[RakutenTechConf2013] [B-3_3] Rakuten Category

Rakuten CategoryVol.01 　 Oct/26/2013Yuhei Nishioka / Suguru SuzukiRakuten Inc.http://www.rakuten.co.jp/

2

Agenda

1.Rakuten Category- Introduction -

2.Measurement/Modification- Approach for Category design -

3.Release- Standardization -

3

Self-Introduction

Suguru SuzukiJapan Ichiba SectionJapan Mall GroupRakuten Ichiba Development Department

• Application Engineer• Joined Rakuten in 2007• Ichiba TOP/ Rakuten

Search(All devices)

Yuhei NishiokaRakuten Institute of Technology

• Chief Technologist• Joined Rakuten in 2008• Semantic Web,

Recommender System

Rakuten Category- Introduction -

Rakuten Category

5

Rakuten Category

What’s Category??

Category??

6

カテゴリーは、事柄の性質を区分する上でのもっとも基本的な分類のことである。

In metaphysics (in particular, ontology), the different kinds or ways of being are called categories of being or simply categories.

Rakuten Category

Source of Quote : wikipedia 　 http://ja.wikipedia.org/wiki/%E3%82%AB%E3%83%86%E3%82%B4%E3%83%AA

Rakuten’s Category is…

Sales area =“ 売り場”

7

Rakuten Category

Rakuten Search

CategoryRanking

Category

ReviewCategory

BooksCategory

CategoryAuction

Category

TOP/Genre TOP

Category

Racoupon/coupon search

Category

And more and more….

8

Rakuten Category

Data Number

Category in Rakuten Ichiba 50,896 genres

Using Category Service 50 service

Using Category Application 100 application

Effective Service of using Category(Genre/Tag)

Auction

RMS

SearchEngine

kobo

Basket

Review

Rakuten Search

Books

Ranking

Advertisement

TOP page

Item Page

Affiliate

Mail

BrowsingHistory

Web Service

Super DB

GMSReport

Auto

Racoupon

A lot of service useCategory data!

9

Rakuten Category

Good Categorize

Catch up the trend

Easy to navigate User

Big factor to increase sales in each items.

10

Rakuten Category

User Come across items

Shop Sell itemsRakuten Sell items

Data analysis

Benefit!!

11

Rakuten Category

Cycle of Category Strategy

Measurements

ReleaseModification

Need toHigh Speed!!

Measurement/Modification

Measurement/Modification- Approach for Category design -

13

Measurement/Modification

Measurements

ReleaseModification

POINT

POINT

14

Measurement/Modification

Measurement on WEB-toolTree view Item countSales volumeRanking data

Show more detail!!

15

Data-Driven Optimization

Modify Category by Analyzing User’s Queries

Past Example of data-driven optimization

….ホットプレート(Hot Plate)…

タジン鍋(Tagines)

No responding genre

You can find “ タジン鍋”without using search

List of high frequency queries

Create new category(a couple of years ago)

Already existing in Rakuten Category Tree

16

Types of queries

Needs browsing function for not only category tree but also other attributes

Ratio of Query Types

Source: User Queries tat Rakuten Ichiba in 2013

Podcut Category

Brand

Merchant

Spec

Character

Others

17

Master Database

Create new master database for brand, color and so on

Data Structure behind Navigation

BrandMaster. a

Unified Brand Master

IntegrationBrand

Master. b

BrandMaster. c

Category…..…..…..

Brand…..…..…..

Color…..…..…..

Color Master

Category Tree

Already Exist

…

NavigationMaster DataData Source

New

New

18

The difficulty identifying brand

Brand name matching is very effective. But must solve 2 major problems

2 major technical problems in brand name matching

• Different Things with the Same Name• カリタ

• The Same Thing with Different Names• Samsung• サムスン

http://www.kalita.co.jp/

http://www.carita.jp/

19

Check by hand

Brand name matching is very effective. But must solve 2 major problems

Data Process

Original Matching Algorithm- Title match- Synonym check- Ambiguous word check- Use other attribute- … chec

k

Result

20

Check by hand with few costs

OpenRefine is very helpful

ID Name

xxx SONY

yyy カリタ

…. ….

