Description of the topic
6 Features and Added Value of Simulation Models Using Different Modelling. . . 115
T ab
le 6.
7 (c
on ti
nu ed
)
V ir
S im
M ic
ro S
im M
E L
-C O
C O
P M
O ’s
K os
ic e
ca se
S K
IN
A pp
li ca
ti on
do m
ai n
H ea
lt h
an d
so ci
al po
li cy
w it
h fo
cu s
on pa
nd em
ic in
fl ue
nz a
H ea
lt h
an d
so ci
al po
li cy
w it
h fo
cu s
on pa
nd em
ic in
fl ue
nz a
H ea
lt h,
Ju st
ic e,
E du
ca ti
on ,S
oc ia
l P
ol ic
y w
it h
fo cu
s on
ea rl
y li
fe -c
ou rs
e
E ne
rg y
po li
cy w
it h
fo cu
s on
us er
be ha
vi ou
r to
w ar
ds en
er gy
ef fi
ci en
cy an
d re
ne w
ab le
s
K no
w le
dg e-
in te
ns iv
e in
du st
ri es
,E ur
op ea
n an
d N
at io
na l
in no
va ti
on po
li ci
es
C on
st ra
in ts
of th
e us
ag e
M is
si ng
im po
rt an
t pa
ra m
et er
s T
o al
l pe
op le
w it
hi n
ag e
gr ou
ps as
si gn
ed sa
m e
pr ob
ab il
it y
U se
s pa
rt ic
ul ar
da ta
fr om
S w
ed en
N ot
ro bu
st to
ch an
ge s
in th
e po
pu la
ti on
da ta
L im
it ed
by av
ai la
bl e
da ta
so ur
ce s
U n-
di re
ct io
na l
re la
ti on
sh ip
s be
tw ee
n va
ri ab
le s
w it
h no
fe ed
ba ck
C ha
ng in
g th
e di
st ri
bu ti
on of
va ri
ab le
s no
t th
e ef
fe ct
s. D
is cr
et e
ti m
e on
ly
A ge
nt -b
as ed
m od
el li
ng ap
pl ic
ab le
fo r
ex am
in in
g so
ci al
be ha
vi ou
r bu
t ca
nn ot
be th
e on
ly so
ur ce
fo r
po li
cy -m
ak in
g M
od el
ca n
be co
m e
qu it
e co
m pl
ex de
pe nd
in g
on th
e nu
m be
r of
ag en
ts
M od
el ap
pl ic
ab le
fo r
a do
m ai
n of
kn ow
le dg
e an
d ag
en t
ne tw
or ks
em be
dd ed
in a
dy na
m ic
en vi
ro nm
en t
E xa
m pl
es of
us e
P ol
ic y-
m ak
in g
un de
r pa
nd em
ic in
fl ue
nz a
in S
w ed
en in
20 09
P ol
ic y-
m ak
in g
un de
r pa
nd em
ic in
fl ue
nz a
in S
w ed
en in
A ut
um n
of 20
09
Il lu
st ra
ti ve
ap pl
ic at
io n
to so
ci al
de te
rm in
an ts
of he
al th
an d
en d-
us er
en ga
ge m
en t
E ne
rg y
po li
cy to
ex pl
or e
st ak
eh ol
de r
be ha
vi ou
r in
K S
R ,S
K F
ur th
er m
od el
s ge
ne ra
te d:
H ou
si ng
P ol
ic y
in L
on do
n, U
K K
no w
le dg
e T
ra ns
fe r
su pp
or t
in C
am pa
ni a
R eg
io n,
It al
y
E U
pr oj
ec ts
S E
IN ,
N E
M O
,M an
E T
E I,
S M
A R
T ,G
R E
A T
T ra
ns fe
ra bi
li ty
an d
re -u
sa bi
li ty
R e-
us ab
le fo
r ot
he r
av ai
la bl
e da
ta bu
t no
t fo
r ot
he r
do m
ai ns
an d
co nt
ex ts
R e-
us ab
le fo
r ex
am in
in g
po li
ci es
ba se
d on
de m
og ra
ph ic
s
G en
er ic
ap pl
ic ab
il it
y in
ea rl
y li
fe -c
ou rs
e an
al ys
is P
os si
bl e
to ex
te nd
m od
el to
la te
r pe
ri od
s in
th e
li fe
-c ou
rs e
an d
to ot
he r
do m
ai ns
O th
er dy
na m
ic so
ci o-
de m
og ra
ph ic
pr oc
es se
s
R e-
us e
li m
it ed
to pa
rt ic
ul ar
po li
cy an
d re
gi on
al co
nt ex
t. M
od el
ca n
se rv
e as
a bl
ue pr
in t
fo r
ne w
m od
el s
S K
IN ca
n be
us ed
in va
ri ou
s di
sc ip
li na
ry co
nt ex
ts w
or ki
ng w
it h
kn ow
le dg
e an
d ag
en t
ne tw
or ks
em be
dd ed
in dy
na m
ic en
vi ro
nm en
ts
w.jager@rug.nl
116 D. Majstorovic et al.
T ab
le 6.
7 (c
on ti
nu ed
)
V ir
S im
M ic
ro S
im M
E L
-C O
C O
P M
O ’s
K os
ic e
ca se
S K
IN
L im
it at
io ns
of m
od el
s/ su
gg es
ti on
s fo
r ex
te ns
io ns
L im
it a
ti o
n s:
m od
el pa
ra m
et er
s an
d eq
ua ti
on s
ar e
fi xe
d an
d ca
nn ot
be m
od ifi
ed S
u gg
es ti
o n
s: E
xt en
d th
e m
od el
to su
pp or
t de
fi ni
ti on
of cu
st om
va ri
ab le
s A
ll ow
fo r
m od
el li
ng si
m il
ar ph
en om
en a
A ll
ow fo
r tr
an sf
er ab
il it
y to
ot he
r do
m ai
ns
L im
it a
ti o
n s:
th e
m od
el is
no t
ro bu
st to
w ar
ds de
m og
ra ph
ic ch
an ge
s. S
u gg
es ti
o n
: al
lo w
fo r
re al
is ti
c si
m ul
at io
n of
in fl
ue nz
a ou
tb re
ak s
L im
it a
ti o
n s:
re st
ri ct
ed to
a no
ti on
al ‘e
vi de
nc e-
ba se
d’ /s
ci en
ce -i
nf or
m ed
ap pr
oa ch
co nc
ep tu
al ly
pr ed
ic at
ed on
th e
pr im
ac y
of so
ci al
de te
rm in
an ts
L im
it ed
ro le
of st
ak eh
ol de
rs .
S u
gg es
ti o
n s:
C om
bi na
ti on
of a
re al
is ti
c da
ta fr
am ew
or k
an d
es ti
m at
es de
ri ve
d fr
om tr
ia ls
an d
sy st
em at
ic re
vi ew
s U
se fu
l fo
r en
d- us
er s
L im
it a
ti o
n s:
D ec
la ra
ti ve
ru le
en gi
ne no
t ea
sy to
pr og
ra m
C on
ce pt
ua l
an d
si m
ul at
io n
m od
el no
t ea
sy to
un de
rs ta
nd fo
r en
d- us
er s.
