Sewol Ferry Disaster Apr 16, 2014 , 294 out of 476 people died 325 of - - PowerPoint PPT Presentation

sewol ferry disaster apr 16 2014 294 out of 476 people
SMART_READER_LITE
LIVE PREVIEW

Sewol Ferry Disaster Apr 16, 2014 , 294 out of 476 people died 325 of - - PowerPoint PPT Presentation

BudgetMap : Engaging Taxpayers in the Issue-Driven Classification of a Government Budget BudgetMap : Engaging Taxpayers in the Issue-Driven Classification of a Government Budget Nam Wook Kim Harvard Eun-Young Ko Songyi Han Jonghyuk Jung


slide-1
SLIDE 1

BudgetMap:

Engaging Taxpayers in the Issue-Driven Classification of a Government Budget

slide-2
SLIDE 2

Nam Wook Kim Harvard Jonghyuk Jung KAIST Eun-Young Ko KAIST Songyi Han KAIST Changwon Lee KAIST Juho Kim Stanford & KAIST Jihee Kim KAIST

BudgetMap:

Engaging Taxpayers in the Issue-Driven Classification of a Government Budget

slide-3
SLIDE 3

Sewol Ferry Disaster Apr 16, 2014, 294 out of 476 people died

slide-4
SLIDE 4

325 of the passengers were secondary school students on a field trip.

slide-5
SLIDE 5

Citizens became suspicious about where the government spends their taxes.

Public safety

slide-6
SLIDE 6

Difficult to explore government programs for a particular issue

Source: US Office of Management and Budget, 2016

slide-7
SLIDE 7

Source: US Office of Management and Budget, 2016

Programs related to the issue are spread out across various budget categories

slide-8
SLIDE 8

“The traditional budget classification system is mainly designed for administrative efficiency.” — A former congressman

slide-9
SLIDE 9

Simply opening up more budget information to the public may not be sufficient.

Open Government

slide-10
SLIDE 10

Supporting the navigation of a government budget through social issues

slide-11
SLIDE 11

A social problem or conflict raised by a society.

Supporting the navigation of a government budget through social issues

from Wikipedia

slide-12
SLIDE 12

Supporting the navigation of a government budget through social issues

  • Use the language of the public
  • Reflect timely public interests
slide-13
SLIDE 13

Challenge

How can we classify budget

data using social issues?

slide-14
SLIDE 14

Challenge

How can we classify budget data using social Issues?

  • Huge amount of budget data
  • Constantly changing social issues
slide-15
SLIDE 15

Challenge

How can we classify budget data using social Issues?

  • Huge amount of budget data
  • Constantly changing social issues

Unsustainable for a government to handle this problem every time a new issue arises.

slide-16
SLIDE 16

Challenge

How can we classify budget data using social Issues?

  • Huge amount of budget data
  • Constantly changing social issues

Automated solution may not be able to understand nuances in social issues.

slide-17
SLIDE 17

Citizensourcing + Tagging

To turn to taxpayers to classify budget programs with social issues.

slide-18
SLIDE 18

BudgetMap

Solicits taxpayers to identify issue-budget tags and facilitates visual exploration of the tagged budget data

slide-19
SLIDE 19

Budget Data from the City of Seoul

Budget spending and Budget categories for 4013 programs in 2015

slide-20
SLIDE 20

Budget Data Model

Program is

a unit of operating budget for a set of services.

slide-21
SLIDE 21

Social Issue

is a form of tag that may connect to multiple programs.

Budget Data Model

slide-22
SLIDE 22

Social

Issue

Budget

Program

N-to-N relationship

Budget Data Model

slide-23
SLIDE 23

“Related”

Social

Issue

Budget

Program “Unrelated” “Unclear”

N-to-N relationship

Tag Types

Budget Data Model

slide-24
SLIDE 24

BudgetMap Tagging Interface

Implementation

slide-25
SLIDE 25

BudgetMap Tagging Interface

and the list goes on

Explore randomly selected budget programs

Implementation

slide-26
SLIDE 26

Majority Voting

Public Safety Special Rescue Team Operation Issue Program

Implementation

slide-27
SLIDE 27

Majority Voting

Public Safety Special Rescue Team Operation Related if

10 people

‘Related’ Tags

>

‘Unrelated’ Tags

5 people

Threshold (x)

