Upgrade to Pro
— share decks privately, control downloads, hide ads and more …
Speaker Deck
Features
Speaker Deck
PRO
Sign in
Sign up for free
Search
Search
how to improve search
Search
genta kaneyama
October 17, 2014
Technology
8
2k
how to improve search
at AWS cloudsearch seminor in japan (
http://kokucheese.com/event/index/220172
)
genta kaneyama
October 17, 2014
Tweet
Share
More Decks by genta kaneyama
See All by genta kaneyama
MOSHでの生成AI活用の取り組み
penguinco
0
120
search and community in cookpad 2019
penguinco
2
1.8k
行動ログでプロダクトを改善するには/exploit user behavior for product
penguinco
4
9.1k
Solr @ CROSS2015 C4
penguinco
1
1.3k
solrとelasticsearchの比較
penguinco
15
5.4k
Other Decks in Technology
See All in Technology
Adopting Jetpack Compose in Your Existing Project - GDG DevFest Bangkok 2024
akexorcist
0
110
【Startup CTO of the Year 2024 / Audience Award】アセンド取締役CTO 丹羽健
niwatakeru
0
990
ハイパーパラメータチューニングって何をしているの
toridori_dev
0
140
【若手エンジニア応援LT会】ソフトウェアを学んできた私がインフラエンジニアを目指した理由
kazushi_ohata
0
150
Why does continuous profiling matter to developers? #appdevelopercon
salaboy
0
190
rootlessコンテナのすゝめ - 研究室サーバーでもできる安全なコンテナ管理
kitsuya0828
3
380
第1回 国土交通省 データコンペ参加者向け勉強会③- Snowflake x estie編 -
estie
0
130
The Role of Developer Relations in AI Product Success.
giftojabu1
0
120
データプロダクトの定義からはじめる、データコントラクト駆動なデータ基盤
chanyou0311
2
300
The Rise of LLMOps
asei
7
1.4k
複雑なState管理からの脱却
sansantech
PRO
1
140
なぜ今 AI Agent なのか _近藤憲児
kenjikondobai
4
1.4k
Featured
See All Featured
The MySQL Ecosystem @ GitHub 2015
samlambert
250
12k
What's new in Ruby 2.0
geeforr
343
31k
Why You Should Never Use an ORM
jnunemaker
PRO
54
9.1k
GraphQLの誤解/rethinking-graphql
sonatard
67
10k
Fantastic passwords and where to find them - at NoRuKo
philnash
50
2.9k
Writing Fast Ruby
sferik
627
61k
Art, The Web, and Tiny UX
lynnandtonic
297
20k
Building an army of robots
kneath
302
43k
5 minutes of I Can Smell Your CMS
philhawksworth
202
19k
Building Applications with DynamoDB
mza
90
6.1k
Rails Girls Zürich Keynote
gr2m
94
13k
The Success of Rails: Ensuring Growth for the Next 100 Years
eileencodes
44
6.8k
Transcript
)PXUPJNQSPWFTFBSDI ! !"%4+ ! $PPLQBETFBSDIHSPVQ (FOUB,BOFZBNB
)PXUPJNQSPWFTFBSDI ! !"%4+ ! $PPLQBETFBSDIHSPVQ (FOUB,BOFZBNB
Cookpad SFDJQFTIBSJOHQMBUGPSN ɾ+BQBO4QBJO6OJUFETUBUFT*OEPOFTJB
Cookpad SFDJQFTIBSJOHQMBUGPSN ɾ+BQBO4QBJO6OJUFETUBUFT*OEPOFTJB
None
Monthly Unique Browser 0ສ 1,000ສ 2,000ສ 3,000ສ 4,000ສ 5,000ສ 4݄
5݄ 6݄ 7݄ 8݄ 9݄ 10݄ 11݄ 12݄ 1݄ 4݄ PC Mobile 44M ʢ20144݄ظ ୈ3࢛ظܾࢉิઆ໌ࢿྉʣ
Recipes 0ສ 30ສ 60ສ 90ສ 120ສ 150ສ 180ສ 7݄ 10݄
1݄ 4݄ 7݄ 10݄ 1݄ 4݄ recipe.count 1.71M ʢ20144݄ظ ୈ3࢛ظܾࢉิઆ໌ࢿྉʣ
None
None
Recipe Search ɾ3FDJQFTUIBUWJFXFEUISPVHITFBSDI ˠPGBMM .POUIMZ ɾ6TFSJOUFOUJTWFSZEJWFSTF
Recipe Search ɾ'JOEPVUUIFCFTUSFDJQFGPSFBDIVTFS ɾPSEFSSFDFODZQPQVMBSJUZ ɾGBDFUDBMPSJFDVJTJOFDPPLJOHUJNFFUD
Recipe Search 2VFSZEJWFSTJUZɿ ɾJOHSFEJFOUɿ NFOVɿ ɾTDFOFɿ FUDɿ
Architecture ɾ4PMS NBTUFSTMBWFSFQMJDBUJPO ɾ3BJMT 4VOTQPU JOIPVTFXSBQQFS ɾNFNDBDIF WBSOJTI ɾ;BCCJY
/BHJPT
Architecture BQQ DBDIF T SFQFBUFST SFDJQFTMBWF T 7FSOJTIDBDIJOHMPBECBMBODJOH 4PMSNBTUFSTMBWF NBTUFS
KVTUNBDIJOF TVOTQPUTMBWF T "VUPDPNQMFUFTMBWF T HSPVQCZ$PSF IUUQMPDBMIPTUTPMSTVOTQPUTFMFDU RIFMMPXUKTPO
Architecture BQQ DBDIF T SFQFBUFST SFDJQFTMBWF T 7FSOJTIDBDIJOHMPBECBMBODJOH 4PMSNBTUFSTMBWF NBTUFS
KVTUNBDIJOF TVOTQPUTMBWF T "VUPDPNQMFUFTMBWF T HSPVQCZ$PSF IUUQMPDBMIPTUTPMSTVOTQPUTFMFDU RIFMMPXUKTPO BWH.CQTNBY.CQT TFBSDIUBSHFUNPEFM SFDJQFTpFMET .EPDT TVOTQPUYpFMET .EPDT
pros/cons 1SPT ɾ&BTZUPTDBMFQSPEVDU FBDICZFBDICBTJT ɾOP410' ɾTJNQMFBQQDPOpH BQQPOMZLOPXT-# !
