论文“具有非均匀缺失数据的快速矩阵分解”的实验代码。.zip

# Fast Matrix Factorization with Non-Uniform Missing Data This is our official implementation for the paper [[url](https://arxiv.org/abs/1811.04411)]: > Xiangnan He, Jinhui Tang, Xiaoyu Du, Richang Hong, Tongwei Ren and Tat-Seng Chua, Fast Matrix Factorization with Non-Uniform Weights on Missing Data, TNNLS. It is an extended work of the SIGIR'16 paper "Fast Matrix Factorization for Online Recommendation with Implicit Feedback". If you use the codes, please cite our paper . Thanks! ## Models The core model is `src/algorithms/MF_extALS.java`. A contrast model based on sparse SVD is `svd.py`. ## Dataset We exhibit a small dataset `data/yelp.rating`, in which the columns indicate the `user_id`, `item_id`, `rating score` and `rating time`, respectively. ## Quick Start Run the main function in `src/main/main_MF.java`. ## Results The outputs with default settings are as below: ``` HoldOne out splitting. Sort items for each user.[00:00:00.470] Generate rating matrices.[00:00:01.056] Data data/yelp.rating #Users 25677 #Items 25815 #Ratings 672765 (train), 25677(test) extALS: showProgress=false, factors=64, maxIter=500, reg=0.010000, w0=10.00, alpha=0.75 ==================================================== Popularity <hr, ndcg, prec>: 0.0705 0.0176 0.0062 [00:00:02.233] training size: 662851755 user: 25677, item: 25815 relates: 1 Iter=0 [00:00:06.892] [-]loss: 261836.1007 [00:00:00.338] Iter=1 [00:00:04.821] [-]loss: 96839.4645 [00:00:00.496] Iter=2 [00:00:04.554] [-]loss: 67288.0789 [00:00:00.242] Iter=3 [00:00:04.531] [-]loss: 54636.3216 [00:00:00.225] Iter=4 [00:00:04.514] [-]loss: 47446.8384 [00:00:00.224] Iter=5 [00:00:04.506] [-]loss: 42775.1669 [00:00:00.222] Iter=6 [00:00:04.485] [-]loss: 39475.7438 [00:00:00.227] Iter=7 [00:00:04.505] [-]loss: 37002.0581 [00:00:00.226] Iter=8 [00:00:04.591] [-]loss: 35060.2324 [00:00:00.229] Iter=9 [00:00:04.546] [-]loss: 33479.9425 [00:00:00.229] Iter=10 [00:00:04.559] [-]loss: 32157.4193 [00:00:00.228] Iter=11 [00:00:04.613] [-]loss: 31027.1596 [00:00:00.231] Iter=12 [00:00:04.572] [-]loss: 30046.2365 [00:00:00.233] Iter=13 [00:00:04.538] [-]loss: 29185.1848 [00:00:00.224] Iter=14 [00:00:04.557] [-]loss: 28422.7083 [00:00:00.237] Iter=15 [00:00:04.564] [-]loss: 27742.7244 [00:00:00.225] Iter=16 [00:00:04.547] [-]loss: 27132.6735 [00:00:00.226] Iter=17 [00:00:04.588] [-]loss: 26582.4973 [00:00:00.228] Iter=18 [00:00:04.566] [-]loss: 26083.9762 [00:00:00.225] Iter=19 [00:00:04.550] [-]loss: 25630.2859 [00:00:00.228] Iter=20 [00:00:04.574] [-]loss: 