bors[bot]
ad4c982c68
Merge #439
...
439: Optimize typo criterion r=Kerollmops a=MarinPostma
This pr implements a couple of optimization for the typo criterion:
- clamp max typo on concatenated query words to 1: By considering that a concatenated query word is a typo, we clamp the max number of typos allowed o it to 1. This is useful because we noticed that concatenated query words often introduced words with 2 typos in queries that otherwise didn't allow for 2 typo words.
- Make typos on the first letter count for 2. This change is a big performance gain: by considering the typos on the first letter to count as 2 typos, we drastically restrict the search space for 1 typo, and if we reach 2 typos, the search space is reduced as well, as we only consider: (2 typos ∩ correct first letter) ∪ (wrong first letter ∩ 1 typo) instead of 2 typos anywhere in the word.
## benches
```
group main typo
----- ---- ----
smol-songs.csv: asc + default/Notstandskomitee 2.51 5.8±0.01ms ? ?/sec 1.00 2.3±0.01ms ? ?/sec
smol-songs.csv: asc + default/charles 2.48 3.0±0.01ms ? ?/sec 1.00 1190.9±1.29µs ? ?/sec
smol-songs.csv: asc + default/charles mingus 5.56 10.8±0.01ms ? ?/sec 1.00 1935.3±1.00µs ? ?/sec
smol-songs.csv: asc + default/david 1.65 3.9±0.00ms ? ?/sec 1.00 2.4±0.01ms ? ?/sec
smol-songs.csv: asc + default/david bowie 3.34 12.5±0.02ms ? ?/sec 1.00 3.7±0.00ms ? ?/sec
smol-songs.csv: asc + default/john 1.00 1849.7±3.74µs ? ?/sec 1.01 1875.1±4.65µs ? ?/sec
smol-songs.csv: asc + default/marcus miller 4.32 15.7±0.01ms ? ?/sec 1.00 3.6±0.01ms ? ?/sec
smol-songs.csv: asc + default/michael jackson 3.31 12.5±0.01ms ? ?/sec 1.00 3.8±0.00ms ? ?/sec
smol-songs.csv: asc + default/tamo 1.05 565.4±0.86µs ? ?/sec 1.00 539.3±1.22µs ? ?/sec
smol-songs.csv: asc + default/thelonious monk 3.49 11.5±0.01ms ? ?/sec 1.00 3.3±0.00ms ? ?/sec
smol-songs.csv: asc/Notstandskomitee 2.59 5.6±0.02ms ? ?/sec 1.00 2.2±0.01ms ? ?/sec
smol-songs.csv: asc/charles 6.05 2.1±0.00ms ? ?/sec 1.00 347.8±0.60µs ? ?/sec
smol-songs.csv: asc/charles mingus 14.46 9.4±0.01ms ? ?/sec 1.00 649.2±0.97µs ? ?/sec
smol-songs.csv: asc/david 3.87 2.4±0.00ms ? ?/sec 1.00 618.2±0.69µs ? ?/sec
smol-songs.csv: asc/david bowie 10.14 9.8±0.01ms ? ?/sec 1.00 970.8±1.55µs ? ?/sec
smol-songs.csv: asc/john 1.00 546.5±1.10µs ? ?/sec 1.00 547.1±2.11µs ? ?/sec
smol-songs.csv: asc/marcus miller 11.45 10.4±0.06ms ? ?/sec 1.00 907.9±1.37µs ? ?/sec
smol-songs.csv: asc/michael jackson 10.56 9.7±0.01ms ? ?/sec 1.00 919.6±1.03µs ? ?/sec
smol-songs.csv: asc/tamo 1.03 43.3±0.18µs ? ?/sec 1.00 42.2±0.23µs ? ?/sec
smol-songs.csv: asc/thelonious monk 4.16 10.7±0.02ms ? ?/sec 1.00 2.6±0.00ms ? ?/sec
smol-songs.csv: basic filter: <=/Notstandskomitee 1.00 95.7±0.20µs ? ?/sec 1.15 109.6±10.40µs ? ?/sec
