Dependable Data Repairing with Fixing Rules Jiannan Wang Nan Tang - - PowerPoint PPT Presentation

▶
dependable data repairing with fixing rules
SMART_READER_LITE
LIVE PREVIEW

Dependable Data Repairing with Fixing Rules Jiannan Wang Nan Tang - - PowerPoint PPT Presentation

Feb 03, 2024 377 likes •1.1k views

Dependable Data Repairing with Fixing Rules Jiannan Wang Nan Tang 1 Data is Dirty 2 incomplete inconsistent inaccurate Data is Dirty 2 incomplete 25% companies: flawed data inconsistent 3+ trillion $: US economy 20%: labor

slide-1
SLIDE 1

Dependable Data Repairing with Fixing Rules

Jiannan Wang Nan Tang

1

slide-2
SLIDE 2

Data is Dirty

2

slide-3
SLIDE 3

Data is Dirty

incomplete inconsistent inaccurate …

2

slide-4
SLIDE 4

Data is Dirty

incomplete inconsistent inaccurate …

25% companies: flawed data 3+ trillion $: US economy 20%: labor productivity … …

2

slide-5
SLIDE 5

Data is Dirty

incomplete inconsistent inaccurate …

25% companies: flawed data 3+ trillion $: US economy 20%: labor productivity … …

Big (clean) data: new oil

2

slide-6
SLIDE 6

D a t a t r a n s f

  • r

m a t i

  • n

( E T L r u l e s )

  • Entity resolution (deduplication)

S t a t i s t i c a l / M L

  • Truth discovery

T y p

  • s

( s y n t a c t i c e r r

  • r

s )

  • State-of-the-art

3

slide-7
SLIDE 7

D a t a t r a n s f

  • r

m a t i

  • n

( E T L r u l e s )

  • Entity resolution (deduplication)

S t a t i s t i c a l / M L

  • Truth discovery

T y p

  • s

( s y n t a c t i c e r r

  • r

s )

  • Constraint (dependency) based data cleaning

State-of-the-art

3

slide-8
SLIDE 8
  • Data dependencies (a.k.a. integrity constraints)

Dependency Theory

4

slide-9
SLIDE 9
  • Data dependencies (a.k.a. integrity constraints)

Dependency Theory

name country capital city conf r1 George China Beijing Beijing SIGMOD r2 Ian China Shanghai Hongkong ICDE r3 Peter China Tokyo Tokyo ICDE r4 Mike Canada Toronto Toronto VLDB

4

slide-10
SLIDE 10
  • Data dependencies (a.k.a. integrity constraints)

Dependency Theory

name country capital city conf r1 George China Beijing Beijing SIGMOD r2 Ian China Shanghai Hongkong ICDE r3 Peter China Tokyo Tokyo ICDE r4 Mike Canada Toronto Toronto VLDB

FD: [country] -> [capital]

4

slide-11
SLIDE 11
  • Data dependencies (a.k.a. integrity constraints)

Dependency Theory

name country capital city conf r1 George China Beijing Beijing SIGMOD r2 Ian China Shanghai Hongkong ICDE r3 Peter China Tokyo Tokyo ICDE r4 Mike Canada Toronto Toronto VLDB

FD: [country] -> [capital]

4

slide-12
SLIDE 12
  • Data dependencies (a.k.a. integrity constraints)

Dependency Theory

name country capital city conf r1 George China Beijing Beijing SIGMOD r2 Ian China Shanghai Hongkong ICDE r3 Peter China Tokyo Tokyo ICDE r4 Mike Canada Toronto Toronto VLDB

FD: [country] -> [capital]

Data dependencies are not sufficient to guide dependable data repairing

4

slide-13
SLIDE 13

User Guidance

name country capital city conf r1 George China Beijing Beijing SIGMOD r2 Ian China Shanghai Hongkong ICDE r3 Peter China Tokyo Tokyo ICDE r4 Mike Canada Toronto Toronto VLDB country capital s1 China Beijing s2 Canada Ottawa s3 Japan Tokyo

5

slide-14
SLIDE 14

User Guidance

name country capital city conf r1 George China Beijing Beijing SIGMOD r2 Ian China Shanghai Hongkong ICDE r3 Peter China Tokyo Tokyo ICDE r4 Mike Canada Toronto Toronto VLDB country capital s1 China Beijing s2 Canada Ottawa s3 Japan Tokyo

editing rule: ((country, country) -> (capital, capital))