S u
gg es
ti o
n s:
E xt
en d
th e
C C
D to
ol an
d tr
an sf
or m
at io
n to
ol to
co nn
ec t
to ot
he r
A B
M to
ol s
su ch
as A
ny L
og ic
,N et
L og
o et
c. E
xt en
d th
e m
od el
tr an
sf or
m at
io n
to su
pp or
t al
so ba
ck w
ar d
co ns
is te
nc y
fr om
si m
ul at
io n
m od
el to
co nc
ep tu
al m
od el
as pr
og ra
m m
er s
m ay
ch an
ge si
m ul
at io
n co
de ea
si ly
L im
it a
ti o
n s:
T he
ul ti
m at
e po
li cy
ob je
ct iv
es fo
r th
e su
pp or
t of
re se
ar ch
an d
de ve
lo pm
en t
ne ed
ye t
to be
de te
rm in
ed S
u gg
es ti
o n
s: Im
po rt
an t
to de
ve lo
p m
et ho
ds of
th e
re se
ar ch
pr es
en ta
ti on
s to
be in
te re
st in
g an
d co
m pr
eh en
si bl
e to
a po
li cy
-m ak
in g
au di
en ce
w.jager@rug.nl
6 Features and Added Value of Simulation Models Using Different Modelling. . . 117
The main advantage of system dynamics models is that they are fast to run and technologically not demanding while providing useful information about the real- world processes and insights into possible impacts of different macro-level policies. However, these models face a number of restrictions. For example, VirSim initially assumes infection probabilities where elderly people (age group 60 and more) have considerably fewer chances of being infected with influenza compared to the other two age groups (Fasth et al. 2010). However, the SEIR model that was applied cannot predict this and cannot explain why this occurs. The authors of VirSim used this result from the micro-simulation model MicroSim and assumed this phenomenon happens because of less social contacts of elderly people or some prior immunity. VirSim cannot explain this phenomenon because system dynamics does not include modelling of various social interactions and other similar dependencies between actors since all variables are averaged over particular groups or the population in general - in the case of VirSim within the members of a particular age group. Apart from the categories of people based on their age, VirSim cannot identify fine-grain groups that have higher probability to be infected. For example, a student has more chances to be exposed and therefore infected than a researcher working in the same university but more in the closed environment of an office while students usually have more frequent social interactions among their groups and communities. It is important to identify closed environments that have high risk of spreading influenza, for example boarding and nursing homes. From the policy modelling point of view, it is important to identify high-risk groups to start the vaccination from there. One could define refined categories of actors by defining more variables, but in general, it would not be possible to represent relations between subcategories, such as taxonomies or ontologies needed to represent social contacts or interactions among actors, due to the lack of representation apparatus in system dynamics models.
As Gilbert and Troitzsch argue, due to social complexity and non-linearity, it is difficult to describe processes and systems analytically. To be able to examine interactions between simulation units, other modelling techniques such as ABM or micro-simulation models need to be applied for exploring the social heterogeneity and structures (Gilbert and Troitzsch 2005).
Micro-simulation models, usually based on a weighted sum of a representative sample of the population, consider characteristics of individuals and are able to re- produce social reality (Martini and Trivellato 1997). They are beneficial in predicting both, short-term as well as long-term impact of policies (Gilbert and Troitzsch 2005). However, micro-simulation models are costly to build and complex, especially at the level of data analysis requirements. In the case of MicroSim, the Swedish population of approximately nine million people was modelled in many details (Brouwers et al. 2009). In addition, in ‘simple’ cases, especially in demographics, a micro-simulation model produces similar results as a system dynamics-based model (Gilbert and Troitzsch, 2005). This proved true in the case of MicroSim and VirSim: The lat- ter confirmed the results of the former, although with a greater difference between vaccination and non-vaccination results (The National Board of Health and Wel- fare 2011). According to Spielauer, micro-simulation is best to use when population heterogeneity matters; when there are too many possible combinations to split the
w.jager@rug.nl
118 D. Majstorovic et al.
population into a manageable number of groups; in situations when the micro level explains complex macro-behaviours, or when individual history is important for the model’s outcomes (Spielauer 2011).
Although agent-based models lack clear predictive possibilities, they are consid- ered a highly valuable tool for describing and explaining complex social interactions and behaviours, contributing to the understanding of real-world social systems and to a better management of different social processes. Schindler argues that agent-based simulations are capable of representing real-world systems, where small changes in parameter values induce big changes in the model’s outputs. This property shifts attention from the importance of predictions of the system’s future behaviour to the management of critical (social) processes responsible for the changes. However, agent-based simulations alone are not sufficient to model reality. Another possi- ble problem is a high degree of freedom in modelling agents, which amplifies the importance of a proper validation of a simulation model (Schindler 2013).
While agent-based and micro-simulation models would be able to show that an elderly person has less infection probability, it is questionable whether they would be able to answer why an elderly person is less infected by influenza. Knowing ‘why’ can help in building a successful strategy for protection against the disease. It might happen that hidden variables and parameters influence this age group. For this reason, in order to model correct probabilities for different age groups, several authors suggest that uncertainty models, such as (dynamic) Bayesian models or Markov chains could be used. In addition, if the past should be also considered (for example, a person has less chances to be infected now because he/she was infected in the recent past), then we have to use more complex probability models, such as the Dempster–Shafer model (Ronald and Halpern 1991, Jameson 1996). Gilbert and Troitzsch argue that statistical models can also be used to predict values of some dependent variables. However, statistical models assume linear relationships between parameters, which becomes a restrictive assumption in the case of (complex) social systems (Gilbert and Troitzsch 2005).
Comparing the three different paradigms of social and policy modelling explored in this chapter, the three approaches can be examined according to the level of gran- ularity they are focussing on, the complexity of the models, the demand for the amount of data needed to generate a valuable simulation model and whether social behaviour is modelled. Table 6.8 provides this comparison, which is adapted from (Gilbert and Troitzsch 2005). Micro-simulation models represent particular ontolo- gies of the population or its representative subset based on individuals and are most demanding regarding data needed for developing a model. Agent-based models are less data demanding, less complex and well suited for representing groups of actors (which can represent individuals, groups as well as a system as a whole) and their social behaviour. ABM is the only one of the three paradigms studied which models social behaviour. However, social behaviour cannot be the only source for policy- making (Gilbert and Troitzsch 2005) Macro-models, in this chapter represented by system dynamics, are the least demanding—they model a situation at a global level and require the least data. Nevertheless, they are better for the analysis of short-term policy impacts than for longer-term perspectives (Astolfi et al. 2012).
w.jager@rug.nl
6 Features and Added Value of Simulation Models Using Different Modelling. . . 119
Table 6.8 Comparison of simulation modelling theories along level of granularity, the complexity of the models, the amount of data needed to generate simulation models, and the modelling of social behaviour of agents
Simulation paradigm
Granularity Complexity Data needed Behavioural
System dynamics Macro-focusing on the system as a whole
Low Aggregated data No
Micro-simulation Micro-focusing on individuals
High High amount at individual level
No
Agent-based modelling
Micro-macro— focusing on interaction of agents (which can be individuals as well as a system)
Medium-high Low to moderate (depending on the number of agents and the policy context)
Yes
The analysis of three different modelling paradigms with the comparison of five different simulation models has shown that each of the modelling approaches has strengths and weaknesses that constrain their usage in policy-making. For example, micro-simulation can be used for representing social structures while ABM examines interactions between the agents. Astolfi et al argue that none of the theories alone is able to address complex policy interactions (Astolfi et al. 2012). In consequence, a necessary step in the development of simulation modelling is to build and explore ways of maintaining complex simulation models consisting of a few sub-models built on different modelling theories, which communicate with each other by set- ting up and propagating particular parameters after each reasoning iteration (Astolfi et al. 2012). These hybrid models can be considered as modelling platforms or com- plex systems consisting of sub-models. Yet, it is necessary to study methodologies and possibilities of combining different modelling paradigms in order to provide reliable simulation platforms. Current research indicates this trend, as an example of micro-macro combination in a Chronic Disease Prevention Model developed in Australia shows (Brown et al. 2009). However, more research is needed to better understand the implications of combined modelling paradigms, to develop innova- tive simulation platforms that support easy adjustment and development of different models based on different modelling paradigms and to bring evangelists of particular modelling paradigms closer to each other to support mutual understanding and the exploration of the added value and benefits of particular simulation models. Further recommendations and indications of research needs include, but are not exhaustively listed:
• Providing guidelines for how to best choose and arrange a collection of smaller (sub) models each describing certain aspects of a given domain of modelling;
• Finding the junction points of those models of distinct modelling paradigms with each other by defining input and output parameters for each of the sub-models;
• Developing meta-models that reflect the combinatory use of distinct modelling approaches;
w.jager@rug.nl
120 D. Majstorovic et al.
• Determining the workflow of a simulation process by means of, e.g. a sequence and timing of exchanging the input and output parameters between sub-models in a combined hybrid meta-model;
• Exploring more extensive engagement of stakeholders in the policy development9; • Developing more comprehensive simulation platforms that enable the combina-
tion of different simulation paradigms in an easy way.