25215.7115 [00:00:00.224] Iter=21 [00:00:04.535] [-]loss: 24835.4411 [00:00:00.227] Iter=22 [00:00:04.506] [-]loss: 24485.4029 [00:00:00.225] Iter=23 [00:00:04.510] [-]loss: 24162.1307 [00:00:00.223] Iter=24 [00:00:04.523] [-]loss: 23862.6545 [00:00:00.226] Iter=25 [00:00:04.503] [-]loss: 23584.4155 [00:00:00.228] Iter=26 [00:00:04.514] [-]loss: 23325.1992 [00:00:00.229] Iter=27 [00:00:04.566] [-]loss: 23083.0798 [00:00:00.227] Iter=28 [00:00:04.603] [-]loss: 22856.3758 [00:00:00.225] Iter=29 [00:00:04.632] [-]loss: 22643.6149 [00:00:00.226] Iter=30 [00:00:04.524] [-]loss: 22443.5050 [00:00:00.225] Iter=31 [00:00:04.663] [-]loss: 22254.9070 [00:00:00.227] Iter=32 [00:00:04.623] [-]loss: 22076.8149 [00:00:00.229] Iter=33 [00:00:04.663] [-]loss: 21908.3398 [00:00:00.225] Iter=34 [00:00:04.580] [-]loss: 21748.6950 [00:00:00.224] Iter=35 [00:00:04.528] [-]loss: 21597.1818 [00:00:00.227] Iter=36 [00:00:04.535] [-]loss: 21453.1786 [00:00:00.224] Iter=37 [00:00:04.541] [-]loss: 21316.1297 [00:00:00.223] Iter=38 [00:00:04.523] [-]loss: 21185.5367 [00:00:00.225] Iter=39 [00:00:04.525] [-]loss: 21060.9505 [00:00:00.227] Iter=40 [00:00:04.523] [-]loss: 20941.9652 [00:00:00.226] Iter=41 [00:00:05.074] [-]loss: 20828.2120 [00:00:00.334] Iter=42 [00:00:04.910] [-]loss: 20719.3558 [00:00:00.237] Iter=43 [00:00:04.868] [-]loss: 20615.0902 [00:00:00.225] Iter=44 [00:00:04.525] [-]loss: 20515.1353 [00:00:00.224] Iter=45 [00:00:04.791] [-]loss: 20419.2337 [00:00:00.238] Iter=46 [00:00:04.867] [-]loss: 20327.1489 [00:00:00.417] Iter=47 [00:00:04.936] [-]loss: 20238.6622 [00:00:00.239] Iter=48 [00:00:04.645] [-]loss: 20153.5715 [00:00:00.225] Iter=49 [00:00:04.534] [-]loss: 20071.6890 [00:00:00.222] Iter=50 [00:00:04.689] [-]loss: 19992.8403 [00:00:00.250] Iter=51 [00:00:04.882] [-]loss: 19916.8628 [00:00:00.322] Iter=52 [00:00:04.776] [-]loss: 19843.6050 [00:00:00.244] Iter=53 [00:00:04.818] [-]loss: 19772.9253 [00:00:00.240] Iter=54 [00:00:04.820] [-]loss: 19704.6913 [00:00:00.225] Iter=55 [00:00:04.790] [-]loss: 19638.7789 [00:00:00.227] Iter=56 [00:00:04.571] [-]loss: 19575.0719 [00:00:00.253] Iter=57 [00:00:04.769] [-]loss: 19513.4613 [00:00:00.314] Iter=58 [00:00:04.753] [-]loss: 19453.8449 [00:00:00.238] Iter=59 [00:00:04.912] [-]loss: 19396.1266 [00:00:00.232] Iter=60 [00:00:05.110] [-]loss: 19340.2162 [00:00:00.246] Iter=61 [00:00:04.720] [-]loss: 19286.0290 [00:00:00.226] Iter=62 [00:00:04.732] [-]loss: 19233.4853 [00:00:00.231] Iter=63 [00:00:04.784] [-]loss: 19182.5101 [00:00:00.228] Iter=64 [00:00:04.760] [-]loss: 