smol-songs.csv: basic filter: <=/charles 1.00 27.8±0.15µs ? ?/sec 1.01 27.9±0.18µs ? ?/sec
smol-songs.csv: basic filter: <=/charles mingus 1.72 119.2±0.67µs ? ?/sec 1.00 69.1±0.13µs ? ?/sec
smol-songs.csv: basic filter: <=/david 1.00 22.3±0.33µs ? ?/sec 1.05 23.4±0.19µs ? ?/sec
smol-songs.csv: basic filter: <=/david bowie 1.59 86.9±0.79µs ? ?/sec 1.00 54.5±0.31µs ? ?/sec
smol-songs.csv: basic filter: <=/john 1.00 17.9±0.06µs ? ?/sec 1.06 18.9±0.15µs ? ?/sec
smol-songs.csv: basic filter: <=/marcus miller 1.65 102.7±1.63µs ? ?/sec 1.00 62.3±0.18µs ? ?/sec
smol-songs.csv: basic filter: <=/michael jackson 1.76 128.2±1.85µs ? ?/sec 1.00 72.9±0.19µs ? ?/sec
smol-songs.csv: basic filter: <=/tamo 1.00 17.9±0.13µs ? ?/sec 1.05 18.7±0.20µs ? ?/sec
smol-songs.csv: basic filter: <=/thelonious monk 1.53 157.5±2.38µs ? ?/sec 1.00 102.8±0.88µs ? ?/sec
smol-songs.csv: basic filter: TO/Notstandskomitee 1.00 100.9±4.36µs ? ?/sec 1.04 105.0±8.25µs ? ?/sec
smol-songs.csv: basic filter: TO/charles 1.00 28.4±0.36µs ? ?/sec 1.03 29.4±0.33µs ? ?/sec
smol-songs.csv: basic filter: TO/charles mingus 1.71 118.1±1.08µs ? ?/sec 1.00 68.9±0.26µs ? ?/sec
smol-songs.csv: basic filter: TO/david 1.00 24.0±0.26µs ? ?/sec 1.03 24.6±0.43µs ? ?/sec
smol-songs.csv: basic filter: TO/david bowie 1.72 95.2±0.30µs ? ?/sec 1.00 55.2±0.14µs ? ?/sec
smol-songs.csv: basic filter: TO/john 1.00 18.8±0.09µs ? ?/sec 1.06 19.8±0.17µs ? ?/sec
smol-songs.csv: basic filter: TO/marcus miller 1.61 102.4±1.65µs ? ?/sec 1.00 63.4±0.24µs ? ?/sec
smol-songs.csv: basic filter: TO/michael jackson 1.77 132.1±1.41µs ? ?/sec 1.00 74.5±0.59µs ? ?/sec
smol-songs.csv: basic filter: TO/tamo 1.00 18.2±0.14µs ? ?/sec 1.05 19.2±0.46µs ? ?/sec
smol-songs.csv: basic filter: TO/thelonious monk 1.49 150.8±1.92µs ? ?/sec 1.00 101.3±0.44µs ? ?/sec
smol-songs.csv: basic placeholder/ 1.00 27.3±0.07µs ? ?/sec 1.03 28.0±0.05µs ? ?/sec
smol-songs.csv: basic with quote/"Notstandskomitee" 1.00 122.4±0.17µs ? ?/sec 1.03 125.6±0.16µs ? ?/sec
smol-songs.csv: basic with quote/"charles" 1.00 88.8±0.30µs ? ?/sec 1.00 88.4±0.15µs ? ?/sec
smol-songs.csv: basic with quote/"charles" "mingus" 1.00 685.2±0.74µs ? ?/sec 1.01 689.4±6.07µs ? ?/sec
smol-songs.csv: basic with quote/"david" 1.00 161.6±0.42µs ? ?/sec 1.01 162.6±0.17µs ? ?/sec
smol-songs.csv: basic with quote/"david" "bowie" 1.00 731.7±0.73µs ? ?/sec 1.02 743.1±0.77µs ? ?/sec
smol-songs.csv: basic with quote/"john" 1.00 267.1±0.33µs ? ?/sec 1.01 270.9±0.33µs ? ?/sec
smol-songs.csv: basic with quote/"marcus" "miller" 1.00 138.7±0.31µs ? ?/sec 1.02 140.9±0.13µs ? ?/sec
smol-songs.csv: basic with quote/"michael" "jackson" 1.01 841.4±0.72µs ? ?/sec 1.00 833.8±0.92µs ? ?/sec
smol-songs.csv: basic with quote/"tamo" 1.01 189.2±0.26µs ? ?/sec 1.00 188.2±0.71µs ? ?/sec