6.5 Conclusions
In this chapter, we have examined and compared five different simulation mod- els, which were built on three different modelling paradigms: system dynamics, micro-simulation, and ABM. The chapter first provided an overview of the main char- acteristics of each of the modelling paradigms and then described the five simulation models by outlining them according to a framework elaborated by eGovPoliNet for comparative analysis of knowledge assets. The simulation models are each suitable for representing different aspects of socio-political and/or socio-economic phenom- ena, such as demographic processes (education, social contacts, spread of diseases, etc.), innovation processes or natural resource consumptions (e.g. energy consump- tion). The comparison has revealed the major differences as well as added value and limitations of the different approaches and simulation models. Some lessons from the comparative analysis are that the main strengths of using simulation models in policy- making are the possibilities of exploring and creating understanding of real-world systems and relationships, of experimenting with new situations and of forecasting outputs of alternative policy options or situations based on the given values of pa- rameters. Another key added value of simulation models in policy-making is that simulation models enable the exploration of social processes to evaluate potential impacts of alternative policy options on real-world situations and thus to identify the most suitable policy option.
Current paradigms of policy modelling using simulation models are however constrained by their particular focus. Yet, our real-world systems and social processes are complex and require the consideration of parameters at different levels: macro- level, micro-level as well as social behaviour and interconnections between actors. Accordingly, applying one singular approach to modelling a real-world problem is constrained by the particular modelling approach it focuses on: A simulation model of system dynamics may therefore lack precision and social interactions because the missing factors are not accounted for. While the demand for meeting the appropriate level of details included in a model’s description, being not too complex and also not too simple, determines the success of a simulation model, there is a rising need for integrating and combining different modelling paradigms to accommodate the diverse aspects to be considered in complex social world policy contexts. Unifying
9 A more detailed discussion of stakeholder engagement in policy making is given in (Helbig et al. 2014).
w.jager@rug.nl
6 Features and Added Value of Simulation Models Using Different Modelling. . . 121
different modelling theories under an umbrella of comprehensive policy modelling platforms is a research need identified in this chapter. Such research should put forward a meta-model of how individual simulation paradigms can be combined, and suggestions of ‘clever’junctions of individual smaller (and self-contained) simulation models dedicated to individual aspects to be modelled.
While this chapter selected three widely used simulation paradigms for the study, it does not claim to be exhaustive nor comprehensive. Further research is needed to ex- tend the study to involve other important modelling approaches such as theory-based macro-economic forecasting for instance Dynamic Stochastic General Equilibrium (DSGE) modelling. DSGE is exemplified by the Global Economy Model (GEM) which provides support in policy analysis to central banks and the International Monetary Fund (IMF) (Bayoumi 2004). This will further add to understanding the scope and limitations of different modelling paradigms, as for example Farmer and Foley argue, too, that instead of DSGE models, agent-based models should be used to model the world economy (Farmer and Foley 2009). Thus, the authors of this chapter recognise the need to continue comparative analysis as carried out in this contribution and to expand the research to incorporate further modelling paradigms as well as other public policy domains. Insights gained will help build up better hybrid models of social simulation paradigms that are better able to cope with the complexity of our social and dynamic world systems and that are more reliable as they are covering the various social, policy and economic aspects at various levels of abstraction and giving consideration in a more comprehensive way. Accordingly, better social simulation modelling platforms will emerge.
References
Ahrweiler P, Pyka A, Gilbert N (2004) Simulating knowledge dynamics in innovation networks (SKIN). In: Leombruni R, Richiardi M (eds) The agent-based computational approach. World Scientific Press, Singapore, pp 284–296
Ahrweiler P, Gilbert N, Pyka, A (2011a) Agency and structure. A socials simulation of knowledge- intensive industries. Comput Math Org Theor 17(1):59–76
Ahrweiler P, Pyka A, Gilbert N (2011b) A new model for university-industry links in knowledge- based economies. J Prod Innov Manage 28(2):218–235
Anderson RE, Hicks C (2011) Highlights of contemporary microsimulation. Soc Sci Comput Rev 29(1):3–8
Astolfi R, Lorenzoni L, Oderkirk J (2012) A comparative analysis of health forecasting methods. OECD Health Working Papers, No. 59. OECD Publishing. doi:10.1787/18152015
Bayoumi T (2004) GEM: a new international macroeconomic model, Occasional Paper 239. Washington, D.C.: International Monetary Fund. http://213.154.74.164/invenio/record/ 12437/files/bayoumiop239.pdf. Accessed 29th July 2014
Brouwers L, Cakici B, Camitz M, Tegnell A, Boman M (2009a) Economic consequences to society of pandemic H1N1 influenza 2009– preliminary results for Sweden. Euro Surveil- lance 14(37):pii= 19333. http://www.eurosurveillance.org/ViewArticle.aspx?ArticleId = 19333 Accessed 29 July 2014
Brouwers L, Camitz M, Cakici B, Mäkilä K, Saretok P (2009b) MicroSim: modeling the Swedish Population. http://arxiv.org/abs/0902.0901. Accessed 29 July 2014
w.jager@rug.nl
122 D. Majstorovic et al.
Brown L, Harris A, Picton M, Thurecht L, Yap M, Harding A et al. (2009). Linking micro-simulation and macro-economic models to estimate the economic impact of chronic disease prevention. In: Zaidi A, Harding A, Williamson P (eds) New frontiers in microsimulation modelling. Ashgate, European Centre Vienna, pp 527–556
Butka P, Mach M, Furdik K, Genci J (2011) Design of a system architecture for support of col- laborative policy modelling processes. In: Proceedings of SACI 2011, 6th IEEE International Symposium on Applied Computational Intelligence and Informatics. Red Hook, NY, USA, Òbuda University, Hungary, IEEE, Curran Associates, Inc, pp 193–198
Carrat F, Vergu E, Ferguson N, Lemaitre M, Cauchemez S, Leach S, Valleron A (2008) Time lines of infection and disease in human influenza: a review of volunteer challenge studies. Am J Epidemiol 167(7):775–785
Collins A, Joseph D, Bielaczyc K (2004) Design research: theoretical and methodological issues. J Learn Sci 13(1):15–42
Davis P, Lay-Yee R, Pearson J (2010) Using micro-simulation to create a synthesized data set and test policy options: the case of health service effects under demographic ageing. Health Policy 97:267–274
Farmer JD, Foley D (2009) The economy needs agent-based modelling. Nature 460:685–686 Fasth T, Ihlar M, Brouwers L (2010) VirSim—a model to support pandemic policy making. PLOS
Currents Influenza. 2010 Sep 22, 2: RRN1181. doi:10.1371/currents.RRN1181 Fergusson D, Horwood L, Shannon F, Lawton J (1989) The Christchurch child development study:
a review of epidemiological findings. Paediatr Perinat Epidemiol 3:278–301 Forrester JW (1961) Industrial dynamics. MIT Press, Cambridge Fraser C, Donnelly CSC, WP H, MD VK, TD H et al. (2009). WHO rapid pandemic assessment
collaboration. Pandemic potential of a strain of influenza A (H1N1): early findings. Science 324(5934). PubMed PMID: 19433588, pp. 1557–1561
Gilbert N (2007) Agent-based models. Sage, London Gilbert N, Troitzsch K (2005) Simulation for the social scientist. Open University Press, Maidenhead Gilbert N, Pyka A, Ahrweiler P (2001) Innovation networks—a simulation approach. J Artif Soc
Soc Simul 4(3) Gilbert N, Ahrweiler P, Pyka A (2007) Learning in innovation networks—some simulation
experiments. Physica A: Stat Mech Appl 374(1):100–109 Gilbert N, Ahrweiler P, Pyka A (2014) Simulating knowledge dynamics in innovation networks.
Springer, Heidelberg Greenwood DJ, Levin, M. (2006) Introduction to action research: social research for social change.
