Seman ic Me ada a Schema o Risks and Mi iga ions Associa ed wi h
Ou pu s om T us ed Resea ch En i onmen s
T up i Padiya
1
, Jim Smi h, Felix Ri chie, Elizabe h G een,
Uni e si y o he Wes o England
2
1. Execu i e Summa y
The documen p oposes a seman ically s uc u ed me ada a schema adhe ing o W3C
s anda ds, o ep esen isks and mi iga ions associa ed wi h s a is ical ou pu s, o he
p ocess o s a is ical disclosu e con ol. The seman ic me ada a schema ( om now
e e ed o as “schema”) is designed o add ess he ollowing co e objec i es:
• P o ide a modula , scalable design ha can adap o changing needs and be
easily in eg a ed in o au oma ed p ocessing sys ems.
• Enable p incipled compa isons be ween manual p ocesses and he unc ionali y
p o ided by di e en e sions o au oma ed so wa e.
• Suppo ex ensibili y and se e as a basis o suppo ing complex ou pu
s uc u es such as AI models.
Table o Con en s
1. Execu i e Summa y ........................................................................................... 1
2. Pu pose and Scope ........................................................................................... 2
3. Modelling he S a ba n Taxonomy ....................................................................... 5
4. Schema o di e en s a ba ns ........................................................................... 7
5. Key Bene i s o he Schema .............................................................................. 20
6. Conclusion ..................................................................................................... 20
7. Re e ences ..................................................................................................... 20
8. Appendix: Classes and p ope ies in he Seman ic Me ada a Schema o S a is ical
Risks and Mi iga ions.............................................................................................. 21
1
Co esponding au ho . T up [email protected]
2
The au ho s acknowledge wi h g a i ude aluable insigh s and cons uc i e eedback om Amy Tilb ook,
Ben De ick and Paul Whi e.
2. Pu pose and Scope
This documen p oposes a seman ically s uc u ed me ada a schema adhe ing o W3C
s anda ds, o ep esen isks and mi iga ions associa ed wi h esea ch ou pu s, o he
p ocess o s a is ical disclosu e con ol. The seman ic me ada a schema ( om now
e e ed o as “schema”) is designed o add ess he ollowing co e objec i es:
• P o ide a modula , scalable design ha can adap o changing needs and be
easily in eg a ed in o au oma ed p ocessing sys ems.
• Enable p incipled compa isons be ween manual p ocesses and he
unc ionali y p o ided by di e en e sions o au oma ed so wa e.
• Suppo ex ensibili y and se e as a basis o suppo ing complex ou pu
s uc u es such as AI models.
To c ea e his schema we build on wo complemen a y de elopmen s: he s a ba n
(G een, Ri chie and Whi e 2024) (Ri chie, e al. 2023), and he Da a P i acy Vocabula y
(DPV) (J. Pandi , e al. 2024) and i s Risk Ex ension (Es e es, Golpayegani, e al., Da a
P i acy Vocabula y 2025).
We also make an impo an dis inc ion be ween (i) conduc ing isk assessmen s, and (ii)
making disclosu e con ol decisions. This dis inc ion allows us o delibe a ely side-s ep
wo issues:
• The sensi i i y o di e en da a se s: Risks may esul om a gi en o m o
analysis, e en i he na u e o he da a means hey ha e no meaning ul impac .
• Rules s. P inciples-based ou pu checking: Rega dless o whe he a TRE will
allow ‘excep ions’, disclosu e con ol decisions should be in o med by a
ho ough isk assessmen .
2.1. Backg ound: The S a ba n F amewo k
The S a ba ns axonomy (G een, Ri chie and Whi e 2024) (Ri chie, e al. 2023)
… is a amewo k o classi y all s a is ical e ms by hei disclosu e
cha ac e is ics, including isk, excep ions and mi iga ion measu es. This s a ba n
massi ely educes he dimensionali y o he disclosu e checking p oblem, as
well as p o iding imp o ed cla i y. I also c ea es a easible basis o au oma ic
disclosu e con ol checking (G een e al 2024)
The axonomy g oups di e en ypes o analyses in o 14 di e en ‘S a ba ns’ acco ding
o he na u e o he ou pu s hey p oduce and he a ious associa ed disclosu e isks. I
is consis en wi h exis ing bes p ac ice as highligh ed in, o example, he Secu e Da a
Access P o essionals handbook (G i i hs e al 2019). Impo an ly, i aims o p o ides a
ou e o Na ional S a is ics Ins i u es (NSIs), and a wide ange o TREs o ind
consensus on manual p ac ice. Howe e , many NSIs, and inc easingly TREs make use
o (semi) au oma ed ools as pa o hei OSDC p ac ice, such as Tau A gus, Table -
builde , Da aSHIELD o SACRO. The goal o ensu ing consis ency c ea es a demand o
a se o unambiguous, human- eadable and machine ac ionable s a emen s desc ibing
bes p ac ice.