smol-songs.csv: basic with quote/"thelonious" "monk" 1.00 1100.5±1.36µs ? ?/sec 1.01 1111.7±2.17µs ? ?/sec
smol-songs.csv: basic without quote/Notstandskomitee 3.40 7.9±0.02ms ? ?/sec 1.00 2.3±0.02ms ? ?/sec
smol-songs.csv: basic without quote/charles 2.57 494.4±0.89µs ? ?/sec 1.00 192.5±0.18µs ? ?/sec
smol-songs.csv: basic without quote/charles mingus 1.29 2.8±0.02ms ? ?/sec 1.00 2.1±0.01ms ? ?/sec
smol-songs.csv: basic without quote/david 1.95 623.8±0.90µs ? ?/sec 1.00 319.2±1.22µs ? ?/sec
smol-songs.csv: basic without quote/david bowie 1.12 5.9±0.00ms ? ?/sec 1.00 5.2±0.00ms ? ?/sec
smol-songs.csv: basic without quote/john 1.24 1340.9±2.25µs ? ?/sec 1.00 1084.7±7.76µs ? ?/sec
smol-songs.csv: basic without quote/marcus miller 7.97 14.6±0.01ms ? ?/sec 1.00 1826.0±6.84µs ? ?/sec
smol-songs.csv: basic without quote/michael jackson 1.19 3.9±0.00ms ? ?/sec 1.00 3.3±0.00ms ? ?/sec
smol-songs.csv: basic without quote/tamo 1.65 737.7±3.58µs ? ?/sec 1.00 446.7±0.51µs ? ?/sec
smol-songs.csv: basic without quote/thelonious monk 1.16 4.5±0.02ms ? ?/sec 1.00 3.9±0.04ms ? ?/sec
smol-songs.csv: big filter/Notstandskomitee 3.27 7.6±0.02ms ? ?/sec 1.00 2.3±0.01ms ? ?/sec
smol-songs.csv: big filter/charles 8.26 1957.5±1.37µs ? ?/sec 1.00 236.8±0.34µs ? ?/sec
smol-songs.csv: big filter/charles mingus 18.49 11.2±0.06ms ? ?/sec 1.00 607.7±3.03µs ? ?/sec
smol-songs.csv: big filter/david 3.78 2.4±0.00ms ? ?/sec 1.00 622.8±0.80µs ? ?/sec
smol-songs.csv: big filter/david bowie 9.00 12.0±0.01ms ? ?/sec 1.00 1336.0±3.17µs ? ?/sec
smol-songs.csv: big filter/john 1.00 554.2±0.95µs ? ?/sec 1.01 560.4±0.79µs ? ?/sec
smol-songs.csv: big filter/marcus miller 18.09 12.0±0.01ms ? ?/sec 1.00 664.7±0.60µs ? ?/sec
smol-songs.csv: big filter/michael jackson 8.43 12.0±0.01ms ? ?/sec 1.00 1421.6±1.37µs ? ?/sec
smol-songs.csv: big filter/tamo 1.00 86.3±0.14µs ? ?/sec 1.01 87.3±0.21µs ? ?/sec
smol-songs.csv: big filter/thelonious monk 5.55 14.3±0.02ms ? ?/sec 1.00 2.6±0.01ms ? ?/sec
smol-songs.csv: desc + default/Notstandskomitee 2.52 5.8±0.01ms ? ?/sec 1.00 2.3±0.01ms ? ?/sec
smol-songs.csv: desc + default/charles 3.04 2.7±0.01ms ? ?/sec 1.00 893.4±1.08µs ? ?/sec
smol-songs.csv: desc + default/charles mingus 6.77 10.3±0.01ms ? ?/sec 1.00 1520.8±1.90µs ? ?/sec
smol-songs.csv: desc + default/david 1.39 5.7±0.00ms ? ?/sec 1.00 4.1±0.00ms ? ?/sec
smol-songs.csv: desc + default/david bowie 2.34 15.8±0.02ms ? ?/sec 1.00 6.7±0.01ms ? ?/sec
smol-songs.csv: desc + default/john 1.00 2.5±0.00ms ? ?/sec 1.02 2.6±0.01ms ? ?/sec
smol-songs.csv: desc + default/marcus miller 5.06 14.5±0.02ms ? ?/sec 1.00 2.9±0.01ms ? ?/sec
smol-songs.csv: desc + default/michael jackson 2.64 14.1±0.05ms ? ?/sec 1.00 5.4±0.00ms ? ?/sec