Sage, Thousand Oaks Harrison RJ, Lin Z, Carroll GR, Carley KM (2007) Simulation modeling in organizational and
management research. Acad Manage Rev 32(4)1229–1245 Helbig N, Dawes SS, Dzhusupova Z, Klievink B, Mkude CG (2014) Stakeholder engagement in
policy development: observations and lessons from international experience. In: Janssen M, Wimmer MA, Deljoo A (eds) Policy practice and digital science—integrating complex systems, social simulation and public administration in policy research. Springer Science, Berlin
Jameson A (1996) Numerical uncertainty management in user and student modeling: an overview of systems and issues. User modeling and user adapted interaction, 5, special issue on numerical uncertainty management in user and student modeling, pp 193–251
Kamateri E, Panopoulou E, Tambouris E, Tarabanis K, Ojo A, Lee D, Price D (2014) A comparative analysis of tools and technologies for policy making. In: Janssen M, Wimmer MA, Deljoo A (eds) Policy practice and digital science—integrating complex systems, social simulation and public administration in policy research. Springer Science, Berlin
Lay-Yee R, Milne B, Davis P, Pearson J, McLay J (2014) Determinants and disparities: a simulation approach to the case of child health care. Under Review
Lengwiler Y (2004) A monetary policy simulation game. J Econ Educ 35(2):175–183 Lotzmann U, Meyer R (2011) DRAMS—a declarative rule-based agent modelling system. In:
Proceedings of 25th European Conference on Modelling and Simulation (pp 77–83)
w.jager@rug.nl
6 Features and Added Value of Simulation Models Using Different Modelling. . . 123
Lotzmann U, Wimmer MA (2013) Evidence traces for multi-agent declarative rule-based policy simulation. In Proceedings of the 17th IEEE/ACM International Symposium on Distributed Simulation and Real Time Applications (DS-RT 2013). IEEE Computer Society, pp115–122
Mannion O, Lay-Yee R, Wrapson W, Davis P, Pearson J (2012) JAMSIM: a microsimulation modelling policy tool. J Artif Soc Soc Simul 15(1):8
Maria A (1997) Introduction to modelling and simulation. Proceedings of the 29th Conference on Winter Simulation, pp 7–13
Martini AP, Trivellato U (1997) The role of survey data in microsimulation models for social policy analysis. Labour 02/1997, 11(1):83–112. (doi:10.1111/1467–9914.00030)
Mayring P (2011) Qualitative Inhaltsanalyse: Grundlagen und Techniken. Beltz, Weinheim und Basel
McLay J, Lay-Yee R, Milne B et al (2014) Statistical modelling techniques for dynamic microsimulation: An empirical performance assessment. Under review
Milne BJ, Lay-Yee R, McLay J, Tobias MTP, Armstrong A, Lynn R et al (2014) A collaborative approach to bridging the research-policy gap through the development of policy advice software. Evid Policy 10(1):127–136
Orcutt G (1957) A new type of socio-economic system. Rev Econ Stat 39(2):116–123 Pyka A, Gilbert N, Ahrweiler P (2007) Simulating knowledge-generation and—distribution
processes. Innov Collab Netw, Cybern Syst 38:667–693 Ronald F, Halpern JY (1991) Uncertainty, belief and probability. Computational Intell 7:160–173 Scherer S, Wimmer MA, Markisic S (2013) Bridging narrative scenario texts and formal policy
modeling through conceptual policy modeling. Artif Intell Law 21(4):455–484 Schindler J (2013) About the uncertainties in model design and their effects: an illustration with a
land-use model. J Artif Soc Soc Simul 16(4):6 Scholz R, Nokkala T, Ahrweiler P, Pyka A, Gilbert N (2010) The agent-based NEMO model
(SKEIN): simulating European framework programmes. In: Ahrweiler P (ed) Innovation in complex social systems. Routledge Studies in Global Competition (pp 300–314)
Sedehi H (2006) GAIM (Gestione Accoglienza IMmigrati): a System Dynamics Model for Im- migration “housing” Management. In: Proceedings of International Conference of the System Dynamics Society. Nijmegen, The Netherlands (pp 3359–3367)
Solar O, Irwin A (2010) A conceptual framework for action on the social determinants of health. Social Determinants of Health Discussion, Paper 2: Policy and Practice. World Health Organization, Geneva
Spielauer M (2011) What is social science microsimulation. Soc Sci Comput Rev 29(1):9–20 Srbljanovic A, Skunca O (2003) An introduction to agent-based modelling and simulation of social
processes. Interdiscip Descr Complex Syst 1(1–2):1–8 Statistics Sweden (Statistiska centralbyrån, SCB). (2009). Sweden’s population by sex and age on
31/12/2008. http://www.scb.se/Pages/TableAndChart____262460.aspx. Accessed 28 July 2014 The National Board of Health and Welfare. (2011). A(H1N1) 2009– An evaluation of Sweden’s
preparations for and management of the pandemic. Evaluation Report number 2011–8–4 van Egmond S., Zeiss R (2010) Modeling for policy science-based models as performative boundary
objects for Dutch policy making. Sci Stud 23(1):58–78 Wallinga JTP (2006) Using data on social contacts to estimate age-specific transmission parameters
for respiratory-spread infectious agents. Am J Epidemiol 164(10):936–944 Wimmer MA (2011) Open government in policy development: from collaborative scenario texts to
formal policy models. Proceedings of 7th International Conference on Distributed Computing and Internet Technologies (ICDCIT—2011) (pp 76–91). Berlin, Springer
Wimmer MA, Scherer S, Moss S, Bicking M (2012) Method and tools to support stake- holder engagement in policy development. The OCOPOMO project. Int J Electron Gov Res 8(3):98–119
Zaidi A, Harding A, Williamson P (2009) New frontiers in microsimulation modeling (Public policy and social welfare. Ashgate Publishing Ltd, England
Zeigler B (1976) Theory of modeling and simulation. Wiley, New York
w.jager@rug.nl
Chapter 7 A Comparative Analysis of Tools and Technologies for Policy Making
Eleni Kamateri, Eleni Panopoulou, Efthimios Tambouris, Konstantinos Tarabanis, Adegboyega Ojo, Deirdre Lee and David Price
Abstract Latest advancements in information and communication technologies of- fer great opportunities for modernising policy making, i.e. increasing its efficiency, bringing it closer to all relevant actors, and enhancing its transparency and acceptance levels. In this context, this chapter aims to present, analyse, and discuss emerging information and communication technologies (ICT) tools and technologies present- ing the potential to enhance policy making. The methodological approach includes the searching and identification of relevant tools and technologies, their system- atic analysis and categorisation, and finally a discussion of potential usage and recommendations for enhancing policy making.
E. Kamateri (�) · E. Panopoulou · E. Tambouris · K. Tarabanis Information Technologies Institute, Centre for Research & Technology—Hellas, Thessaloniki, Greece e-mail: ekamater@iti.gr
E. Panopoulou e-mail: epanopou@iti.gr
E. Tambouris · K. Tarabanis University of Macedonia, Thessaloniki, Greece e-mail: tambouris@iti.gr, tambouris@uom.gr
K. Tarabanis e-mail: kat@iti.gr, kat@uom.gr
A. Ojo · D. Lee INSIGHT Centre for Data Analytics, NUIG, Galway, Ireland e-mail: adegboyega.ojo@deri.org
D. Lee e-mail: deirdre.lee@deri.org
D. Price Thoughtgraph Ltd, Somerset, UK e-mail: david@debategraph.org © Springer International Publishing Switzerland 2015 125 M. Janssen et al. (eds.), Policy Practice and Digital Science, Public Administration and Information Technology 10, DOI 10.1007/978-3-319-12784-2_7
w.jager@rug.nl
126 E. Kamateri et al.
7.1 Introduction
Policy making may be defined as “the process by which governments translate their political vision into programmes and actions to deliver ‘outcomes’ desired changes in the real world” (UK Government 1999). Policy making encompasses any activity relevant to discussing political issues, identifying areas of improvement or solutions, creating and implementing laws and regulations, monitoring and evaluating current policies, etc.
Policy making is a multidisciplinary scientific field referring mainly to politi- cal science, but it may also refer to social, economics, statistics, information, and computer sciences. These diverse scientific fields are essential in order to perform policy making in a more effective and informed manner. Information and communi- cation technologies (ICTs), specifically, have supported decision-making processes for many years. However, the current ICT advancements and good practices of- fer even greater opportunities for modernising policy making, i.e. increasing its efficiency, bringing it closer to all relevant actors and increasing participation, facil- itating its internal processes (e.g. decision making), and enhancing its transparency and acceptance levels.