5

slide-15
SLIDE 15

User Guidance

name country capital city conf r1 George China Beijing Beijing SIGMOD r2 Ian China Shanghai Hongkong ICDE r3 Peter China Tokyo Tokyo ICDE r4 Mike Canada Toronto Toronto VLDB country capital s1 China Beijing s2 Canada Ottawa s3 Japan Tokyo

editing rule: ((country, country) -> (capital, capital))

5

slide-16
SLIDE 16

User Guidance

name country capital city conf r1 George China Beijing Beijing SIGMOD r2 Ian China Shanghai Hongkong ICDE r3 Peter China Tokyo Tokyo ICDE r4 Mike Canada Toronto Toronto VLDB country capital s1 China Beijing s2 Canada Ottawa s3 Japan Tokyo

editing rule: ((country, country) -> (capital, capital)) Is r2[country] China? YES.

5

slide-17
SLIDE 17

User Guidance

name country capital city conf r1 George China Beijing Beijing SIGMOD r2 Ian China Shanghai Hongkong ICDE r3 Peter China Tokyo Tokyo ICDE r4 Mike Canada Toronto Toronto VLDB country capital s1 China Beijing s2 Canada Ottawa s3 Japan Tokyo

editing rule: ((country, country) -> (capital, capital)) Is r2[country] China? YES. Beijing

5

slide-18
SLIDE 18

User Guidance

name country capital city conf r1 George China Beijing Beijing SIGMOD r2 Ian China Shanghai Hongkong ICDE r3 Peter China Tokyo Tokyo ICDE r4 Mike Canada Toronto Toronto VLDB country capital s1 China Beijing s2 Canada Ottawa s3 Japan Tokyo

editing rule: ((country, country) -> (capital, capital)) Is r2[country] China? YES. Beijing Is r1[country] China? Is r3[country] China? Is r4[country] Canada? … … … …

5

slide-19
SLIDE 19

User Guidance

name country capital city conf r1 George China Beijing Beijing SIGMOD r2 Ian China Shanghai Hongkong ICDE r3 Peter China Tokyo Tokyo ICDE r4 Mike Canada Toronto Toronto VLDB country capital s1 China Beijing s2 Canada Ottawa s3 Japan Tokyo

editing rule: ((country, country) -> (capital, capital)) Is r2[country] China? YES. Beijing Is r1[country] China? Is r3[country] China? Is r4[country] Canada? … … … …

check each tuple: not cheap !!

5

slide-20
SLIDE 20

Heuristic

(Automated)

Certain

(User guided)

precision: + recall: ++ precision: ++ recall: ++

6

slide-21
SLIDE 21

Heuristic

(Automated)

Certain

(User guided)

precision: + recall: ++ precision: ++ recall: ++

  • (Automated)

precision: ++ recall: +

Fixing Rules

6

slide-22
SLIDE 22

7

slide-23
SLIDE 23

7

slide-24
SLIDE 24

7

slide-25
SLIDE 25

7

negative evidence

slide-26
SLIDE 26

7

negative evidence

slide-27
SLIDE 27

8

country capital

China Shanghai

Data patterns

slide-28
SLIDE 28

8

country capital

China Shanghai

Data patterns

evidence negative

slide-29
SLIDE 29

8

country capital

China Shanghai

Data patterns

China T

  • kyo

evidence negative

slide-30
SLIDE 30

8

country capital

China Shanghai

Data patterns

China T

  • kyo

evidence negative

?

(China, Beijing) (Japan, T

  • kyo)
slide-31
SLIDE 31

8

country capital

China Shanghai

Data patterns

China T

  • kyo

name work mail

Ian ian@gmail.com

evidence negative

?

(China, Beijing) (Japan, T

  • kyo)
slide-32
SLIDE 32

8

country capital

China Shanghai

Data patterns

China T

  • kyo

name work mail

Ian ian@gmail.com

evidence negative evidence negative

?

(China, Beijing) (Japan, T

  • kyo)
slide-33
SLIDE 33

8

country capital

China Shanghai

Data patterns

China T

  • kyo

name work mail

Ian ian@gmail.com

evidence negative evidence negative

?