2.2. Backg ound: Da a P i acy Vocabula y
The schema uses a ich seman ic ep esen a ion o model he S a ba n Taxonomy
adhe ing o he Wo ld Wide Web Conso ium (W3C) s anda ds. I includes co e
concep s o s a ba n (e.g., s a is ical e ms, isks, mi iga ions) o ganised as class
hie a chies ex ended om he Da a P i acy Vocabula y (DPV) (J. Pandi , e al. 2024) and
i s Risk Ex ension (Es e es, Golpayegani, e al., Da a P i acy Vocabula y 2025). DPV is
an es ablished, in e na ionally ecognised ocabula y and on ology, which ep esen
concep s abou use and p ocessing o da a, hei associa ed isks and impac s, and can
aid easoning ele an o da a p i acy. The DPV Risk ex ension consis s o concep s
ela ed o in o ma ion associa ed wi h isks. The schema mainly builds on wo key
concep s:
• dp :Risk ep esen s he concep o ' isk' i.e. a possibili y o po en ial o nega i e
e en s o occu and is indica ed using he ela ion dp :hasRisk.
• dp :RiskMi iga ionMeasu e ep esen s a me hod o p ocess o con ol o mi iga e
he isk, and is associa ed using he ela ion dp :isMi iga edByMeasu e
and hese a e accompanied by a ich se o ela ionships, cons ain s, and a ibu es
such as likelihoods, and measu es, acili a ing machines o unde s and and eason
abou isks and mi iga ions associa ed wi h p ocesses.
The schema se es as a key componen o a ligh weigh on ology using OWL/RDF o
enable s uc u ed ep esen a ion and easoning abou p i acy- ela ed disclosu e isks
and mi iga ion measu es wi hou needing o e ly complex logical s uc u es.
In addi ion, SHACL will be used o speci y some me a-da a (e.g. he ca dinali y o
ela ionships) and alida e knowledge-g aphs ( ep esen ing se s o analyses) and
en o ce ules. Fo example, we use SHACL o speci y ha o a gi en da ase he e is
exac ly one `minimumTh eshold’ o he numbe o esponden s in a cell.
Ou p oposed schema c ea es a new class S a ba n, and le e ages exis ing DPV
p ope ies, and SDC-speci ic ex ensions o classes/concep s o c ea e a seman ic
model o OSDC. The concep ual o e iew is illus a ed in Figu e 1.
Figu e 1 Concep ual O e iew o Risks and Mi iga ions o OSDC
The majo classes/concep s ha a e ex ended om DPV a e dp :Risk, dp :Likelihood,
dp :RiskMi iga ionMeasu e, and dp :O ganisa ionalMeasu e. Figu e 1 shows likelihood
wi h 5 likelihood le els e.g. e y high, high, mode a e, low and e y low. I is possible o
add likelihood le els a he scale o 3 e.g. high, mode a e, low o a he scale o 7 e.g.
ex emely high, e y high, high, mode a e, low, e y low, ex emely low.
The p ope ies o he DPV Vocabula y eused in his schema a e dp :hasRisk,
dp :hasRiskAssessmen , dp :hasLikelihood, dp :hasO ganisa ionalMeasu e, and
dp :isMi iga edByMeasu e. Class le el es ic ions a e applied o he p ope ies ha a e
eused om DPV, o emphasis hei seman ic con ex o he domain o SDC. The
schema is easie o unde s and, in eg a e, and ex end.
3. Modelling he S a ba n Taxonomy
“S a ba n” se es as he base class in he schema. The s a ba n amewo k includes
14 s a is ical e ms, and hey a e modelled as subclasses o he base class – S a ba n,
as shown in Figu e 2. S a ba n has ollowing subclasses: 1) F equencies, 2)
S a is icalHypo hesisTes , 3) Posi ion, 4) Shape, 5) Linea Agg ega ions, 6) Mode, 7)
EndPoin s, 8) Nonlinea Concen a ionRa ios, 9) Calcula edRa ios, 10)
Haza dSu i alTables, 11) GiniCoe icien ,12) LinkedMul ile elTables, 13) Clus e s, and
14) Co ela ionCoe icien s.
