Approximation and Visualization of Interactive Decision Maps Short course of lectures

1
Approximation and Visualization of Interactive
Decision Maps Short course of lectures

• Alexander V. Lotov
• Dorodnicyn Computing Center of Russian Academy of
  Sciences and
• Lomonosov Moscow State University

2
Lecture 9. Application of Pareto frontier
visualization in Web and in e-democracy

• Plan of the lecture
• 1) Standard Instruments of E-Democracy
• 2) Democratic paradigm of environmental decision
  making
• 3) Few words concerning RGM/IDM technique
• 4) Several applications of the RGM/IDM technique
  on Web
• Participatory Decision Support for
  Integrated River Basin Planning
• e-DEMOCRACIA-CM (Madrid community)
• 5) Modified Pareto frontier visualization
  interactive MCO procedures in Web

3
E-Governance and E-Democracy
E-Governance
E-Democracy
E-Government
E-Participation E-Voting
E-Administration
4
Standard Instruments of E-Democracy
Political Process (iii) Decision (ii)
Formation of opinion (i) Information
acquisition
E-Voting
E-Mail
Chat
Web-sites
0
Infor- Uni- Bi- Trans- mation directional
actional
Technical Complexity
5
The technocratic and the democratic paradigms of
environmental decision making
6
The technocratic paradigm of environmental
decision making

• The technocratic paradigm is a usual concept of
  environmental decision making experts develop a
  water management project, and professional
  decision makers approve, conditionally approve
  with minor changes or reject the project.

7
Democratic paradigm of environmental decision
making

• the power to make decisions must be placed as
  far as possible in the hands of the persons who
  are the most directly influenced by the decision
  concerned, and not in the hands of individual
  decision makers and their experts. The
  expert-oriented paradigm is seen increasingly
  as counterproductive in this respect (M.Abbott
  et al., 1999).

8
Example of a failure of the technocratic paradigm
in the USSR

• An illustration of the failure of the
  technocratic paradigm is can provided by the
  story of the large-scale water management project
  based on partial diversion of the flow of
  Northern Russian rivers into the Volga River
  basin. In 1981 the USSR communist party has made
  a decision to approve the project and start its
  implementation. However, mass protests of
  environmentalists, researchers, writers and
  practically all educated people have resulted
  first in suspension of the project and then in
  its final stop in 1986.

9
Informing non-experts

• Non-experts usually have minimal knowledge on the
  ways, how to solve environmental problems.
  Nevertheless, they want and often are involved
  into political actions related to such problems.
  It is clear that the gap between knowledge and
  actions of non-experts can be misused by
  irresponsible politicians and is dangerous.
• Internet can help non-experts understand the
  environmental problems and base their
  problem-related legal and political actions on
  such knowledge.

10
The democratic paradigm requires special tools
for elaboration of information in a form
accessible for all people
11
Internet tools that support the democratic
paradigm possible requirements

• simplicity
• transparent form
• objectivity of the tools

12
Objectivity of the tools

• Information must be supplied under the same form
  for all stakeholders and must be considered by
  all of them as objective. (Cunge and Erlich,
  1999).

13
Comment A small list of possible alternatives
developed by experts results in an asymmetric
relation between experts and non-experts experts
can develop alternatives and non-experts cannot.
This asymmetric situation is not equitable, the
objectivity principle may be violated. Experts
may use it to thrust their preferences on
non-experts, and non-experts understand it.
14
Once again, decision screening versus final
decision making in environmental problems
15
We try to make the situation symmetric, i.e. to
help non-experts to develop the decision
alternatives by themselves on the basis of
graphic exploration of the whole variety of
feasible decision alternatives. An independent
search for preferable decision alternatives can
make the situation symmetric and objective. The
process of independent decision screening can be
considered as the learning process.
16
Multi-criteria graphic techniques for decision
screening

• The main principles of the methodology are
• Application of a simplified integrated model of a
  environmental system
• Application of a multi-criteria decision support
  tool based on visualization of Pareto frontier

17
Two main tasks to be solved in the framework of
e-participation in public decision problems

• informing lay stakeholders on public decision
  problems (especially on possible strategies for
  solving the problems) and
• supporting the decision making (aggregating
  stakeholders preferences or even negotiations).

18
INFORMING the lay stakeholders

• Web tools based on the IDM technique can help lay
  stakeholders better understand the feasibility
  frontiers and express preferences by selecting
  one or several strategies that best fit their
  concerns.
• It important that it can be done independently of
  mass media that can help thrusting the strategies
  selected by an expert on lay stakeholders.
• The lay stakeholders can base their
  problem-related legal and political actions
  (including e-participation) on such knowledge.