In this context, this chapter aims to present, analyse, and discuss emerging ICT tools and technologies presenting the potential to enhance policy making. Our ap- proach includes searching and identification of relevant tools and technologies, their systematic analysis and categorisation, and finally a discussion of potential usage and recommendations for enhancing policy making. The chapter is structured in the following way: Sect. 7.2 describes our methodological approach, Sect. 7.3 provides the comparative analysis, and Sect. 7.4 discusses the findings and concludes the chapter.
Before proceeding to the rest of the chapter, we should provide further clarifica- tions with regard to its scope. First, for work presented in this chapter, policy making is considered as a broad and continuous process that commences from the need to create a policy and ends when a policy is abandoned or replaced. In this context, the policy-making process is usually described with a circular-staged model called “the policy cycle”. There are differences in the number, names, and boundaries of the stages adopted in each proposed policy cycle (e.g. Jann and Wegrich 2006; Northern Ireland Government 2013); however, every policy cycle includes an initiation stage, a drafting stage, an implementation stage, and an evaluation stage. The scope of our work refers to all these stages of the policy cycle.
Second, we consider all stakeholders relevant to policy making within the scope of work presented in this chapter. Obviously, the main actor involved in policy making is the government with its different roles, bodies, and institutions. However, noninstitutional actors are also involved such as political parties, political consultants and lobbyists, the media, nongovernmental organisations, civil organisations, and other interested parties depending also on the policy topic at hand. Last but not least, individual citizens are also actors of policy making; as the final policy recipients and beneficiaries, they should actively participate in policy making. Hence, in this
w.jager@rug.nl
7 A Comparative Analysis of Tools and Technologies for Policy Making 127
chapter, we do not consider policy making as a close, internal government process, but rather as an open, deliberative process relevant to the whole society.
7.2 Methodology
In order to analyse the existing ICT tools and technologies that can be used to enhance the policy-making process, we adopted a simple methodology consisting of four main steps.
Before introducing the adopted methodology, we provide a short description with regard to the difference between ICT tools and technologies. ICT tools normally in- clude software applications, web-based environments, and devices that facilitate the way we work, communicate, and solve problems. These are developed by individual software developers, big software providers, researchers, and scientists (Phang and Kankanhalli 2008). Technology, on the other hand, refers to knowledge and know- how, skills, processes, tools and/or practices.1 Therefore, technology not only refers to tools but also the way we employ them to build new things. In the current survey, we organise the findings of our literature analysis based on tool categories.
Step 1: Identification During this step, we surveyed the current state of the art to identify ICT tools and technologies that have been (or have a clear potential to be) used to reinforce the policy-making process. These tools have been collected mainly from project deliverables, posts, electronic articles, conference papers, scientific journals, and own contacts and expertise.
In particular, we searched for tools and technologies that have been highlighted, used, or created by existing research and coordination projects in the area of e-government and policy modelling, i.e. CROSSOVER2, e-Policy3, FuturICT4, OCOPOMO5, COCKPIT6 and UbiPol7, OurSpace8, PuzzledbyPolicy9, etc. This investigation resulted in a collection of more than 30 ICT tools and technolo- gies mainly coming from project deliverables, posts, electronic articles, conference papers, scientific journals, and own contacts and expertise.
Thereafter, we expanded our research on the web to include additional tools that were not previously identified. To this end, we tried multiple searches in the major research databases of computer science, e.g. Association for Computing Machin- ery (ACM) Digital Library and Google Scholar using a combination of different
1 http://en.wikipedia.org/wiki/Technology. 2 http://crossover-project.eu. 3 http://www.epolicy-project.eu/node. 4 http://www.futurict.eu. 5 http://www.ocopomo.eu. 6 http://www.cockpit-project.eu. 7 http://www.ubipol.eu/. 8 http://www.ep-ourspace.eu/. 9 http://www.puzzledbypolicy.eu/.
w.jager@rug.nl
128 E. Kamateri et al.
keywords such as tools, technologies, policy modelling, online participation, en- gagement, government, policy making, decision making, policy formulation, etc. The references of the selected papers were checked and additional papers were found. Some of the journals that have been reviewed include Government Infor- mation Quarterly, International Journal of Electronic Government Research. In addition, we surveyed similar initiatives that summarise tools or/and methods, i.e. the Participation Compass10 launched by Involve11 (not-for-profit organisation in public participation), the ParticipateDB12 by Intellitics13, and the ReformCompass by Bertelsmann Stiftung14 (providers of digital engagement solutions). The final result of this exercise was a list of 75 tools and technologies.
Step 2: Categorisation Analysing the identified tools and technologies, it was ev- ident that most of them fall under a number of categories. We defined, therefore, 11 categories of tools and technologies for policy making. Each category has a spe- cific application focus, e.g. opinion mining, serious games, etc., and may be further divided into one or more subcategories.
We then organised tools and technologies’ analysis according to the defined cate- gories. There are few cases, however, where the same tool could be classified under more than one category, i.e. in the case of visualisation and argumentation tools and in the case of serious games and simulation tools. In the first case, argumentation tools represent and structure arguments and debates, and usually exploit visual means in order to clearly represent the arguments. However, the main focus remains the representation of arguments. On the other hand, the visualisation tools present, in a graphical form, any type of input data. Thus, it was selected for the sake of simplicity to analyse each tool in one category according to its most prominent feature. Similar difficulties in categorisation have also arisen in the case of simulation tools and seri- ous games. Serious games are created for educational and entertainment purposes, or for helping citizens to further understand some processes by playing the role of a key stakeholder. On the other hand, simulation tools are usually created on a more serious context (e.g. within a research project, taking into account accurate real-world data) in order to help real policy makers or governments to simulate long-term impacts of their actions. Therefore, the categorisation of tools in these two categories was made based on the context and the goal of the tool.
Step 3: Comparative Analysis A comparative analysis of identified ICT tools and technologies per category was then performed. Initially, we analysed tools’ function- ality to identify core capabilities per category. Then, we examined the key features for each tool. The outcome of this analysis is a comparative table for each category
10 http://participationcompass.org/. 11 http://www.involve.org.uk. 12 http://participatedb.com/. 13 http://www.intellitics.com/. 14 http://www.reformkompass.de.
w.jager@rug.nl
7 A Comparative Analysis of Tools and Technologies for Policy Making 129
that shows, at a glance, an overview of different features found in each tool of the category.
Step 4: Conceptualisation During this step, we performed an overall discussion of the presented tools and technologies and their potential for enhancing policy making. To this end, we examined three main aspects for policy making—the type of facilitated activities, the type of targeted stakeholders, and the stages of the policy cycle. Finally, we drafted overall recommendations and conclusions.
7.3 Tools and Technologies for Policy Making
Based on the literature survey, we identified 11 main categories of ICT tools and technologies that can be used for policy making purposes as follows:
• Visualisation tools help users better understand data and provide a more meaningful view in context, especially by presenting data in a graphical form.
• Argumentation tools visualise the structure of complex argumentations and debates as a graphical network.
• eParticipation tools support the active engagement of citizens in social and political processes including, e.g. voting advice applications and deliberation tools.
• Opinion mining tools help analyse and make sense of thousands of public comments written in different application contexts.
• Simulation tools represent a real-world system or phenomenon and help users understand the system and the effects of potential actions in order to make better decisions.
• Serious games train users through simulation and virtual environments. • Tools specifically developed for policy makers have been recently developed to
facilitate the design and delivery of policies. • Persuasive tools aim to change users’ attitudes or behaviours. • Social network analysis (SNA) tools analyse social connections and identify
patterns that can be used to predict users’ behaviour. • Big data analytics tools support the entire big data exploitation process from
discovering and preparing data sources, to integration, visualisation, analysis, and prediction.
• Semantics and linked data tools enable large amounts of data to become easily published, linked to other external datasets, and analysed.
