Bot 甘苦談，使用者至上

Bot 甘苦談，使用者至上Maxis Kao @ Yahoo Search

什麼是 Bot

程序員鼓勵師

Nov. 6, 2015

PTT 表特版

Heroku + MongoLab + Slack APINode MongoDB Webhook

Heroku + MongoLab + Slack APINode MongoDB Webhook

Heroku + MongoLab + Slack APINode MongoDB Webhook

> db.beauty.lists.findOne()

{

"_id" : ObjectId("563b17bef26ec7b20aabe962"),

"push" : 26,

"full_title" : "[正妹] 又正又會讀書",

"tag" : "正妹",

"title" : "又正又會讀書",

"url" : "https://www.ptt.cc/bbs/Beauty/M.1446701090.A.658.html",

"post_id" : "M.1446701090.A.658",

"mark" : "",

"fetched" : true,

"raw_meta" : {

"作者" : "pk698326889 (JiMiHua)",

"看板" : "Beauty",

"標題" : "[正妹] 又正又會讀書",

"時間" : "Thu Nov 5 13:24:47 2015"

},

"author" : "pk698326889 (JiMiHua)",

"image_count" : 8,

"datetime" : ISODate("2015-11-05T05:24:47Z"),

"display" : 2

}

> db.beauty.posts.findOne()

{

"_id" : ObjectId("563b5a3af3526d750e61f5ad"),

"img_total" : 35,

"post_id" : "M.1446558454.A.167",

"img_idx" : 0,

"img_url" : "http://i.imgur.com/h4f2E15.jpg",

"display" : 1

}

鼓勵 +長腿 >50

4,735 posts

37,849 images

ButSide effects

鼓勵 + 新加坡 = 悲劇

> db.beauty.lists.findOne()

{

"_id" : ObjectId("563b17bef26ec7b20aabe962"),

"push" : 26,

"full_title" : "[正妹] 又正又會讀書",

"tag" : "正妹",

"title" : "又正又會讀書",

"url" : "https://www.ptt.cc/bbs/Beauty/M.1446701090.A.658.html",

"post_id" : "M.1446701090.A.658",

"mark" : "",

"fetched" : true,

"raw_meta" : {

"作者" : "pk698326889 (JiMiHua)",

"看板" : "Beauty",

"標題" : "[正妹] 又正又會讀書",

"時間" : "Thu Nov 5 13:24:47 2015"

},

"author" : "pk698326889 (JiMiHua)",

"image_count" : 8,

"datetime" : ISODate("2015-11-05T05:24:47Z"),

"display" : 2

}

使用者要什麼，就給他什麼

「X，給我男的是要鼓勵什麼？」

版本控制也救不了你

$ git checkout tags/v1.7

$ npm install slack-monkey@1.7.0

BDDBehavior-driven development

重質不重量

Quality

VolumeRevenue

Quality

VolumeRevenue

Quality

VolumeRevenue

為使用者設想

$ clear

https://www.npmjs.com/package/lorem-ipsum

簡化常見的指令

4,600searches in the query log

> _________

> _________Alan

> _________Conrad

> _________布萊恩

user log 是珍貴的資產

看著儀表板開車

鼓勵師 v2.0

Heroku + MongoLab + Slack API

Heroku + MongoLab + Slack API

3~5 s (including dyno sleeping)

Heroku + MongoLab + Slack API

3~5 s 1 s3~5 s

Heroku + MongoLab + Slack API

1 s3~5 s

NodeMongoDB

PREPROCESSING: 0ms

DB_CONNECT: 8ms

QUERY_LOGGING: 4ms

RECORD_COUNT: 1ms

CONDITION_SEARCH: 57ms

Ping: 8ms Webhook: 30ms

server mongodb slack

30x faster3s → 100ms

鼓勵師 v3.0

Elasticsearch

● Full-text search● Analytics● High Availability

Feedback

● Personalization● User preference

Recommendation

preprocessor

slackbot

middleware