(China, Beijing) (Japan, T

  • kyo)

city area code

Beijing 110002

slide-34
SLIDE 34

8

country capital

China Shanghai

Data patterns

China T

  • kyo

name work mail

Ian ian@gmail.com

evidence negative evidence negative evidence negative

?

(China, Beijing) (Japan, T

  • kyo)

city area code

Beijing 110002

slide-35
SLIDE 35
  • Syntax

Fixing Rules

fR1: (([country], [China]), (capital, {Shanghai, Hongkong})) -> Beijing

country {capital capital China Shanghai Beijing Hongkong evidence negative fact

9

slide-36
SLIDE 36
  • Syntax

Fixing Rules

fR1: (([country], [China]), (capital, {Shanghai, Hongkong})) -> Beijing

country {capital capital China Shanghai Beijing Hongkong evidence negative fact name country capital city conf r1 George China Beijing Beijing SIGMOD r2 Ian China Shanghai Hongkong ICDE r3 Peter China Tokyo Tokyo ICDE r4 Mike Canada Toronto Toronto VLDB

9

slide-37
SLIDE 37
  • Syntax

Fixing Rules

fR1: (([country], [China]), (capital, {Shanghai, Hongkong})) -> Beijing

country {capital capital China Shanghai Beijing Hongkong evidence negative fact name country capital city conf r1 George China Beijing Beijing SIGMOD r2 Ian China Shanghai Hongkong ICDE r3 Peter China Tokyo Tokyo ICDE r4 Mike Canada Toronto Toronto VLDB

9

slide-38
SLIDE 38
  • Syntax

Fixing Rules

fR1: (([country], [China]), (capital, {Shanghai, Hongkong})) -> Beijing

country {capital capital China Shanghai Beijing Hongkong evidence negative fact name country capital city conf r1 George China Beijing Beijing SIGMOD r2 Ian China Shanghai Hongkong ICDE r3 Peter China Tokyo Tokyo ICDE r4 Mike Canada Toronto Toronto VLDB

Beijing

9

slide-39
SLIDE 39
  • Syntax

Fixing Rules

fR1: (([country], [China]), (capital, {Shanghai, Hongkong})) -> Beijing

country {capital capital China Shanghai Beijing Hongkong evidence negative fact name country capital city conf r1 George China Beijing Beijing SIGMOD r2 Ian China Shanghai Hongkong ICDE r3 Peter China Tokyo Tokyo ICDE r4 Mike Canada Toronto Toronto VLDB

Beijing

9

Deterministic Conservative

slide-40
SLIDE 40

Applying One Fixing Rule

r2 Ian China Shanghai Hongkong ICDE country {capital capital China Shanghai Beijing Hongkong

10

slide-41
SLIDE 41

Applying One Fixing Rule

r2 Ian China Shanghai Hongkong ICDE r2’ Ian China Beijing Hongkong ICDE country {capital capital China Shanghai Beijing Hongkong

10

slide-42
SLIDE 42
  • Fixes

Applying Multiple Fixing Rules

capital city conf {country country Tokyo Tokyo ICDE China Japan

fR1’ fR3

country {capital capital China Shanghai Beijing Hongkong Tokyo

11

slide-43
SLIDE 43
  • Fixes

Applying Multiple Fixing Rules

capital city conf {country country Tokyo Tokyo ICDE China Japan

fR1’ fR3

r2 Ian China Shanghai Hongkong ICDE r2’ Ian China Beijing Hongkong ICDE

fR1’

country {capital capital China Shanghai Beijing Hongkong Tokyo

11

slide-44
SLIDE 44
  • Fixes

Applying Multiple Fixing Rules

capital city conf {country country Tokyo Tokyo ICDE China Japan

fR1’ fR3

  • Unique fixes

r2 Ian China Shanghai Hongkong ICDE r2’ Ian China Beijing Hongkong ICDE

fR1’

country {capital capital China Shanghai Beijing Hongkong Tokyo

r3 Peter China Tokyo Tokyo ICDE r3’ Peter China Beijing Tokyo ICDE

fR1’

