Run1 report

split method condition Hit@3 (95% CI over seeds) MRR@10 nDCG@10 Δ Hit@3 vs none [95% CI] Δ Hit@3 vs shuffled [95% CI]
within none none 0.382 [0.353, 0.411] 0.334 0.384
within contextual genuine 0.383 [0.350, 0.415] 0.338 0.387 +0.001 [-0.003, +0.004] -0.001 [-0.005, +0.004]
within contextual shuffled 0.383 [0.350, 0.416] 0.332 0.383 +0.001 [-0.004, +0.006]
within contextual noisy 0.384 [0.353, 0.415] 0.336 0.385 +0.002 [-0.003, +0.006] +0.001 [-0.004, +0.005]
within contextual+ts genuine 0.389 [0.362, 0.415] 0.335 0.385 +0.007 [+0.002, +0.012] +0.004 [-0.001, +0.009]
within contextual+ts shuffled 0.384 [0.356, 0.413] 0.334 0.384 +0.002 [-0.002, +0.007]
within contextual+ts noisy 0.384 [0.360, 0.409] 0.335 0.385 +0.002 [-0.002, +0.006] +0.000 [-0.005, +0.005]
within global genuine 0.323 [0.288, 0.357] 0.273 0.332 -0.059 [-0.079, -0.039] +0.015 [-0.006, +0.035]
within global shuffled 0.309 [0.290, 0.327] 0.257 0.317 -0.074 [-0.093, -0.054]
within global noisy 0.313 [0.265, 0.362] 0.264 0.324 -0.069 [-0.089, -0.050] +0.004 [-0.016, +0.025]
within ranker genuine 0.421 [0.394, 0.448] 0.363 0.409 +0.039 [+0.024, +0.055] +0.141 [+0.120, +0.162]
within ranker shuffled 0.280 [0.244, 0.315] 0.234 0.293 -0.102 [-0.121, -0.083]
within ranker noisy 0.330 [0.232, 0.428] 0.287 0.342 -0.055 [-0.073, -0.036] +0.047 [+0.025, +0.069]
test none none 0.386 [0.361, 0.411] 0.334 0.378
test contextual genuine 0.386 [0.361, 0.411] 0.334 0.378 +0.000 [+0.000, +0.000] +0.000 [+0.000, +0.000]
test contextual shuffled 0.386 [0.361, 0.411] 0.334 0.378 +0.000 [+0.000, +0.000]
test contextual noisy 0.386 [0.361, 0.411] 0.334 0.378 +0.000 [+0.000, +0.000] +0.000 [+0.000, +0.000]
test contextual+ts genuine 0.386 [0.361, 0.411] 0.334 0.378 +0.000 [+0.000, +0.000] +0.000 [+0.000, +0.000]
test contextual+ts shuffled 0.386 [0.361, 0.411] 0.334 0.378 +0.000 [+0.000, +0.000]
test contextual+ts noisy 0.386 [0.361, 0.411] 0.334 0.378 +0.000 [+0.000, +0.000] +0.000 [+0.000, +0.000]
test global genuine 0.386 [0.361, 0.411] 0.334 0.378 +0.000 [+0.000, +0.000] +0.000 [+0.000, +0.000]
test global shuffled 0.386 [0.361, 0.411] 0.334 0.378 +0.000 [+0.000, +0.000]
test global noisy 0.386 [0.361, 0.411] 0.334 0.378 +0.000 [+0.000, +0.000] +0.000 [+0.000, +0.000]
test ranker genuine 0.414 [0.386, 0.442] 0.361 0.404 +0.028 [+0.018, +0.038] +0.028 [+0.018, +0.038]
test ranker shuffled 0.386 [0.361, 0.411] 0.334 0.378 +0.000 [+0.000, +0.000]
test ranker noisy 0.397 [0.374, 0.420] 0.348 0.389 +0.011 [+0.005, +0.017] +0.011 [+0.005, +0.017]
method feedback phase (s, mean) eval search p50 (ms)
none 0 10.3
global 3.4 2.8
contextual 3.5 2.8
contextual+ts 3.5 2.7
ranker 3.7 2.1