smol-songs.csv: desc + default/tamo 1.00 567.0±0.65µs ? ?/sec 1.00 565.7±0.97µs ? ?/sec
smol-songs.csv: desc + default/thelonious monk 3.55 11.6±0.02ms ? ?/sec 1.00 3.3±0.00ms ? ?/sec
smol-songs.csv: desc/Notstandskomitee 2.58 5.6±0.02ms ? ?/sec 1.00 2.2±0.02ms ? ?/sec
smol-songs.csv: desc/charles 6.04 2.1±0.00ms ? ?/sec 1.00 348.1±0.57µs ? ?/sec
smol-songs.csv: desc/charles mingus 14.51 9.4±0.01ms ? ?/sec 1.00 646.7±0.99µs ? ?/sec
smol-songs.csv: desc/david 3.86 2.4±0.00ms ? ?/sec 1.00 620.7±2.46µs ? ?/sec
smol-songs.csv: desc/david bowie 10.10 9.8±0.01ms ? ?/sec 1.00 973.9±3.31µs ? ?/sec
smol-songs.csv: desc/john 1.00 545.5±0.78µs ? ?/sec 1.00 547.2±0.48µs ? ?/sec
smol-songs.csv: desc/marcus miller 11.39 10.3±0.01ms ? ?/sec 1.00 903.7±0.95µs ? ?/sec
smol-songs.csv: desc/michael jackson 10.51 9.7±0.01ms ? ?/sec 1.00 924.7±2.02µs ? ?/sec
smol-songs.csv: desc/tamo 1.01 43.2±0.33µs ? ?/sec 1.00 42.6±0.35µs ? ?/sec
smol-songs.csv: desc/thelonious monk 4.19 10.8±0.03ms ? ?/sec 1.00 2.6±0.00ms ? ?/sec
smol-songs.csv: prefix search/a 1.00 1008.7±1.00µs ? ?/sec 1.00 1005.5±0.91µs ? ?/sec
smol-songs.csv: prefix search/b 1.00 885.0±0.70µs ? ?/sec 1.01 890.6±1.11µs ? ?/sec
smol-songs.csv: prefix search/i 1.00 1051.8±1.25µs ? ?/sec 1.00 1056.6±4.12µs ? ?/sec
smol-songs.csv: prefix search/s 1.00 724.7±1.77µs ? ?/sec 1.00 721.6±0.59µs ? ?/sec
smol-songs.csv: prefix search/x 1.01 212.4±0.21µs ? ?/sec 1.00 210.9±0.38µs ? ?/sec
smol-songs.csv: proximity/7000 Danses Un Jour Dans Notre Vie 18.55 48.5±0.09ms ? ?/sec 1.00 2.6±0.03ms ? ?/sec
smol-songs.csv: proximity/The Disneyland Sing-Along Chorus 8.41 56.7±0.45ms ? ?/sec 1.00 6.7±0.05ms ? ?/sec
smol-songs.csv: proximity/Under Great Northern Lights 15.74 38.9±0.14ms ? ?/sec 1.00 2.5±0.00ms ? ?/sec
smol-songs.csv: proximity/black saint sinner lady 11.82 40.1±0.13ms ? ?/sec 1.00 3.4±0.02ms ? ?/sec
smol-songs.csv: proximity/les dangeureuses 1960 6.90 26.1±0.13ms ? ?/sec 1.00 3.8±0.04ms ? ?/sec
smol-songs.csv: typo/Arethla Franklin 14.93 5.8±0.01ms ? ?/sec 1.00 390.1±1.89µs ? ?/sec
smol-songs.csv: typo/Disnaylande 3.18 7.3±0.01ms ? ?/sec 1.00 2.3±0.00ms ? ?/sec
smol-songs.csv: typo/dire straights 5.55 15.2±0.02ms ? ?/sec 1.00 2.7±0.00ms ? ?/sec
smol-songs.csv: typo/fear of the duck 28.03 20.0±0.03ms ? ?/sec 1.00 713.3±1.54µs ? ?/sec
smol-songs.csv: typo/indochie 19.25 1851.4±2.38µs ? ?/sec 1.00 96.2±0.13µs ? ?/sec
smol-songs.csv: typo/indochien 14.66 1887.7±3.18µs ? ?/sec 1.00 128.8±0.18µs ? ?/sec
smol-songs.csv: typo/klub des loopers 37.73 18.0±0.02ms ? ?/sec 1.00 476.7±0.73µs ? ?/sec
smol-songs.csv: typo/michel depech 10.17 5.8±0.01ms ? ?/sec 1.00 565.8±1.16µs ? ?/sec
smol-songs.csv: typo/mongus 15.33 1897.4±3.44µs ? ?/sec 1.00 123.8±0.13µs ? ?/sec
smol-songs.csv: typo/stromal 14.63 1859.3±2.40µs ? ?/sec 1.00 127.1±0.29µs ? ?/sec