We present an analysis of each category of tools and technologies in the rest of this section.15
15 All tools mentioned in this section are summarized in the end of the chapter along with their links.
w.jager@rug.nl
130 E. Kamateri et al.
7.3.1 Visualisation Tools
Visualisation tools enable large amounts of “raw” data to become visually represented in an interpretable form. Moreover, they provide appropriate means to uncover pat- terns, relationships, and observations that would not be apparent from looking at it in a nonvisual format. Therefore, users can explore, analyse, and make sense of data that, otherwise, may be of limited value (Osimo and Mureddu 2012). Today, there are many data visualisation tools, desktop- or web-based, free or proprietary, that can be used to visualise and analyse raw data provided by the user. Examples include Google Charts, Visokio Omniscope, R, and Visualize Free. Besides visual presentation and exploration of raw data, they provide additional features such as data annotation (e.g. Visokio Omniscope), data handling, and other statistical computations on raw data (e.g. R).
Over recent years, geovisualisation (shortened form of the term geographic vi- sualisation) has gained considerable momentum within the fields of geographic information systems (GIS), cartography, and spatial statistics. Some consider it to be a branch of data visualisation (Chang 2010). However, geovisualisation inte- grates different approaches including data visualisation, such as cartography, GIS, image analysis, exploratory data analysis, and dynamic animations, to provide visual exploration, analysis, synthesis, and presentation of geospatial data (MacEachren and Kraak 2001). Geovisualisation tools have been widely used to visualise societal statistics in combination with geographic data.
Several visualisation and geovisualisation tools have been developed to visualise and analyse demographic and social statistics in several countries across the world. Most tools are used for data coming from the USA. However, many efforts have been made, lately, to visualise statistics coming from all over the world (e.g. Google Public Data Explorer and World Bank eAtlas). The most important source of information for these tools is governmental reports which are made available by each state. Most tools support data transparency, mainly for downloading data and figures, while uploading of users’ data is available only in few cases. Visualisation tools are organised into static and interactive based on a categorisation proposed for web- mapping tools (Kraak and Brown 2001). A static tool contains a figure or a map displayed as a static image (Mitchell 2005), while interactive tools allow users to access a set of functions to have some interaction with the tool or the map, such as zooming in and out (Mitchell 2005). Table 7.1 summarises well-known visualisation and geovisualisation tools and compares their main characteristics. In particular, the table provides information on: (a) the number and subject of indicators, e.g. if they deal with demographic, health, environmental, or other social issues, (b) the coverage, namely, the countries supported, (c) the period for which statistics are available, (d) data transparency, and (e) whether it is a static or interactive tool.
w.jager@rug.nl
7 A Comparative Analysis of Tools and Technologies for Policy Making 131
Table 7.1 Visualisation and geovisualisation tools for analyzing regional statistics
Indicators and Topic
Coverage (countries)
Period Data transparency
Static/ Interactive
Gapminder > 400 Demographics, social, economic, en- vironmental, health
> 200 Over the past 200 years
Download and upload
Interactive
Worldmapper ∼ 696 maps Demographics
All N/A Download (No custom maps)
Datasets, static
Dynamic Choropleth Maps
Multiple social, economic, and environ- mental
USA N/A Download (free to adjust the threshold criteria)
Interactive
DataPlace ∼ 2360 Demographics, health, arts, real estate
USA After 1990 N/A Interactive
Data Visualizer- World Bank
∼ 49 Social, economic, financial, IT, and environ- mental
209 1960–2007 N/A Interactive
World Bank eAtlas
∼ 175 Development challenges
200 After 1960 Download and upload
Interactive
State Cancer Profiles
Demographic data related to cancer
USA 2006–2010 N/A Interactive
Health Infoscape
Health conditions
USA January 2005– July 2010
N/A Interactive
OECD eXplorer
∼ 40 Demographics, economic, labour market, environment, social, and innovation
34 (335 large regions 1679 small regions)
1990–2005 Download and upload
Interactive (time animation) storytelling
Other tools investigated, but not included, in the above table include STATcompiler, Google Public Data Explorer, NComVA, Social Explorer (USA), PolicyMap (USA), All-Island Research Observatory (UK), and China Geo-Explorer II
w.jager@rug.nl
132 E. Kamateri et al.
Demographic, social, environmental, health, and other public data, provided by governmental and public authorities, in raw form, can be transformed and presented through visualisation and geovisualisation tools into a more interpretable way. Thus, information and current trends hidden in this data can easily become apparent. This can assist policy stakeholders and decision makers to make more informed decisions. In addition, incorporating geographical knowledge into planning and formation of social and political policies can help derive more accurate spatial decisions. Obvious fields where visualisation and geovisualisation tools can be applied for policy making are investment, population, housing, environmental assessment, public health, etc.
7.3.2 Argumentation Tools
Argumentation tools visualise the structure of complex arguments and debates as a graphical network. In particular, they allow a large number of stakeholders to partici- pate, discuss, and contribute creative arguments and suggestions which then become visualised. This visual representation provides a better and deeper understanding of topics discussed. Thus, complex debates can become easily analysed, refined, or evaluated, e.g. by pinpointing possible gaps and inconsistencies or strong and weak points in the arguments, etc. (Benn and Macintosh 2011).
Table 7.2 summarises well-known argumentation tools and depicts their main characteristics (i.e. whether they are open source, whether they enable import- ing/exporting data, whether they are Web-based or collaborative, the argument framework, whether they support visual representation argumentation structure mod- ification and manipulation of layouts). DebateGraph, Rationale, Cope It!, and bCisive constitute proprietary solutions, while Cohere, Araucaria, Compendium, and Carneades were developed during research studies within universities and re- search projects. Most argumentation tools enable users to share ideas and collaborate upon “wicked problems”. For example, DebateGraph allows users to collaboratively modify the structure and the content of debate maps in the same way they can collaboratively edit a wiki. In addition, MindMeister and Compendium constitute desktop-based solutions that support collaborative argument analysis, while Mind- Meister and bCisive also enable real-time collaboration. Though most argumentation tools provide, even partially, a visual representation of discussions, only few sup- port an easy layout manipulation; such tools are Compendium, Araucaria, Cohere, and DebateGraph. Besides argument analysis, argumentation tools offer additional features, such as argument reconstruction, discussion forums, argument evaluation, etc. For example, Araucaria and Argunet enable users to reconstruct and map de- bates, Cohere enables any content on the web to serve as a node of information in the argument map, and Rationale allows users to judge the strength of an argu- ment by evaluating its elements. These judgments are also represented on the map. Similarly, Carneades allows users to evaluate and compare arguments as well as to apply proof standards. Finally, Cope It! supports a threaded discussion forum, while
w.jager@rug.nl
7 A Comparative Analysis of Tools and Technologies for Policy Making 133
Table 7.2 Argumentation tools. (Source: Benn and Macintosh 2011)
Tool Open source
Import/ export
Web- based
Collab- orative
Argument frame- work
Visual represen- tation
Modify argument structure
Manipulate layouts
Araucaria Yes Yes No No Walton, Toulmin, Wigmore, Classical
Partially Yes Partially
Argunet Yes Yes Yes Yes Classical Yes Partially N/A
Carneades Yes Yes Yes No Walton Partially Yes N/A
Cohere Yes Yes Yes Yes IBIS Yes Partially Partially
Compendium Yes Yes No Yes IBIS Yes Partially Partially
Cope_it! N/A No Yes Yes IBIS Yes Partially N/A
DebateGraph No No Yes Yes Multiple (including IBIS)
Partially Partially Partially
Rationale No No No No Classical Partially Partially N/A
bCisive No No Yes Yes IBIS Partially Partially N/A
MindMeister No Yes Yes Yes N/A Yes Yes Partially
IBIS Issue-Based Information System
bCisive incorporates group planning, decision making, and team problem-solving capabilities.
Argumentation tools facilitate better-informed public debate, policy deliberation, and dialogue mapping on the web about complex political issues. For example, DebateGraph has been used by the Dutch Foreign Ministry in its recent consultation on its human rights policy16, the UK Prime Minister’s Office17, and the White House’s Open Government Brainstorming.18 Compendium has been used in a case study for consultation on regional planning in southeast Queensland (Ohl 2008). Carneades has been developed during the European Estrella project19 that aims to help both citizens and government officials to take part, more effectively, in dialogues for assessing claims and has been used in several applications.