Original Matching Algorithm

Other Master Data

SONY [ Matched ]

Karita [ Candidate1 ] CARITA [ Candidate 2]

API for Open Refine

Web Interface

Server side

http://openrefine.org/

Useful Open Source Tool

http://www.carita.jp/source

21

Color Master

Building color dictionary automatically as much as possible

Color Dictionary

黒

black.

1,871 color names

黒色• Image Processing

• Natural Language Processing

Black

Blue

blue

navy

.

16 color groups

..

.

22

Tagging Data for each item

Structured Data

Category Brand Color

Item ID Category Brand Color …

xxxx

….

Merchant Input

Extract AutomaticallyFrom item description

(in research)

23

Attribute value extraction

• Generate extraction rules using attribute value database constructed from table data

Item page includinga dictionary entry

Rulewine from x => x is a Region

Chateau d’Issan 1994This is a wine from Margaux....

:<Region, Margaux><Color, White> :

DatabaseAnnotation

Values not included in the database can be captured.

Table data

24

Measurement/Modification

Modification on WEB-tool

Drag and DropEasy to modify!!

25

Measurement/Modification

Old modification styleExtra

Hand-made…!

26

Measurement/ModificationExtra

Problem

Achieved limit counts by excel

…orz

Old modification style

27

Measurement/Modification

Good Categorize =A huge benefit Very Important phase Need to survey trend and data

optimization

Release

Release- Standardization -

29

Release

Measurements

ReleaseModification

POINT

30

Release

Measurements

ReleaseModification

Need it more rapidly!!

31

Release

Hard to release Category dataCategory data has over 15 DB…Deliver its data to all 50 service.

Auction

RMS

SearchEngine

kobo

Basket

Review

Rakuten Search

Books

Ranking

Advertisement

TOP page

Item Page

Affiliate

Mail

BrowsingHistory

Web Service

Super DB

GMSReport

Auto

Racoupon

Have over 15 DB....

Deliver data to all service

Add new servicesometime

Extra - Before

32

Extra - Before Release

Show the Maintenance time tableWhen Category Restructuring maintenance.Complicated!!!Related Category Restructuring task

is almost 300!!

33

Release

Easy to release by all servicemore speedy

ServiceA

・・・・

ServiceB

ServiceC

ServiceD

ServiceE

Already Automation In Progress for Automation

Now improving!

API

CategoryData

34

Release

■System Reconstruction used by APIBefore In

Progress

Test and operate by each service

ServiceA serviceB

serviceD serviceE

serviceC

・・・・

Release in Regular Maintenance

6monthMaking data by management tool

Release in week

ServiceA serviceB serviceC

・・・・

APIReflect new Data used by API

Making data by handmade

Share data by dump or excel

Every week

serviceD serviceE

CategoryData

35

Release

More easily more Speedy!!

Auction

RMS

SearchEngine

kobo

Basket

Review

Rakuten Search

Books

Ranking

Advertisement

TOP page

Item Page

Affiliate

Mail

BrowsingHistory

Web Service

Super DB

GMSReport

Auto

Racoupon

CategoryAPI

For operation freeGet rid of dependency in each service

CategoryData

36

■Real Time reflection

iPhone5s

Register

Real Time releasedwhen needed.

Real Time reference

Can be released Category Data andsearch it by “Real Time” on Rakuten Search.

Release

37

■Real Time reflection

iPhone5s

Register

Can be released Category Data andsummarize it on Ranking.

Releasedas a daily/weekly

Ranking data.

Release

38

■Real Time reflection

iPhone5s

Register

Can be createdLanding-page

used bynew Categorydata

Can be released Category Data andCreate Landing-page.

Release

39

Finally

Measurements

ReleaseModification

Standardization forcycle of Improvement

40

Finally

User Come across items

Shop Sell itemsRakuten Sell items

Data analysis

Benefit!!Category optimization ismade everyone happy!!

41

Thank you for your Listening!!Finally

If you have any idea or question, Please contact us.Let’s talk about Category with us!!

Yuhei Nishioka

@nishiokamegane

yuhei.nishioka@mail.rakuten.com

Suguru Suzuki

@sugsuzuki

sugru.suzuki@mail.rakuten.com