! $POT ɾ)BSEUPVOEFSTUBOE ɾpHIUBHBJOTU($JTQBJOGVM
How to improve search
1. Monitoring
1. Monitoring *UFOBCMFVTUP ! ɾLOPXUPEBZTVTFSTOFFE ɾLOPX66PGFBDILFZXPSE QSPYZPGJNQPSUBODF ɾRVJDLMZBDDFTTUPBDUVBMTFBSDISFTVMU ɾpOELFZXPSETXIJDISFUVSOFEUIFCMBOL
1. Monitoring 8FTIPVMECFBCMFUPBOTXFSʜ ! ɾ)PXNBOZQFPQMFVTFUIJTLFZXPSE zPSlJOXIBUXBZ
1. graph of response time 8FDIFDLUIJTHSBQIFWFSZXFFL JGUIFSFTQPOTFUJNFHFUTXPSTF JOWFTUJHBUFDIBOHFMPHT
2. Graph of response time 8FDIFDLUIJTHSBQIFWFSZXFFL JGUIFSFTQPOTFUJNFHFUTXPSTF JOWFTUJHBUFDIBOHFMPHT
3. report from users and staff. 8FDIFDLBMMSFQPSUTGSPN VTFSTBOETUB⒎TSFMBUFEUP TFBSDIFWFSZEBZ !
:PVDBOMFBSOGSPNNPSF UIBOSFQPSUTGSPNTUB⒎
4. Log analysis ɾ$MJDLMPHTJTWJUBM ɾ—5SFBTVSF%BUB 3FETIJGU
4. Log analysis
4. Log analysis ɾ3FTVMUFYBNQMF ɾTPSULFZXPSECZ)*5!)*5 ɾPS/%$( 85'!FUDʜ !
*GZPVOPUJDFE ZPVDBOTUBSUUIJOLJOHMJLFʜ z"SFVTFSXBOUTΧϨʔ͏ͲΜ z
4. Log analysis -session- ɾ%FpOFTVDDFTTTFTTJPO ɾ:PVXJMMOPUJDFEJ⒎FSFODFCFUXFFOLFZXPSET ɾBWHUJNF
ɾBWHTVDDFTTSBUF ɾRVFSZQBUUFSOFUDʜ ! #VUʜ/PUJNQMZTPMVUJPO NPTUDBTF
Find out problems .POJUPSJOH (SBQIPGSFTQPOTFUJNF 3FQPSUGSPNVTFSTBOETUB⒎
-PHBOBMZTJT ! ˠTVNNBSJ[FˠQSJPSJUJ[FˠUBLFBDUJPO
Dictionary 8FDBOpYXPSEDIVOLJOHXJUIEJDUJPOBSZ 8FpYFEUIPVTBOETPGLFZXPSEJOUIJTZFBS FHlΘΒͼṷzlΘΒͼz "EElΘΒͼṷzJOUPEJDUJPOBSZ"OEDIFDLTFBSDISFTVMUT
Hierarchical Synonym 8FDBOpYTZOPOZNFBTJMZ 8FpYFEUIPVTBOETPGTZOPOZNTJOUIJTZFBS
Speed up indexing 3FJOEFYXIPMFJOEFYFWFSZEBZ 8FDBOBQQMZDIBOHFTJOEBJMZCBTJT
Query modification .BLFBTQBDFGPSRVFSZQSFQSPDFTT 2VFSZDBOCFEJ⒎FSFOUGPSFBDI IUUQXXXTMJEFTIBSFOFUEUVOLFMBOHTFBSDIRVBMJUZBUMJOLFEJO
(cloudsearch)
Exploit user’s log &YQMPJUVTFS`TTVDDFTTGVMRVFSZDIBJOBTSFMBUFETFBSDIFST
Summary 'JOEPVUQSPCMFNT TVNNBSJ[FQSJPSJUJ[F UBLFBDUJPO ! 2VFTUJPOT