11

slide-45
SLIDE 45
  • Fixes

Applying Multiple Fixing Rules

capital city conf {country country Tokyo Tokyo ICDE China Japan

fR1’ fR3

  • Unique fixes

r2 Ian China Shanghai Hongkong ICDE r2’ Ian China Beijing Hongkong ICDE

fR1’

country {capital capital China Shanghai Beijing Hongkong Tokyo

r3 Peter China Tokyo Tokyo ICDE r3’ Peter China Beijing Tokyo ICDE

fR1’

r3’’ Peter Japan Tokyo Tokyo ICDE

fR3

11

slide-46
SLIDE 46

Fundamentals

12

slide-47
SLIDE 47

Fundamental Problems

13

T ermination

Y es

Consistency

PTIME

Implication

coNP-complete

Determinism

Y es

slide-48
SLIDE 48
  • Tuple enumeration

Ensuring Consistency

country {capital capital China Shanghai Beijing Hongkong country {capital capital Canada Toronto Ottawa

fR1 fR2

14

slide-49
SLIDE 49
  • Tuple enumeration

Ensuring Consistency

country {capital capital China Shanghai Beijing Hongkong country {capital capital Canada Toronto Ottawa

fR1 fR2 (◦, China, Shanghai, ◦, ◦) (◦, China, Hongkong, ◦, ◦) (◦, China, Toronto, ◦, ◦) (◦, Canada, Shanghai, ◦, ◦) (◦, Canada, Hongkong, ◦, ◦) (◦, Canada, Toronto, ◦, ◦)

14

slide-50
SLIDE 50
  • Tuple enumeration

Ensuring Consistency

country {capital capital China Shanghai Beijing Hongkong country {capital capital Canada Toronto Ottawa

fR1 fR2 (◦, China, Shanghai, ◦, ◦) (◦, China, Hongkong, ◦, ◦) (◦, China, Toronto, ◦, ◦) (◦, Canada, Shanghai, ◦, ◦) (◦, Canada, Hongkong, ◦, ◦) (◦, Canada, Toronto, ◦, ◦)

14

  • Rule characterization

Xi, Xj Bi, Bj tpi, tpj two rules

slide-51
SLIDE 51

Repair

15

slide-52
SLIDE 52

Repairing with Fixing Rules

name country capital city conf r1 George China Beijing Beijing SIGMOD r2 Ian China Shanghai Hongkong ICDE r3 Peter China Tokyo Tokyo ICDE r4 Mike Canada Toronto Toronto VLDB

fR1 fR3

country {capital capital China Shanghai Beijing Hongkong country {capital capital Canada Toronto Ottawa capital city conf {country country Tokyo Tokyo ICDE China Japan capital conf {city city Beijing ICDE Hongkong Shanghai

fR2 fR4

16

slide-53
SLIDE 53

Repairing with Fixing Rules

name country capital city conf r1 George China Beijing Beijing SIGMOD r2 Ian China Shanghai Hongkong ICDE r3 Peter China Tokyo Tokyo ICDE r4 Mike Canada Toronto Toronto VLDB

fR1 fR3

country {capital capital China Shanghai Beijing Hongkong country {capital capital Canada Toronto Ottawa capital city conf {country country Tokyo Tokyo ICDE China Japan capital conf {city city Beijing ICDE Hongkong Shanghai

fR2 fR4 country, China country, Canada conf, ICDE capital, Tokyo city, Tokyo capital, Beijing fR1 fR2 fR3, fR4 fR3 fR3 fR4 Key List

16

slide-54
SLIDE 54

Repairing with Fixing Rules

name country capital city conf r1 George China Beijing Beijing SIGMOD r2 Ian China Shanghai Hongkong ICDE r3 Peter China Tokyo Tokyo ICDE r4 Mike Canada Toronto Toronto VLDB

fR1 fR3

country {capital capital China Shanghai Beijing Hongkong country {capital capital Canada Toronto Ottawa capital city conf {country country Tokyo Tokyo ICDE China Japan capital conf {city city Beijing ICDE Hongkong Shanghai

fR2 fR4 country, China country, Canada conf, ICDE capital, Tokyo city, Tokyo capital, Beijing fR1 fR2 fR3, fR4 fR3 fR3 fR4 Key List cnt(fR1) = 1, cnt(fR4) = 1, rules = {fR1} r1’ = r1, rules = {}

r1: itr1:

16

slide-55
SLIDE 55

Repairing with Fixing Rules

name country capital city conf r1 George China Beijing Beijing SIGMOD r2 Ian China Shanghai Hongkong ICDE r3 Peter China Tokyo Tokyo ICDE r4 Mike Canada Toronto Toronto VLDB

fR1 fR3

country {capital capital China Shanghai Beijing Hongkong country {capital capital Canada Toronto Ottawa capital city conf {country country Tokyo Tokyo ICDE China Japan capital conf {city city Beijing ICDE Hongkong Shanghai

fR2 fR4 country, China country, Canada conf, ICDE capital, Tokyo city, Tokyo capital, Beijing fR1 fR2 fR3, fR4 fR3 fR3 fR4 Key List cnt(fR1) = 1, cnt(fR4) = 1, rules = {fR1} r1’ = r1, rules = {}

r1: itr1:

cnt(fR1, fR3, fR4) = 1, rules = {fR1} r2’[capital] = Beijing, cnt(fR3) = 1, cnt(fR4) = 2, rules = {fR4} r2’[city] = Shanghai, rules = {}

r2: itr1: itr2:

16

slide-56
SLIDE 56

Repairing with Fixing Rules

name country capital city conf r1 George China Beijing Beijing SIGMOD r2 Ian China Shanghai Hongkong ICDE r3 Peter China Tokyo Tokyo ICDE r4 Mike Canada Toronto Toronto VLDB

fR1 fR3

country {capital capital China Shanghai Beijing Hongkong country {capital capital Canada Toronto Ottawa capital city conf {country country Tokyo Tokyo ICDE China Japan capital conf {city city Beijing ICDE Hongkong Shanghai

fR2 fR4 country, China country, Canada conf, ICDE capital, Tokyo city, Tokyo capital, Beijing fR1 fR2 fR3, fR4 fR3 fR3 fR4 Key List cnt(fR1) = 1, cnt(fR4) = 1, rules = {fR1} r1’ = r1, rules = {}

r1: itr1:

cnt(fR1, fR3, fR4) = 1, rules = {fR1} r2’[capital] = Beijing, cnt(fR3) = 1, cnt(fR4) = 2, rules = {fR4} r2’[city] = Shanghai, rules = {}

r2: itr1: itr2:

16

slide-57
SLIDE 57

Repairing with Fixing Rules

name country capital city conf r1 George China Beijing Beijing SIGMOD r2 Ian China Shanghai Hongkong ICDE r3 Peter China Tokyo Tokyo ICDE r4 Mike Canada Toronto Toronto VLDB

fR1 fR3

country {capital capital China Shanghai Beijing Hongkong country {capital capital Canada Toronto Ottawa capital city conf {country country Tokyo Tokyo ICDE China Japan capital conf {city city Beijing ICDE Hongkong Shanghai

fR2 fR4 country, China country, Canada conf, ICDE capital, Tokyo city, Tokyo capital, Beijing fR1 fR2 fR3, fR4 fR3 fR3 fR4 Key List cnt(fR1) = 1, cnt(fR4) = 1, rules = {fR1} r1’ = r1, rules = {}

r1: itr1:

cnt(fR1, fR3, fR4) = 1, rules = {fR1} r2’[capital] = Beijing, cnt(fR3) = 1, cnt(fR4) = 2, rules = {fR4} r2’[city] = Shanghai, rules = {}

r2: itr1: itr2:

Beijing Shanghai 16

slide-58
SLIDE 58

Repairing with Fixing Rules

name country capital city conf r1 George China Beijing Beijing SIGMOD r2 Ian China Shanghai Hongkong ICDE r3 Peter China Tokyo Tokyo ICDE r4 Mike Canada Toronto Toronto VLDB

fR1 fR3

country {capital capital China Shanghai Beijing Hongkong country {capital capital Canada Toronto Ottawa capital city conf {country country Tokyo Tokyo ICDE China Japan capital conf {city city Beijing ICDE Hongkong Shanghai

fR2 fR4 country, China country, Canada conf, ICDE capital, Tokyo city, Tokyo capital, Beijing fR1 fR2 fR3, fR4 fR3 fR3 fR4 Key List cnt(fR1) = 1, cnt(fR4) = 1, rules = {fR1} r1’ = r1, rules = {}

r1: itr1:

cnt(fR1, fR3, fR4) = 1, rules = {fR1} r2’[capital] = Beijing, cnt(fR3) = 1, cnt(fR4) = 2, rules = {fR4} r2’[city] = Shanghai, rules = {}

r2: itr1: itr2:

cnt(fR3) = 3, cnt(fR4) = 1, rules = {fR3} r3’[country] = Japan, rules = {}

r3: itr1:

Beijing Shanghai Japan 16

slide-59
SLIDE 59

Repairing with Fixing Rules

name country capital city conf r1 George China Beijing Beijing SIGMOD r2 Ian China Shanghai Hongkong ICDE r3 Peter China Tokyo Tokyo ICDE r4 Mike Canada Toronto Toronto VLDB

fR1 fR3

country {capital capital China Shanghai Beijing Hongkong country {capital capital Canada Toronto Ottawa capital city conf {country country Tokyo Tokyo ICDE China Japan capital conf {city city Beijing ICDE Hongkong Shanghai

fR2 fR4 country, China country, Canada conf, ICDE capital, Tokyo city, Tokyo capital, Beijing fR1 fR2 fR3, fR4 fR3 fR3 fR4 Key List cnt(fR1) = 1, cnt(fR4) = 1, rules = {fR1} r1’ = r1, rules = {}

r1: itr1:

cnt(fR1, fR3, fR4) = 1, rules = {fR1} r2’[capital] = Beijing, cnt(fR3) = 1, cnt(fR4) = 2, rules = {fR4} r2’[city] = Shanghai, rules = {}

r2: itr1: itr2:

cnt(fR3) = 3, cnt(fR4) = 1, rules = {fR3} r3’[country] = Japan, rules = {}

r3: itr1:

cnt(fR3) = 1, rules = {fR2} r4’[capital] = Ottawa, rules = {}

r4: itr1:

Beijing Shanghai Japan Ottawa 16

slide-60
SLIDE 60

Experiment

17

slide-61
SLIDE 61
  • Efficiency of checking consistency
  • Accuracy
  • Efficiency of repairing algorithms

Experimental Study

18

slide-62
SLIDE 62

Efficiency of Checking Consistency

100 101 102 103 104 105 106 1 2 3 4 5 6 7 8 9 10 Time (msec) # of rules (* 100)

isConsistt (worst case) isConsistr (worst case)

10-3 10-2 10-1 100 101 102 103 1 2 3 4 5 6 7 8 9 10 Time (msec) # of rules (* 10)

isConsistt (worst case) isConsistr (worst case)

Hospital data UIS

19

slide-63
SLIDE 63

Accuracy

10 20 30 40 50 60 20 40 60 80 100 # of errors corrected Top 100 rules

0.2 0.4 0.6 0.8 1 Precision Recall

Fix Edit

Hospital data

20

slide-64
SLIDE 64

Efficiency of Repairing Algorithms

2 4 6 8 10 12 1 2 3 4 5 6 7 8 9 10 Time (sec) # of rules (* 100) cRepair lRepair 0.05 0.1 0.15 0.2 1 2 3 4 5 6 7 8 9 10 Time (sec) # of rules (* 10) cRepair lRepair

Hospital data UIS

21

slide-65
SLIDE 65
  • Heuristic

(Automated)

  • Certain

(User guided)

  • Fixing Rules

precision: + recall: ++ precision: ++ recall: ++ precision: ++ recall: +

22

slide-66
SLIDE 66
  • Heuristic

(Automated)

  • Certain

(User guided)

  • Fixing Rules

precision: + recall: ++ precision: ++ recall: ++ precision: ++ recall: +

Conclusion:

Automated Dependable Fundamentals Repair

22

slide-67
SLIDE 67
  • Heuristic

(Automated)

  • Certain

(User guided)

  • Fixing Rules

precision: + recall: ++ precision: ++ recall: ++ precision: ++ recall: +

Conclusion:

Automated Dependable Fundamentals Repair

Future work:

Discovery Generalized fixing rules

22

slide-68
SLIDE 68

How to Get My Rules?

23

slide-69
SLIDE 69

Generating Fixing Rules

country {capital capital China Shanghai Beijing Hongkong

24

slide-70
SLIDE 70

Generating Fixing Rules

country {capital capital China Shanghai Beijing Hongkong

[{ "type": "/location/country", "name": null, "/location/country/capital": [] }]

MQL1

24

slide-71
SLIDE 71

Generating Fixing Rules

country {capital capital China Shanghai Beijing Hongkong

[{ "type": "/location/country", "name": null, "/location/country/capital": [] }]

MQL1

[{ "/location/country/iso3166 1 shortname": "CHINA", "/location/location/contains": [{
 "name": null,
 "type": "/location/citytown"

}]

}]

MQL2

24