+

Issue Program

Implementation

e.g.,x=3

slide-28
SLIDE 28

BudgetMap Visualization

Automatically Translated Implementation

slide-29
SLIDE 29

BudgetMap Visualization

Social Issues

Implementation

slide-30
SLIDE 30

BudgetMap Visualization

Budget Categories

Implementation

slide-31
SLIDE 31

BudgetMap Visualization

Budget Programs for the current category

Implementation

slide-32
SLIDE 32

BudgetMap Visualization

Selected Issue: Air Pollution

Implementation

slide-33
SLIDE 33

BudgetMap Visualization

Budget programs related to Air Pollution

Implementation

slide-34
SLIDE 34

BudgetMap Visualization

Span multiple budget categories

Implementation

Transportation Health Environment …

slide-35
SLIDE 35

Evaluation

Tagging results and perception changes after tagging tasks

slide-36
SLIDE 36

Evaluation: Procedure

Pre-task Survey Tagging Task Post-task Survey

slide-37
SLIDE 37

Pre-task Survey Tagging Task Post-task Survey

+ Interviews with budget experts

Evaluation: Procedure

slide-38
SLIDE 38

Pre-task Survey Tagging Task Post-task Survey

+ Interviews with budget experts

Evaluation: Procedure

Paid for participation

slide-39
SLIDE 39

Pre-task Survey Tagging Task Post-task Survey Informedness & perspective on the city’s budget.

Evaluation: Procedure

slide-40
SLIDE 40

Pre-task Survey Tagging Task Post-task Survey Changes in informedness & perspective

  • n the city’s budget.

Evaluation: Procedure

slide-41
SLIDE 41

Pre-task Survey Tagging Task Post-task Survey Tagging budget programs with: 1) public safety, 2) air pollution, and 3) youth unemployment.

Evaluation: Procedure

slide-42
SLIDE 42

104 citizens participated through

  • nline communities & social media,

generated 730 tags per participant.

Evaluation: Tag Results

slide-43
SLIDE 43

104 citizens participated through

  • nline communities & social media,

generated 730 tags per participant. Compared with a reference solution constructed by 3 budget experts

Evaluation: Tag Results

slide-44
SLIDE 44

Evaluated by varying the threshold (x) in the majority voting formula:

’Related’ Tags > ‘Unrelated’ Tags + threshold (x)

Evaluation: Tag Results

threshold(x) accuracy(%)

slide-45
SLIDE 45

84% (x=3) of the issue-budget tags were correctly matched to the reference solution.

Evaluation: Tag Results

slide-46
SLIDE 46

Coverage of Participants’ Tags

Public Safety: 7% (843 tags) Air Pollution: 14% (1685 tags) Youth Unemployment: 33% (3973 tags)

threshold(x=3)

Evaluation: Tag Results

Out of Total Possible Tags 3 issues x 4,013 programs = 12,039 tags

slide-47
SLIDE 47

People are capable of contributing expert-quality tags

Evaluation: Summary

slide-48
SLIDE 48

Participants’ level of informedness significantly improved after tagging tasks

“I know in which budget programs Seoul’s tax revenues are spent.”

Evaluation: Survey Results

slide-49
SLIDE 49

Their perspective on government spending marginally improved, remaining on the negative side.

“The city government is executing its budget in the right direction regarding public safety.”

Evaluation: Survey Results

slide-50
SLIDE 50

People gain informational benefits from doing tagging tasks.

Evaluation: Summary

slide-51
SLIDE 51

Lessons

slide-52
SLIDE 52

Confirmed the Feasibility of Issue-Driven Approach

slide-53
SLIDE 53

“We encounter social issues in our daily lives. They made it easier for me to evaluate the suitability of budget programs.” — p045

slide-54
SLIDE 54

“Understanding government budget spending through issues would be effective in solving social problems.” — p026

slide-55
SLIDE 55

“The issue-driven approach would be more intuitive to the general public.” — Officer in the Seoul government

slide-56
SLIDE 56

More importantly,

By looking into the differences among tags

slide-57
SLIDE 57

No ground truth for issue-program tags

slide-58
SLIDE 58

Individuals have different perspectives

slide-59
SLIDE 59

Dealing with

  • pinionated crowds

calls for a different kind of quality control.

slide-60
SLIDE 60

Important to incorporate diverse, contextual, and localized viewpoints

slide-61
SLIDE 61

Limitations & Future Work

Allow people to create new issues & Live deployment & Data spanning multiple years

slide-62
SLIDE 62

BudgetMap is

a step toward 
 participatory democracy

Improve awareness

  • n budgetary issues

Build systems for 
 civic engagement Leverage open government data

slide-63
SLIDE 63

Make Government Great Again!

slide-64
SLIDE 64

budgetwiser.org

Thank you

namwkim@seas.harvard.edu