP: 568ms (568268995ns)
  • Elapsed time for (Distributed) stochastic-combination model mAP: 1476ms (1476176598ns)
  • (Sequential) user-based model mAP: 0.1521579869
  • (Parallel) user-based model mAP: 0.1521579869
  • (Distributed) user-based model mAP: 0.1521579869
  • (Sequential) item-based model mAP: 0.2049370919
  • (Parallel) item-based model mAP: 0.2049370919
  • (Distributed) item-based model mAP: 0.2049370919
  • (Sequential) linear-combination model mAP: 0.2063306259
  • (Parallel) linear-combination model mAP: 0.2063306259
  • (Distributed) linear-combination model mAP: 0.2063306259
  • (Sequential) aggregation model model mAP: 0.1652810957
  • (Parallel) aggregation model mAP: 0.1652810957
  • (Distributed) aggregation model model mAP: 0.1652810957
  • (Sequential) stochastic-combination model mAP: 0.1789417105
  • (Parallel) stochastic-combination model mAP: 0.1798375538
  • (Distributed) stochastic-combination model mAP: 0.1719222806
GCP with 400 train users and 10 test users
  • Songs: 14497
  • New songs: 155
  • Elapsed time for (Sequential) user-based model: 832376ms (832376707399ns)
  • Elapsed time for (Parallel) user-based model: 384275ms (384275605015ns)
  • Elapsed time for (Distributed) user-based1: 155877ms (155877622249ns)
  • Elapsed time for (Distributed) user-based2: 125346ms (125346254580ns)
  • Elapsed time for (Sequential) item-based model: 625668ms (625668772218ns)
  • Elapsed time for (Parallel) item-based model: 301811ms (301811331916ns)
  • Elapsed time for (Distributed) item-based1: 70841ms (70841287114ns)
  • Elapsed time for (Distributed) item-based2: 66695ms (66695686731ns)
  • Elapsed time for (Sequential) linear-combination model: 81ms (81285222ns)
  • Elapsed time for (Parallel) linear-combination model: 131ms (131164707ns)
  • Elapsed time for (Distributed) linear combination: 1721ms (1721914413ns)
  • Elapsed time for (Sequential) aggregation model: 101ms (101266968ns)
  • Elapsed time for (Parallel) aggregation model: 84ms (84696671ns)
  • Elapsed time for (Distributed) aggregation model: 1851ms (1851172322ns)
  • Elapsed time for (Sequential) stochastic-combination model: 70ms (70677752ns)
  • Elapsed time for (Parallel) stochastic-combination model: 67ms (67535615ns)
  • Elapsed time for (Distributed) stochastic combination model: 1270ms (1270652268ns)
  • Elapsed time for (Sequential) user-based model mAP: 826ms (826859843ns)
  • Elapsed time for (Parallel) user-based model mAP: 550ms (550445058ns)
  • Elapsed time for (Distributed) user-based model mAP: 933ms (933694767ns)
  • Elapsed time for (Sequential) item-based model mAP: 1065ms (1065890482ns)
  • Elapsed time for (Parallel) item-based model mAP: 754ms (754424823ns)
  • Elapsed time for (Distributed) item-based model mAP: 685ms (685739359ns)
  • Elapsed time for (Sequential) linear-combination model mAP: 990ms (990476621ns)
  • Elapsed time for (Parallel) linear-combination model mAP: 685ms (685955805ns)
  • Elapsed time for (Distributed) linear-combination model mAP: 885ms (885055663ns)
  • Elapsed time for (Sequential) aggregation model mAP: 800ms (800535461ns)
  • Elapsed time for (Parallel) aggregation model mAP: 574ms (574064035ns)
  • Elapsed time for (Distributed) aggregation model mAP: 598ms (598422401ns)
  • Elapsed time for (Sequential) stochastic-combination model mAP: 775ms (775346316ns)
  • Elapsed time for (Parallel) stochastic-combination model mAP: 506ms (506862046ns)
  • Elapsed time for (Distributed) stochastic-combination model mAP: 668ms (668666705ns)
  • (Sequential) user-based model mAP: 0.143609831
  • (Parallel) user-based model mAP: 0.143609831
  • (Distributed) user-based model mAP: 0.143609831
  • (Sequential) item-based model mAP: 0.1968356375
  • (Parallel) item-based model mAP: 0.1968356375
  • (Distributed) item-based model mAP: 0.1968356375
  • (Sequential) linear-combination model mAP: 0.2011367127
  • (Parallel) linear-combination model mAP: 0.2011367127