19
Supporting the decision making

• Web applications of the RGM/IDM technique are
  aimed, first of all, at supporting the first,
  pre-negotiation phase and arbitration. However,
  they can be used for supporting negotiations.

20
Few words on the RGM/IDM technique A database
of alternatives in the form of a decision matrix
is considered, i.e., table of N decision
alternatives (rows) given by a finite number of
attributes (columns), a part of which is used as
the selection criteria. One or several preferable
alternatives must be selected.
21
Main features of the problem

• The criteria, which used for selecting a small
  number of alternatives, are assumed to be real
  values. Thus, an alternative is associated with a
  criterion point. The method is based on
  visualization of the Pareto frontier of the
  cloud of criterion points.
• The decision maker has to identify the goal on
  the Pareto frontier of the cloud. Such
  information of the DMs preferences helps to
  select a small number of good alternatives.
  This study can be considered as a special form of
  data mining.

22
Example real estate on sale
23
A simple graphic description of the method
24
For illustrative purposes, let m2 (criterion
points are displayed in the plane). Non-dominated
points are given by crosses.
25
Enveloping the criterion points
26
Approximating the Edgeworth-Pareto hull of the
convex hull (the so-called CEPH)
27
Pareto frontier is analyzed by user and a
preferred combination of criterion values
(reasonable goal) is identified
28
The alternatives that are close to the goal are
selected
29
General case (m from 3 to 8)

• Visualization of the Pareto frontier
• is based on approximation of the CEPH and
  application of the Interactive Decision Maps
  technique for the interactive analysis of the
  frontiers of the slices.

30
Several applications (discussed yesterday)

• Selecting a location for rural health practice in
  Idaho
• Application to local water quality planning in
  Russia (Revival of the Volga River program)
• Mexico APLICACIÓN DE LA MINERÍA DE DATOS EN LA
  LOCALIZACIÓN ÓPTIMA DE INSTALACIONES PETROLERAS
• Mexico Aplicación de la Minería de Datos para la
  exploración óptima de reservas petroleras
• Exploration of pollution abatement cost in the
  Electricity Sector Israeli case study

31
Application of RGM/IDM in Web.Reasonable Goals
for DataBases(RGDB)
32
Concept of the Web RGDB application server
33
Data input
34
Example of the RGDB display
35
Selected alternatives
36
Web RGDB can be found at
http//www.ccas.ru/mmes/mmeda/rgdb/index.htm Or
http//www.rgdb.org/idm/
37
EU Water Framework Directive
38
(No Transcript)
39
Participatory Decision Support for Integrated
River Basin Planning(Funding German Federal
Ministry of Education and Research)

• The Web RGDB was used as a part of DSS
• developed by
• Jörg Dietrich and Andreas H. Schumann,
• Ruhr University Bochum,
• Institute for Hydrology, Water Management and
  Environmental Engineering

40
DSS was calibrated for the Werra River Basin
Ems
Elbe
Weser
Werra
Rhein
41
Dynamically Calculated Decision Matrix
42
For the Participatory Decision Support System, a
special form of the Web RGDB was developed. It
can support negotiations. It applies selecting
several goals and related small groups of
alternatives.
43
(No Transcript)
44
Presentation of RGM/IDM Results
45
Architecture of the Web-based DSS
46
The plan of Werra basin management for the next
five years was developed. Unfortunately, ordinary
people (lay stakeholders) were excluded from the
decision process.Next German project started
now is related to strategies of water management
at the sea shore of Shandong province of China
(Yellow River delta).
47
Another Project E-DEMOCRACIA-CM (Madrid
community)
48
A framework for participatory group decision
support using Pareto frontier visualization, goal
identification and arbitrationR. Efremov, D.
Rios Insua, A. Lotov
49
Application of the research is related to
participatory budget planningGeneral public
must be allowed to have a word and aid in
deciding and approving how public budgets, mainly
in municipalities, are spent.
50
The study is devoted to developing
user-friendly, yet rigorous, Web-based group
decision support methods. The developed methods
are based on interactive Pareto frontier
visualization combined with expression of
preferences in terms of goals and using
goal-based arbitration.
51
A participatory decision making process is
divided into two stages

• At the first stage, instead of providing their
  value functions, the stakeholders express their
  preferences via Web in the form of feasible (or
  reasonable) goals directly on display of the
  Pareto frontier.
• At the second stage, this preference information
  is used in an arbitration procedure to construct
  the group decision.