smol-songs.csv: typo/the white striper 10.83 9.4±0.01ms ? ?/sec 1.00 866.0±0.98µs ? ?/sec
smol-songs.csv: typo/thelonius monk 14.40 3.8±0.00ms ? ?/sec 1.00 261.5±1.30µs ? ?/sec
smol-songs.csv: words/7000 Danses / Le Baiser / je me trompe de mots 5.54 70.8±0.09ms ? ?/sec 1.00 12.8±0.03ms ? ?/sec
smol-songs.csv: words/Bring Your Daughter To The Slaughter but now this is not part of the title 3.48 119.8±0.14ms ? ?/sec 1.00 34.4±0.04ms ? ?/sec
smol-songs.csv: words/The Disneyland Children's Sing-Alone song 8.98 71.9±0.12ms ? ?/sec 1.00 8.0±0.01ms ? ?/sec
smol-songs.csv: words/les liaisons dangeureuses 1793 11.88 37.4±0.07ms ? ?/sec 1.00 3.1±0.01ms ? ?/sec
smol-songs.csv: words/seven nation mummy 22.86 23.4±0.04ms ? ?/sec 1.00 1024.8±1.57µs ? ?/sec
smol-songs.csv: words/the black saint and the sinner lady and the good doggo 2.76 124.4±0.15ms ? ?/sec 1.00 45.1±0.09ms ? ?/sec
smol-songs.csv: words/whathavenotnsuchforth and a good amount of words to pop to match the first one 2.52 107.0±0.23ms ? ?/sec 1.00 42.4±0.66ms ? ?/sec
group main-wiki typo-wiki
----- --------- ---------
smol-wiki-articles.csv: basic placeholder/ 1.02 13.7±0.02µs ? ?/sec 1.00 13.4±0.03µs ? ?/sec
smol-wiki-articles.csv: basic with quote/"film" 1.02 409.8±0.67µs ? ?/sec 1.00 402.6±0.48µs ? ?/sec
smol-wiki-articles.csv: basic with quote/"france" 1.00 325.9±0.91µs ? ?/sec 1.00 326.4±0.49µs ? ?/sec
smol-wiki-articles.csv: basic with quote/"japan" 1.00 218.4±0.26µs ? ?/sec 1.01 220.5±0.20µs ? ?/sec
smol-wiki-articles.csv: basic with quote/"machine" 1.00 143.0±0.12µs ? ?/sec 1.04 148.8±0.21µs ? ?/sec
smol-wiki-articles.csv: basic with quote/"miles" "davis" 1.00 11.7±0.06ms ? ?/sec 1.00 11.8±0.01ms ? ?/sec
smol-wiki-articles.csv: basic with quote/"mingus" 1.00 4.4±0.03ms ? ?/sec 1.00 4.4±0.00ms ? ?/sec
smol-wiki-articles.csv: basic with quote/"rock" "and" "roll" 1.00 43.5±0.08ms ? ?/sec 1.01 43.8±0.06ms ? ?/sec
smol-wiki-articles.csv: basic with quote/"spain" 1.00 137.3±0.35µs ? ?/sec 1.05 144.4±0.23µs ? ?/sec
smol-wiki-articles.csv: basic without quote/film 1.00 125.3±0.30µs ? ?/sec 1.06 133.1±0.37µs ? ?/sec
smol-wiki-articles.csv: basic without quote/france 1.21 1782.6±1.65µs ? ?/sec 1.00 1477.0±1.39µs ? ?/sec
smol-wiki-articles.csv: basic without quote/japan 1.28 1363.9±0.80µs ? ?/sec 1.00 1064.3±1.79µs ? ?/sec
smol-wiki-articles.csv: basic without quote/machine 1.73 760.3±0.81µs ? ?/sec 1.00 439.6±0.75µs ? ?/sec
smol-wiki-articles.csv: basic without quote/miles davis 1.03 17.0±0.03ms ? ?/sec 1.00 16.5±0.02ms ? ?/sec
smol-wiki-articles.csv: basic without quote/mingus 1.07 5.3±0.01ms ? ?/sec 1.00 5.0±0.00ms ? ?/sec
smol-wiki-articles.csv: basic without quote/rock and roll 1.01 63.9±0.18ms ? ?/sec 1.00 63.0±0.07ms ? ?/sec