  • (Distributed) linear-combination model mAP: 0.2011367127
  • (Sequential) aggregation model model mAP: 0.1312442396
  • (Parallel) aggregation model mAP: 0.1312442396
  • (Distributed) aggregation model model mAP: 0.1312442396
  • (Sequential) stochastic-combination model mAP: 0.1812442396
  • (Parallel) stochastic-combination model mAP: 0.181781874
  • (Distributed) stochastic-combination model mAP: 0.1753302611
GCP with 500 train users and 10 test users
  • Songs: 16785
  • New songs: 244
  • Elapsed time for (Sequential) user-based model: 1635081ms (1635081337343ns)
  • Elapsed time for (Parallel) user-based model: 799185ms (799185599123ns)
  • Elapsed time for (Distributed) user-based1: 313024ms (313024202265ns)
  • Elapsed time for (Distributed) user-based2: 250710ms (250710844884ns)
  • Elapsed time for (Sequential) item-based model: 1334183ms (1334183493280ns)
  • Elapsed time for (Parallel) item-based model: 622275ms (622275174180ns)
  • Elapsed time for (Distributed) item-based1: 138590ms (138590795829ns)
  • Elapsed time for (Distributed) item-based2: 114828ms (114828347368ns)
  • Elapsed time for (Sequential) linear-combination model: 80ms (80275271ns)
  • Elapsed time for (Parallel) linear-combination model: 78ms (78162441ns)
  • Elapsed time for (Distributed) linear combination: 2095ms (2095842940ns)
  • Elapsed time for (Sequential) aggregation model: 122ms (122355039ns)
  • Elapsed time for (Parallel) aggregation model: 75ms (75601571ns)
  • Elapsed time for (Distributed) aggregation model: 2055ms (2055686834ns)
  • Elapsed time for (Sequential) stochastic-combination model: 241ms (241857960ns)
  • Elapsed time for (Parallel) stochastic-combination model: 139ms (139012745ns)
  • Elapsed time for (Distributed) stochastic combination model: 1636ms (1636384274ns)
  • Elapsed time for (Sequential) user-based model mAP: 1502ms (1502964618ns)
  • Elapsed time for (Parallel) user-based model mAP: 894ms (894087673ns)
  • Elapsed time for (Distributed) user-based model mAP: 1118ms (1118585507ns)
  • Elapsed time for (Sequential) item-based model mAP: 1426ms (1426947763ns)
  • Elapsed time for (Parallel) item-based model mAP: 835ms (835000030ns)
  • Elapsed time for (Distributed) item-based model mAP: 999ms (999477942ns)
  • Elapsed time for (Sequential) linear-combination model mAP: 1253ms (1253600616ns)
  • Elapsed time for (Parallel) linear-combination model mAP: 847ms (847565595ns)
  • Elapsed time for (Distributed) linear-combination model mAP: 1032ms (1032727448ns)
  • Elapsed time for (Sequential) aggregation model mAP: 1253ms (1253506318ns)
  • Elapsed time for (Parallel) aggregation model mAP: 681ms (681685749ns)
  • Elapsed time for (Distributed) aggregation model mAP: 1011ms (1011418927ns)
  • Elapsed time for (Sequential) stochastic-combination model mAP: 1012ms (1012065861ns)
  • Elapsed time for (Parallel) stochastic-combination model mAP: 564ms (564852782ns)
  • Elapsed time for (Distributed) stochastic-combination model mAP: 928ms (928529292ns)
  • (Sequential) user-based model mAP: 0.1661316029
  • (Parallel) user-based model mAP: 0.1661316029
  • (Distributed) user-based model mAP: 0.1661316029
  • (Sequential) item-based model mAP: 0.280976776
  • (Parallel) item-based model mAP: 0.280976776
  • (Distributed) item-based model mAP: 0.280976776
  • (Sequential) linear-combination model mAP: 0.2721116315
  • (Parallel) linear-combination model mAP: 0.2721116315
  • (Distributed) linear-combination model mAP: 0.2721116315
  • (Sequential) aggregation model model mAP: 0.2748162243
  • (Parallel) aggregation model mAP: 0.2748162243
  • (Distributed) aggregation model model mAP: 0.2748162243
  • (Sequential) stochastic-combination model mAP: 0.2300172391
  • (Parallel) stochastic-combination model mAP: 0.2035031226
  • (Distributed) stochastic-combination model mAP: 0.210296318
GCP with 1000 train users and 10 test users
  • Songs: 17748
  • New songs: 1885