52
The IDM technique is used for supporting the
stakeholders in identifying their personal goals.
Since the stakeholders supported by visualizing
the Pareto frontier are able to identify their
goals consciously, we assume that such goals are
the result of maximizing the value functions of
stakeholders over the Pareto frontier.
53
Simple-minded arbitration
54
Preference information provided by stakeholders
a(Y)
y1

• The stakeholders identify their reasonable goals
• The weights of Tchebycheff functions are found
  that result in the identified goals

y(1)
y2
55
Goal-based arbitration scheme
y1
a(Y)
The arbitration scheme is based on averaging the
weights of Tchebycheff functions
y(2)
y(1)
y2
56
Using of voting procedures for testing the
arbitration rule
a(Y)

• Participants receive several alternatives, which
  are close to the goal according to the
  Tchebysheff distance then, they assign score to
  the alternatives

6
5
4
2
y(1)
1
2

• The lists of alternatives with score are used in
  voting schemes.

y2
3
57
A group decision support tool for selecting a
hostel in London town

• Criteria
• Location
• Security
• Price
• Cleanliness
• Staff (service)

58
The experiment

• First stage

1. Every participant uses Web to study the problem
   alone he/she obtains the IDM Java applet with
   the Pareto frontier and specifies the goal
2. After it, he/she immediately receives a list of
   about five alternatives that are close to the
   goal
3. Participants has to give scores to the
   alternatives (5 for the best one, 1 for the
   worst).

• Second stage

1. The scores given by participants are summed up
   the alternatives are ranked in accordance to the
   sum.
2. The ranked list of alternatives, along with their
   attributes, is displayed to the participants
   during their face-to-face meeting.
3. The participants are asked to approve the
   alternative with the maximal score as the best
   one (by voting). If they do not accept such an
   alternative, the discussion follows and another
   alternative may be proposed to be voted.

59
Additional information

• Participants International group of students at
  faculty for Cybernetics and Computational
  Mathematics of Lomonosov Moscow State University,
  consisting of 5 Vietnamese and 18 Russian
  students. No instructions were given on the
  decision support method and the Web system before
  the experiment.

• Main goals of the experiment

• Comparing the results of the arbitration rules
  with voting procedure.
• Studying the users feedbacks
• whether this resource and the RGM method are
  intelligible?
• is it worth using this Web resource and
  additional procedures or it could be
• easier to chose a hostel from the table?

60
Results

• The list of 3 hostels with the maximal scores
• Admotel 41 points
• YHA London St. Pauls 30 points
• International Student hostel 26 points.
• The results of voting Admotel is the best.
• Let us compare it with the result of the
  arbitration procedure

61
Comparison of arbitration rule and direct voting

• The goal-based arbitration rule gave
• Security Location Price Cleanliness
  Staff
• YHA London St Pauls 83 91
  18 79 81
• Note that it was second place in the list of best
  scored hotels.
• Compare it with the voting winner
• Security Location Price Cleanliness
  Staff
• Admotel 86 93 18
  86 73
• Thus, the arbitration rule turned out to be close
  to the results of scoring and voting. It makes us
  hope that a balanced decision in the case of a
  large number of stakeholders that are not able to
  meet can be provided by the arbitration rule.

62
Experiments with non-mathematicians

• Experiments with karate sportsmen
• Experiments with orgel musicians
• Experiments with people that have got various
  education
• Experiments with energy engineering students
  (freshmen)
• Easy to use 40, not easy to use 40
• Desirable to use in future 53

63
Application of the arbitration procedure in the
case of finite choice in budget problems(Spain,
URJC)
64

• List of alternative budget allocation decisions
  was formulated and evaluated against selected
  attributes
• Any stakeholder uses the RGDB applet to explore
  the Pareto frontier of the envelope and to
  identify his/her individual reasonable goal
• The arbitration reasonable point was found (using
  the Tchebycheff function with averaged
  parameters)
• The arbitration point was used for selecting one
  or several alternatives of budget allocation. In
  the case of several alternatives, they are ranked
  in accordance to the nearness to the arbitration
  point.

65
Demonstration of the Web resource for selecting a
hostel in London(ask Francesca Pianosi)
66
Modified Pareto frontier visualization
interactive MCO procedures in Web
67
Pareto Step
68
Pareto Step -2
69
Pareto Race with Interactive Decision Maps