smol-wiki-articles.csv: basic without quote/spain 2.07 667.4±0.93µs ? ?/sec 1.00 322.8±0.29µs ? ?/sec
smol-wiki-articles.csv: prefix search/c 1.00 343.1±0.47µs ? ?/sec 1.00 344.0±0.34µs ? ?/sec
smol-wiki-articles.csv: prefix search/g 1.00 374.4±3.42µs ? ?/sec 1.00 374.1±0.44µs ? ?/sec
smol-wiki-articles.csv: prefix search/j 1.00 359.9±0.31µs ? ?/sec 1.00 361.2±0.79µs ? ?/sec
smol-wiki-articles.csv: prefix search/q 1.01 102.0±0.12µs ? ?/sec 1.00 101.4±0.32µs ? ?/sec
smol-wiki-articles.csv: prefix search/t 1.00 536.7±1.39µs ? ?/sec 1.00 534.3±0.84µs ? ?/sec
smol-wiki-articles.csv: prefix search/x 1.00 400.9±1.00µs ? ?/sec 1.00 399.5±0.45µs ? ?/sec
smol-wiki-articles.csv: proximity/april paris 3.86 14.4±0.01ms ? ?/sec 1.00 3.7±0.01ms ? ?/sec
smol-wiki-articles.csv: proximity/diesel engine 12.98 10.4±0.01ms ? ?/sec 1.00 803.5±1.13µs ? ?/sec
smol-wiki-articles.csv: proximity/herald sings 1.00 12.7±0.06ms ? ?/sec 5.29 67.1±0.09ms ? ?/sec
smol-wiki-articles.csv: proximity/tea two 6.48 1452.1±2.78µs ? ?/sec 1.00 224.1±0.38µs ? ?/sec
smol-wiki-articles.csv: typo/Disnaylande 3.89 8.5±0.01ms ? ?/sec 1.00 2.2±0.01ms ? ?/sec
smol-wiki-articles.csv: typo/aritmetric 3.78 10.3±0.01ms ? ?/sec 1.00 2.7±0.00ms ? ?/sec
smol-wiki-articles.csv: typo/linax 8.91 1426.7±0.97µs ? ?/sec 1.00 160.1±0.18µs ? ?/sec
smol-wiki-articles.csv: typo/migrosoft 7.48 1417.3±5.84µs ? ?/sec 1.00 189.5±0.88µs ? ?/sec
smol-wiki-articles.csv: typo/nympalidea 3.96 7.2±0.01ms ? ?/sec 1.00 1810.1±2.03µs ? ?/sec
smol-wiki-articles.csv: typo/phytogropher 3.71 7.2±0.01ms ? ?/sec 1.00 1934.3±6.51µs ? ?/sec
smol-wiki-articles.csv: typo/sisan 6.44 1497.2±1.38µs ? ?/sec 1.00 232.7±0.94µs ? ?/sec
smol-wiki-articles.csv: typo/the fronce 6.92 2.9±0.00ms ? ?/sec 1.00 418.0±1.76µs ? ?/sec
smol-wiki-articles.csv: words/Abraham machin 16.63 10.8±0.01ms ? ?/sec 1.00 649.7±1.08µs ? ?/sec
smol-wiki-articles.csv: words/Idaho Bellevue pizza 27.15 25.6±0.03ms ? ?/sec 1.00 944.2±5.07µs ? ?/sec
smol-wiki-articles.csv: words/Kameya Tokujirō mingus monk 26.87 40.7±0.05ms ? ?/sec 1.00 1515.3±2.73µs ? ?/sec
smol-wiki-articles.csv: words/Ulrich Hensel meilisearch milli 11.99 48.8±0.10ms ? ?/sec 1.00 4.1±0.02ms ? ?/sec
smol-wiki-articles.csv: words/the black saint and the sinner lady and the good doggo 4.90 110.0±0.15ms ? ?/sec 1.00 22.4±0.03ms ? ?/sec
```
Co-authored-by: mpostma <postma.marin@protonmail.com>
Co-authored-by: ad hoc <postma.marin@protonmail.com>
2022-03-15 16:43:36 +00:00
ad hoc
3f24555c3d
custom fst automatons
2022-03-15 17:38:35 +01:00
ad hoc
628c835a22
fix tests
2022-03-15 17:38:34 +01:00
bors[bot]
8efac33b53
Merge #467
...
467: optimize prefix database r=Kerollmops a=MarinPostma
This pr introduces two optimizations that greatly improve the speed of computing prefix databases.