16 http://debategraph.org/MR 17 http://debategraph.org/No10. 18 http://debategraph.org/WH. 19 http://www.estrellaproject.org/.
w.jager@rug.nl
134 E. Kamateri et al.
7.3.3 eParticipation Tools
eParticipation tools have been specifically developed to involve citizens in the policy- making process, i.e. to enable citizens to get informed, to provide feedback on different policy issues, and to get actively involved in decision making (Gramberger 2001). These tools are mainly based on Web 2.0 features including a variety of social networking tools such as discussion forums or message boards, wikis, electronic surveys or polls, e-petitions, online focus groups, and webcasting.
eParticipation may entail different types of involvement, which are supported by different tools and functionalities, ranging from the provision of information, to deliberation, community building and collaboration, active involvement through consultations, polling, and decision making. The International Association for Public Participation (IAP2) has produced a public participation spectrum20, which shows how various techniques may be employed to increase the level of public impact.
Recently, eParticipation tools have been widely used by governmental and public authorities. Through actively engaging citizens, in the planning, design, and de- livery process of public policies, they have moved towards improving democratic governance, preventing conflicts, and facilitating citizens’ active participation in the solution of issues affecting their lives. Table 7.3 presents a set of such recently developed eParticipation tools.
7.3.4 Opinion Mining Tools
The Web’s widespread use over the past decade has significantly increased the pos- sibility for users to express their opinion. The users not only can post text messages now but also can see what other users have written about the same subject in a variety of communication channels across the Web. Moreover, with the advent of Twitter and Facebook, status updates, and posts about any subject have become the new norm in social networking. This user-generated content usually contains relevant information on the general sentiment of users concerning different topics including persons, products, institutions, or even governmental policies. Thus, an invaluable, yet scattered, source of public opinion has quickly become available.
Opinion-mining tools (or otherwise called sentiment analysis tools) perform a computational study of large quantities of textual contributions in order to gather, identify, extract, and determine the attitude expressed in them. This attitude may be users’ judgment or evaluation, their affectual state (that is to say, the emotional state of the author when writing), or the intended emotional communication (that is to say, the emotional effect the author wishes to have on the reader; Stylios et al. 2010).
20 Available at: http://www.iap2.org/associations/4748/files/spectrum.pdf.
w.jager@rug.nl
7 A Comparative Analysis of Tools and Technologies for Policy Making 135
Table 7.3 eParticipation tools developed to improve people involvement in government
Typical actions Examples
Citizen Space Consultation and engagement software Create, organise, and publish public consultations across the net on complex policy documents Share consultation data openly in a structured way Provide a way to easily analyse consultation data (both qualitative and quantitative)
Used by government bodies to run e-consultations around the world
Adhocracy.de Participation and voting software Present and discuss issues Collaborate (develop and work on texts together) Make proposals, gather, and evaluate proposals Add polls for decision making Vote on issues
Used in the Munich Open Government Day where citizens could propose policies, projects, and actions of the city
MixedInk.com Collaborative writing software Large groups of people work together to write texts that express collective opinions Post ideas Combine ideas to make new versions Post comments and rate versions to bring the best ideas to the top
Used by the White House to let citizens draft collective policy recommendations for the Open Government Directive
Loomio.org Decision making and collaborative software Initiate discussions and present proposals that can then be discussed, modified, and voted (Agree, Abstain, Disagree, or Block, along with a brief explanation of why) Change their position any time
Used by the Wellington City Council for discussion with their citizens
CitySourced Mobile civil engagement platform Identify and report civic issues (graffiti, trash, potholes, etc.), and comment on existing ones
Used in San Francisco, Los Angeles, and several other cities in California
Puzzledbypolicy Consultation and opinion mapping software Learn about policy issues concerning immigration in the European Union (EU) Give their voice Graphically compare their views on immigration with national and EU immigration policies as well as with the opinions of relevant stakeholders Encourage to join discussions on particular aspects of immigration policy they feel strongly about
Used by the Athens and Torino municipalities and other stakeholders in Tenerife, Hungary, and Slovenia
w.jager@rug.nl
136 E. Kamateri et al.
Table 7.3 (continued)
Typical actions Examples
Opinion Space Opinion mapping software Collect and visualise user opinions on important issues and policies (rate five propositions on the chosen topic and type initial response to a discussion question) Show in a graphical “map” where user’s opinions fall next to the opinions of other participants Display patterns, trends, and insights Employ the wisdom of crowds to identify the most insightful ideas
Used by US State Department to engage global online audiences on a variety of foreign policy issues
CivicEvolution.org Collaboration platform Engage citizens in structured dialogue and deliberation and develop detailed community-written proposals to make constructive changes
Used by the City of Greater Geraldton, in Australia, to facilitate collaboration and deliberation among participants in participatory budgeting community panels
UbiPol Mobile civil engagement platform Identify and report problems or suggestions Report policy issues
Used by TURKSAT, a publicly owned but privately operated company in affiliation with Ministry of Transportation in Turkey
OurSpace Youth eParticipation platform Engage young people in the decision-making process Enable collaboration
European and National Youth Organisations already using OurSpace
Dialogue App Set up a dialogue Share, rate, comment, and discuss ideas and bring the best ideas to the top
Department for Environment, Food and Rural Affairs in the UK is using Dialogue App to get thoughts, ideas and input on how to improve and formulate policy
In social media, opinion mining usually refers to the extraction of sentiments from unstructured text. The recognised sentiments are classified as positive, negative, and neutral, or of a more fine-grained sentiment classification scheme. Examples include Sentiment140, Sentimentor, Repustate, etc. Opinion-mining tools may also integrate a broad area of approaches including natural language processing, computational linguistics, and text mining. Text mining, for example, can provide a deeper analysis of contributions; it summarises contributions, helps highlight areas of agreement and disagreement, and identifies participants’ main concerns—the level of support for draft proposals or suggestions for action that seem necessary to address. Opinion
w.jager@rug.nl
7 A Comparative Analysis of Tools and Technologies for Policy Making 137
mining tools providing such approaches include DiscoverText, RapidMiner, and Weka.
Classifying statements is a common problem in opinion mining, and different techniques have been used to address this problem. These techniques follow two main approaches; those based on lexical resources and neutral-language processing (lexicon-based) and those employing machine-learning algorithms. Lexicon-based approaches rely on a sentiment lexicon—a collection of known and precompiled sentiment/opinion terms. These terms are words that are commonly used to ex- press positive or negative sentiments, e.g. “excellent”, “great”, “poor”, and “bad”. The method basically counts the number of positive and negative terms, and de- cides accordingly the final sentiment. Machine-learning approaches that make use of syntactic and/or linguistic features and hybrid techniques are very common, with sentiment lexicons playing a key role in the majority of methods.
Table 7.4 presents several opinion mining tools that have been recently developed to analyse public opinions.
Opinion mining tools can help derive different inferences on quality control, pub- lic relations, reputation management, policy, strategy, etc. Therefore, opinion mining tools can be used to assist policy stakeholders and decision makers in making more in- formed decisions. In particular, knowing citizens’ opinion about public and political issues, proposed government actions, and interventions or policies under formation can ensure more socially acceptable policies and decisions. Finally, gathering and analysing public opinion can enable us to understand how a certain community re- acts to certain events and even try to discover patterns and predict their reactions to upcoming events based on their behaviour history (Maragoudakis et al. 2011).
7.3.5 Simulation Tools
Simulation tools are based on agent-based modelling. This is a recent technique that is used to model and reproduce complex systems. An agent-based system is formed by a set of interacting and autonomous “agents” (Macal and North 2005) that represent humans. Agents act and interact with their environment, including other agents, to achieve their objectives (Onggo 2010). Agents’ behaviour is described by a set of simple rules. However, agents may also influence each other, learn from their experiences, and adapt their behaviour to be better suited to their environment. Above all, they operate autonomously, meaning that they decide whether or not to perform an operation, taking into account their goals and priorities, as well as the known context. The analysis of interactions between agents results at the creation of patterns that enable visualising and understanding the system or the phenomenon under investigation.
w.jager@rug.nl
138 E. Kamateri et al.
Table 7.4 Opinion mining tools
Purpose Sources Classification
SwiftRiver Aggregate, manage, filter, and validate web data Discover relationship and trends in data
Twitter, SMS, e-mail, and RSS feeds
Machine learning
DiscoverText (Text analytics) Search, filter, collect, and classify data Generate insights
E-mail archives, social media content, and other document collections
Machine learning
Repustate Categorise and visualise social media data Extract text sentiment Predict future trends
Twitter or Facebook Multiple languages
Machine learning
Opinion observer (Opinion mining) Extract text sentiment Discover patterns
Web pages Lexicon-based (feature category)
AIRC Sentiment Analyser
Extract text sentiment N/A Lexicon-based
Social Mention Aggregate and analyse social media data Extract text sentiment Discover patterns
Blogs, comments, social media including Twitter, Facebook, Social bookmarks, microblogging services, Images, News, etc.