  • Elapsed time for (Distributed) user-based1: 1484891ms (1484891183020ns)
  • Elapsed time for (Distributed) item-based1: 574715ms (574715471321ns)
  • Elapsed time for (Distributed) user-based2: 1420594ms (1420594878316ns)
  • Elapsed time for (Distributed) item-based2: 606498ms (606498590813ns)
  • Elapsed time for (Distributed) linear combination: 11212ms (11212173798ns)
  • Elapsed time for (Distributed) aggregation model: 12169ms (12169649857ns)
  • Elapsed time for (Distributed) stochastic combination model: 10572ms (10572380416ns)
  • Elapsed time for (Distributed) user-based model mAP: 12487ms (12487388953ns)
  • Elapsed time for (Distributed) item-based model mAP: 13396ms (13396437813ns)
  • Elapsed time for (Distributed) linear-combination model mAP: 13107ms (13107499441ns)
  • Elapsed time for (Distributed) aggregation model mAP: 12161ms (12161913984ns)
  • Elapsed time for (Distributed) stochastic-combination model mAP: 11728ms (11728255668ns)
  • (Distributed) user-based model mAP: 0.0308797122
  • (Distributed) item-based model mAP: 0.1107832411
  • (Distributed) linear-combination model mAP: 0.1128872417
  • (Distributed) aggregation model model mAP: 0.0713676829
  • (Distributed) stochastic-combination model mAP: 0.0811107242
GCP with 1000 train users and 100 test users
  • Songs: 28720
  • New songs: 2176
  • Elapsed time for (Distributed) user-based1: 6456360ms (6456360527017ns)
  • Elapsed time for (Distributed) item-based1: 2188633ms (2188633285944ns)
  • Elapsed time for (Distributed) user-based2: 5607922ms (5607922651110ns)
  • Elapsed time for (Distributed) item-based2: 2940547ms (2940547519188ns)
  • Elapsed time for (Distributed) linear combination: 24083ms (24083653467ns)
  • Elapsed time for (Distributed) aggregation model: 25625ms (25625904375ns)
  • Elapsed time for (Distributed) stochastic combination model: 23222ms (23222672280ns)
  • Elapsed time for (Distributed) user-based model mAP: 69804ms (69804751039ns)
  • Elapsed time for (Distributed) item-based model mAP: 72479ms (72479420315ns)
  • Elapsed time for (Distributed) linear-combination model mAP: 71194ms (71194940111ns)
  • Elapsed time for (Distributed) aggregation model mAP: 70002ms (70002738736ns)
  • Elapsed time for (Distributed) stochastic-combination model mAP: 69433ms (69433599409ns)
  • (Distributed) user-based model mAP: 0.0385800219
  • (Distributed) item-based model mAP: 0.142942848
  • (Distributed) linear-combination model mAP: 0.1369886698
  • (Distributed) aggregation model model mAP: 0.0950991402
  • (Distributed) stochastic-combination model mAP: 0.101490377
GCP with 2000 train users and 100 test users
  • Songs: 44451
  • New songs: 1865
  • Elapsed time for (Distributed) user-based1: 24727672ms (24727672988947ns)
  • Elapsed time for (Distributed) item-based1: 16157675ms (16157675183404ns)
  • Elapsed time for (Distributed) linear combination: 60038ms (60038994146ns)
  • Elapsed time for (Distributed) aggregation model: 61191ms (61191362477ns)
  • Elapsed time for (Distributed) stochastic combination model: 50780ms (50780938492ns)
  • Elapsed time for (Distributed) user-based model mAP: 68756ms (68756818058ns)
  • Elapsed time for (Distributed) item-based model mAP: 74989ms (74989210957ns)
  • Elapsed time for (Distributed) linear-combination model mAP: 62710ms (62710235048ns)
  • Elapsed time for (Distributed) aggregation model mAP: 62156ms (62156142078ns)
  • Elapsed time for (Distributed) stochastic-combination model mAP: 62575ms (62575104084ns)
  • (Distributed) user-based model mAP: 0.0296820445
  • (Distributed) item-based model mAP: 0.1549078628
  • (Distributed) linear-combination model mAP: 0.0923070961
  • (Distributed) aggregation model model mAP: 0.0770281485
  • (Distributed) stochastic-combination model mAP: 0.0666806146

musicrecommendation's People

Contributors

alberto-paparella avatar gitgabbo avatar

Stargazers

 avatar

Watchers

 avatar  avatar

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.