- The time that it takes to create the prefix FST has been divided by 5 by inverting the way we iterated over the words FST.
- We unconditionally and needlessly checked for documents to remove in `word_prefix_pair`, which caused an iteration over the whole database.
Co-authored-by: ad hoc <postma.marin@protonmail.com>
2022-03-15 16:14:35 +00:00
ad hoc
d127c57f2d
review edits
2022-03-15 17:12:48 +01:00
ad hoc
d633ac5b9d
optimize word prefix pair
2022-03-15 16:37:22 +01:00
ad hoc
d68fe2b3c7
optimize word prefix fst
2022-03-15 16:36:48 +01:00
Clément Renault
0c5f4ed7de
Apply suggestions
...
Co-authored-by: Many <many@meilisearch.com>
2022-03-15 14:18:29 +01:00
Kerollmops
21ec334dcc
Fix the compilation error of the dependency versions
2022-03-15 11:17:45 +01:00
Kerollmops
63682c2c9a
Upgrade the dependencies
2022-03-15 11:17:44 +01:00
Kerollmops
288a879411
Remove three useless dependencies
2022-03-15 11:17:44 +01:00
psvnl sai kumar
5e08fac729
fixes for rustfmt pass
2022-03-14 19:22:41 +05:30
psvnl sai kumar
92e2e09434
exporting heed to avoid having different versions of Heed in Meilisearch
2022-03-14 01:01:58 +05:30
Kerollmops
1ae13c1374
Avoid iterating on big databases when useless
2022-03-09 15:43:54 +01:00
Bruno Casali
66c6d5e1ef
Add a new error message when the valid_fields is empty
...
> "Attribute `{}` is not sortable. This index doesn't have configured sortable attributes."
> "Attribute `{}` is not sortable. Available sortable attributes are: `{}`."
coexist in the error handling
2022-03-05 10:38:18 -03:00
Clémentine Urquizar
d9ed9de2b0
Update heed link in cargo toml
2022-03-01 19:45:29 +01:00
Kerollmops
d5b8b5a2f8
Replace the ugly unwraps by clean if let Somes
2022-02-28 16:31:33 +01:00
Kerollmops
8d26f3040c
Remove a useless grenad file merging
2022-02-28 16:31:33 +01:00
Clément Renault
04b1bbf932
Reintroduce appending sorted entries when possible
2022-02-24 14:50:45 +01:00
bors[bot]
25123af3b8
Merge #436
...
436: Speed up the word prefix databases computation time r=Kerollmops a=Kerollmops
This PR depends on the fixes done in #431 and must be merged after it.
In this PR we will bring the `WordPrefixPairProximityDocids`, `WordPrefixDocids` and, `WordPrefixPositionDocids` update structures to a new era, a better era, where computing the word prefix pair proximities costs much fewer CPU cycles, an era where this update structure can use the, previously computed, set of new word docids from the newly indexed batch of documents.
---
The `WordPrefixPairProximityDocids` is an update structure, which means that it is an object that we feed with some parameters and which modifies the LMDB database of an index when asked for. This structure specifically computes the list of word prefix pair proximities, which correspond to a list of pairs of words associated with a proximity (the distance between both words) where the second word is not a word but a prefix e.g. `s`, `se`, `a`. This word prefix pair proximity is associated with the list of documents ids which contains the pair of words and prefix at the given proximity.
The origin of the performances issue that this struct brings is related to the fact that it starts its job from the beginning, it clears the LMDB database before rewriting everything from scratch, using the other LMDB databases to achieve that. I hope you understand that this is absolutely not an optimized way of doing things.
Co-authored-by: Clément Renault <clement@meilisearch.com>
Co-authored-by: Kerollmops <clement@meilisearch.com>
2022-02-16 15:41:14 +00:00
Clément Renault
ff8d7a810d
Change the behavior of the as_cloneable_grenad by taking a ref
2022-02-16 15:40:08 +01:00
Clément Renault
f367cc2e75
Finally bump grenad to v0.4.1
2022-02-16 15:28:48 +01:00
Irevoire
0defeb268c
bump milli
2022-02-16 13:27:41 +01:00
Irevoire
48542ac8fd
get rid of chrono in favor of time
2022-02-15 11:41:55 +01:00
Clémentine Urquizar
d03b3ceb58
Update version for the next release (v0.22.1)
2022-02-07 18:39:29 +01:00
bors[bot]
5d58cb7449
Merge #442
...