Lexicon-based
Umigon Sentiment analysis Twitter Lexicon-based
Convey API Sentiment analysis Social media records Machine learning Natural-language processing Statistical modelling
Sentiment140 Sentiment analysis for tweets on a subject or keyword
Twitter Machine learning Natural-language processing
Sentimentor Sentiment analysis for tweets on a subject
Twitter Machine learning
Corpora’s Applied Linguistics
Document summarisation and sentiment analysis
Documents Natural-language processing in combination with an extensive English language lexicon
Attentio Sentiment analysis Blogs, news, and discussion forum sites
Lexicon-based Machine learning
w.jager@rug.nl
7 A Comparative Analysis of Tools and Technologies for Policy Making 139
Table 7.4 (continued)
Purpose Sources Classification
Opinmind Sentiment analysis of bloggers opinion
Blogs Not available
ThinkUp Archive and analyse social media life
Twitter and Facebook Machine learning
In this sense, simulation tools are particularly suited to explore the complexity of social systems. A social system consists of a collection of individuals who interact directly or through their social environment. These individuals evolve autonomously as they are motivated by their own beliefs and personal goals, as well as the cir- cumstances of their social environment. Simulating social systems and analysing the effects of individuals’ interactions can result in the construction of social patterns (e.g. how society responds to a change) that can be used for policy analysis and planning as well as for participatory modelling (Bandini et al. 2009).
There are several general-purpose simulation tools. Most of them are open source and free to be accessed by anyone. Some of these are specially designed to focus on social systems. For example, Multi-Agent Simulation Suite (MASS) is a soft- ware package intended to enable modellers to simulate and study complex social environments. To this end, it models the individual together with its imperfections (e.g. limited cognitive or computational abilities), its idiosyncrasies, and personal interactions. Another tool focusing on the development of flexible models for living social agents is Repast.
An increasing number of tools for the simulation and analysis of social inter- actions has been developed in recent years. These aim to help policy stakeholders and decision makers to simulate the long-term impact of policy decisions. Table 7.5 presents such simulation tools that have been used in the field of health, environment, developmental policies, etc.
7.3.6 Serious Games
Agent-based modelling is used also in serious games, providing the opportunity for experiential and interactive learning and exploration of large uncertainties, divergent values, and complex situations through an engaging, active, and critical environment (Raybourn et al. 2005). Serious games enable players to learn from the accurate rep- resentations of real-world phenomena and the contextual information and knowledge and data embedded in the dynamics of the game. Abt (1987) defines serious games as games with “an explicit and carefully thought-out educational purpose and are not intended to be played primarily for amusement”.
w.jager@rug.nl
140 E. Kamateri et al.
Table 7.5 Simulation tools simulating the long-term impact of policy decisions
Purpose Input Interface Scale
Threshold 21 Simulate the long-term impact of socioeconomic development policies
About 800 variables concerning economic, social, and environment factors
Flexible Customisable to suit the needs of any sector and country
GLEaMviz Simulate global epidemics
Detailed population, mobility, and epidemic–infection data (real-world data) Compartmentalised disease models
Visual tool for designing compartmen- tal models
Thirty countries in 5 different continents
The Climate Rapid Overview and Decision- support Simulator (C-ROADS)
Simulate long-term climate impacts of policy scenarios to reduce greenhouse gas emissions (CO2 concentration, temperature, sea-level rise)
Sources of historical data
Flexible equations are available and easily auditable
Six-region and 15-region mode
UrbanSim Simulate the possible long-term effects of different policies on urban development (land use, transportation, and environmental planning)
Historical data Flexible Any country
Modelling the Early Life Course (MEL-C)
Simulate the effects of policy making in the early life course and issues concerning children and young people
Data from existing longitudinal studies to quantify the underlying determinants of progress in the early life course
Flexibly adapted for new data and parameter inputs
N/A
Global Buildings Performance Network (GBPN) Policy Comparative Tool
An interactive tool that enables users to compare the world’s best practice policies for new buildings (residential and commercial)
N/A N/A N/A
w.jager@rug.nl
7 A Comparative Analysis of Tools and Technologies for Policy Making 141
Table 7.5 (continued)
Purpose Input Interface Scale
CLASP’s Policy Analysis Modeling System (PAMS)
Forecast the impacts of energy efficiency standards and labelling programs Assess the benefits of policies, identify the most attractive targets for appliances, and efficiency levels
N/A N/A Support basic modelling inputs for over 150 countries Customisable where country- specific data is available
Scenario Modelling and Policy Assessment Tool (EUREAPA)
Model the effects of policies on environment, consumption, industry, and trade
Detailed carbon, ecological, and water footprint indicators
N/A N/A
Budget simulator
Budget consultation platform that enables to adjust budget items and see the consequences of their allocations on council tax and service areas
N/A Flexible Any country
CLASP Comprehensive, Lightweight Application Security Process, EUREAPA, MEL-C, C-ROADS
In policy making, serious games provide the opportunity for players to assume roles of real-world critical stakeholders whose decisions rely on extensive data col- lected from the world around them. In this way, players get educated on the process of decision making as well as on the limitations and trade-offs involved in policy making. Serious games may be used in fields like defence, education, scientific ex- ploration, health care, emergency management, city planning, engineering, religion, and politics (Caird-Daley et al. 2007).
Table 7.6 summarises a number of serious games aiming to tackle different social and political problems. In some of these, users assume the role of critical stakeholders. For example, in 2050 Pathways, users play as if they were the Energy and Climate Change Minister of the UK, while, in Democracy, users act as the president or the prime minister of a modern country. Other games enable users to apply policies/strategies and explore their potential impact. Such an example is the Maryland Budget Map Game that gives the option to make cost-cutting decisions and consider short-term and long-term budget effects. Serious games also help users gain virtual experiences for solving real-world problems. Thus, such games could be used to train citizens and public authorities on how to enforce a policy, e.g. a disas- ter or crisis management policy. For instance, Breakaway simulates critical incidents
w.jager@rug.nl
142 E. Kamateri et al.
Table 7.6 Serious games focusing on policy making
Purpose Features Scope
2050 Pathways Users act as the energy and climate change minister and explore the complex choices and trade-offs which the UK will have to make to reach the 80 % emission reduction targets by 2050, while matching energy demand and supply
It covers all parts of the economy and all greenhouse gas emissions Users create their emission reduction pathway, and see the impact using real scientific data
Scientific exploration and engineering
Democracy Users are in the position of president (or prime minister) of a modern country and the objective of the game is to stay in power as long as possible
It recreates a modern political system as accurately as possible Users influence the voters and the country by putting in place policies
Education, political strategy
Maryland Budget Map Game
Users act as the administration and general assembly of a state Gives the options to make cost-cutting decisions, weigh revenue options, and consider short-term and long-term budget effects
It explains how budgeting decisions are made
Education, political strategy
NationStates— create your own country
Users build a nation and run it according to their political ideals and care for people
N/A Entertainment, education
Breakaway(disaster management— incident commander)
Helps incident commanders and other public safety personnel train and plan for how they might respond to a wide range of critical incidents
It models acts of terrorism, school hostage situations, and natural disasters
Education, emergency management
The Social Simulator
Trains communications, policy, and frontline staff in a variety of sectors using a number of crisis scenarios Users use the language, tools, and norms of the social web for crisis response
It models terrorist attacks, a leaked report spreads anger about a government policy, etc.
Education, emergency management, political strategy
w.jager@rug.nl