19133.0329 [00:00:00.234] Iter=65 [00:00:04.780] [-]loss: 19084.9873 [00:00:00.227] Iter=66 [00:00:04.754] [-]loss: 19038.3105 [00:00:00.235] Iter=67 [00:00:04.734] [-]loss: 18992.9438 [00:00:00.276] Iter=68 [00:00:04.690] [-]loss: 18948.8313 [00:00:00.243] Iter=69 [00:00:04.632] [-]loss: 18905.9206 [00:00:00.238] Iter=70 [00:00:04.610] [-]loss: 18864.1622 [00:00:00.230] Iter=71 [00:00:04.743] [-]loss: 18823.5089 [00:00:00.258] Iter=72 [00:00:04.758] [-]loss: 18783.9166 [00:00:00.223] Iter=73 [00:00:04.707] [-]loss: 18745.3430 [00:00:00.300] Iter=74 [00:00:04.808] [-]loss: 18707.7481 [00:00:00.223] Iter=75 [00:00:04.831] [-]loss: 18671.0940 [00:00:00.230] Iter=76 [00:00:04.913] [-]loss: 18635.3447 [00:00:00.331] Iter=77 [00:00:05.078] [-]loss: 18600.4658 [00:00:00.241] Iter=78 [00:00:04.781] [-]loss: 18566.4247 [00:00:00.225] Iter=79 [00:00:04.822] [-]loss: 18533.1906 [00:00:00.236] Iter=80 [00:00:04.659] [-]loss: 18500.7338 [00:00:00.226] Iter=81 [00:00:04.599] [-]loss: 18469.0264 [00:00:00.225] Iter=82 [00:00:04.580] [-]loss: 18438.0416 [00:00:00.228] Iter=83 [00:00:04.541] [-]loss: 18407.7543 [00:00:00.225] Iter=84 [00:00:04.760] [-]loss: 18378.1401 [00:00:00.226] Iter=85 [00:00:04.693] [-]loss: 18349.1762 [00:00:00.282] Iter=86 [00:00:05.050] [-]loss: 18320.8408 [00:00:00.308] Iter=87 [00:00:04.885] [-]loss: 18293.1130 [00:00:00.232] Iter=88 [00:00:04.696] [-]loss: 18265.9733 [00:00:00.226] Iter=89 [00:00:04.664] [-]loss: 18239.4026 [00:00:00.247] Iter=90 [00:00:04.505] [-]loss: 18213.3829 [00:00:00.226] Iter=91 [00:00:04.505] [-]loss: 18187.8971 [00:00:00.225] Iter=92 [00:00:04.520] [-]loss: 18162.9285 [00:00:00.226] Iter=93 [00:00:04.557] [-]loss: 18138.4613 [00:00:00.223] Iter=94 [00:00:04.946] [-]loss: 18114.4802 [00:00:00.225] Iter=95 [00:00:04.745] [-]loss: 18090.9703 [00:00:00.285] Iter=96 [00:00:04.725] [-]loss: 18067.9176 [00:00:00.259] Iter=97 [00:00:04.788] [-]loss: 18045.3084 [00:00:00.228] Iter=98 [00:00:04.874] [-]loss: 18023.1294 [00:00:00.234] Iter=99 [00:00:04.574] [-]loss: 18001.3682 [00:00:00.225] Iter=100 [00:00:04.542] [-]loss: 17980.0124 [00:00:00.225] Iter=101 [00:00:04.664] [-]loss: 17959.0506 [00:00:00.270] Iter=102 [00:00:04.820] [-]loss: 17938.4716 [00:00:00.226] Iter=103 [00:00:04.527] [-]loss: 17918.2647 [00:00:00.225] Iter=104 [00:00:04.591] [-]loss: 17898.4195 [00:00:00.240] Iter=105 [00:00:04.884] [-]loss: 17878.9264 [00:00:00.263] Iter=106 [00:00:04.626] [-]loss: 17859.7758 [00:00:00.227] Iter=107 [00:00:04.872] [-]loss: 17840.9588 [00:00:00.227] Iter=108 [00:00:04.615] [