442: fix phrase search r=curquiza a=MarinPostma
Run the exact match search on 7 words windows instead of only two. This makes false positive very very unlikely, and impossible on phrase query that are less than seven words.
Co-authored-by: ad hoc <postma.marin@protonmail.com>
2022-02-07 16:18:20 +00:00
ad hoc
bd2262ceea
allow null values in csv
2022-02-03 16:03:01 +01:00
ad hoc
13de251047
rewrite word pair distance gathering
2022-02-03 15:57:20 +01:00
Many
d59bcea749
Revert "Revert "Change chunk size to 4MiB to fit more the end user usage""
2022-02-02 17:01:13 +01:00
mpostma
7541ab99cd
review changes
2022-02-02 12:59:01 +01:00
mpostma
d0aabde502
optimize 2 typos case
2022-02-02 12:56:09 +01:00
mpostma
55e6cb9c7b
typos on first letter counts as 2
2022-02-02 12:56:09 +01:00
mpostma
642c01d0dc
set max typos on ngram to 1
2022-02-02 12:56:08 +01:00
ad hoc
d852dc0d2b
fix phrase search
2022-02-01 20:21:33 +01:00
Kerollmops
fb79c32430
Compute the new, common and, deleted prefix words fst once
2022-01-27 11:00:18 +01:00
Clément Renault
51d1e64b23
Remove, now useless, the WriteMethod enum
2022-01-27 10:08:35 +01:00
Clément Renault
e9c02173cf
Rework the WordsPrefixPositionDocids update to compute a subset of the database
2022-01-27 10:08:35 +01:00
Clément Renault
dbba5fd461
Create a function to simplify the word prefix pair proximity docids compute
2022-01-27 10:08:35 +01:00
Clément Renault
e760e02737
Fix the computation of the newly added and common prefix pair proximity words
2022-01-27 10:08:35 +01:00
Clément Renault
d59e559317
Fix the computation of the newly added and common prefix words
2022-01-27 10:08:34 +01:00
Clément Renault
2ec8542105
Rework the WordPrefixDocids update to compute a subset of the database
2022-01-27 10:08:34 +01:00
Clément Renault
28692f65be
Rework the WordPrefixDocids update to compute a subset of the database
2022-01-27 10:08:34 +01:00
Clément Renault
5404bc02dd
Move the fst_stream_into_hashset method in the helper methods
2022-01-27 10:06:00 +01:00
Clément Renault
c90fa95f93
Only compute the word prefix pairs on the created word pair proximities
2022-01-27 10:06:00 +01:00
Clément Renault
822f67e9ad
Bring the newly created word pair proximity docids
2022-01-27 10:06:00 +01:00
Clément Renault
d28f18658e
Retrieve the previous version of the words prefixes FST
2022-01-27 10:05:59 +01:00
bors[bot]
38d23546a5
Merge #431
...
431: Fix and improve word prefix pair proximity r=ManyTheFish a=Kerollmops
This PR first fixes the algorithm we used to select and compute the word prefix pair proximity database. The previous version was skipping nearly all of the prefixes. The issue is that this fix made this method to take more time and we were trying to reduce the time spent in it.
With `@ManyTheFish` we found out that we could skip some of the work we were doing by:
- discarding the prefixes that were shorter than a specific threshold (default: 2).
- discarding the word prefix pairs with proximity bigger than a specific threshold (default: 4).
- remove the unused threshold that was specifying a minimum amount of word docids to merge.
We will take more time to do some more optimization, like stop clearing and recomputing from scratch the database, we will compute the subsets of keys to create, keep and merge. This change is a little bit more complex than what this PR does.
I keep this PR as a draft as I want to further test the real gain if it is enough or not if it is valid or not. I advise reviewers to review commit by commit to see the changes bit by bit, reviewing the whole PR can be hard.
Co-authored-by: Clément Renault <clement@meilisearch.com>
2022-01-27 07:04:56 +00:00
Clément Renault
f9b214f34e
Apply suggestions from code review
...
Co-authored-by: Many <legendre.maxime.isn@gmail.com>
2022-01-26 11:28:11 +01:00
bors[bot]
e1cc025cbd
Merge #440
...
440: fix(fuzzer): fix the fuzzer after #430 r=Kerollmops a=irevoire
Co-authored-by: Tamo <tamo@meilisearch.com>
2022-01-25 16:33:57 +00:00
Clément Renault
f04cd19886
Introduce a max prefix length parameter to the word prefix pair proximity update
2022-01-25 17:04:23 +01:00