Figu e 2 S a ba n Hie a chy
3.1. Ex ending dp :Risk
The dp :Risk class is ex ended o ep esen se e al ypes o isks associa ed wi h
di e en s a ba ns. The hie a chy o he isks is depic ed in Figu e 3.
Figu e 3 Types o Risks in SDC
3.2. Ex ending dp :RiskMi iga ionMeasu e
The dp :RiskMi iga ionMeasu e is ex ended o ep esen isk mi iga ion measu es
associa ed wi h di e en isks. Figu e 4 shows he hie a chy o isk mi iga ion measu es
used o he s a ba n. Clea ly his is no comple e ( o example, algo i hms o achie ing
Di e en ial P i acy could be a sub-class o ‘Ta ge ed Noise’). Howe e , as ou aim is o
p o ide a unique namespace o his schema, i can simply be ex ended o encompass
mo e mi iga ions.
3.3. Ex ending dp :RiskAssessmen
The dp :RiskAssessmen is ex ended o ep esen a ious checks associa ed wi h
di e en s a ba ns as shown in Figu e 5. These checks o m he basis o isk
assessmen o ou pu disclosu e con ol. To make he schema gene ally applicable
and o sepa a e dp :RiskAssessmen om TRE and da ase -speci ic choices, hey ha e
associa ed pa ame e s.
Figu e 4 Types o Risk Mi iga ion Measu es in SDC
Figu e 5 Types o Risk Assessmen s in SDC
3.4. Ex ending dp :O ganisa ionalMeasu e
The dp :O ganisa ionalMeasu e is ex ended o ep esen a ious o ganisa ional
measu es - ypically nume ical o Boolean alues - p o ided by TREs. These a e used o
pa ame e ise isk assessmen checks- o example, he size o he smalles g oup
allowed (Minimum Th eshold), o whe he Class Disclosu e is an issue
(Requi edZe ocheck) – and may a y be ween TREs and da ase s. They ep esen i al
aspec s o he TREs (o hei da a owne s) isk appe i e. These a e shown in Figu e 6.
Figu e 6 O ganisa ional measu es om TREs o isk assessmen s.
Appendix 1 p o ides addi ional de ails abou all he classes and p ope ies de ined and
used in he schema. The s a ba ns: Linked mul ile el ables and Clus e s needs
e inemen and hence a e no ep esen ed in his documen .
4. Schema o di e en s a ba ns
The nex sec ions p esen isualisa ions o all he S a ba n schemas, s a ing wi h a
wo ked example o how he mos common – ‘ equencies’ is c ea ed.
4.1. S a ba n: F equencies
Figu e 7 p esen s he schema o he s a ba n “F equencies” de ailing associa ed isks
and mi iga ions. Figu e 8 shows all he subclass hie a chy o F equencies: Allu ialFlow,
C ossTab, F equencyTable, His og am, and so on. These subclasses inhe i all he isk
associa ions, isk likelihood and isk mi iga ion measu es.
These igu es a e g aph models ep esen ing he schema and p o ides ounda ion o a
knowledge g aph, bu his is less amenable o human inspec ion. We ha e colou -
coded i o make i mo e human- eadable:
• The s a ba n ( ype o que y) is in yellow
• Po en ial isks a e in o ange
• Mi iga ions a e in g een
• Associa ed wi h each isk is a check (pu ple) wi h associa ed pa ame e s (blue).
• The isk likelihood (beige) is calcula ed h ough o combina ion o (i) check esul s,
and (ii) whe he a mi iga ion has been applied.
Figu e 7 S a ba n: F equencies
Figu e 7 desc ibes ha e e y ins ance o F equencies is associa ed wi h speci ic ypes
o p i acy isks: LowCoun s, Di e encing and ClassDisclosu e. In u n, each isk
ep esen s de ails abou hei isk assessmen s and mi iga ions. In his case i can be
seen ( isually) o in e ed (p og amma ically) ha Supp ession and Noise a e alid
mi iga ions o all h ee isks.
Di ing deepe , LowCoun is a isk associa ed wi h F equencies, and mus be checked
o he minimum h eshold and i i ails he check, he e is a highe likelihood o isk,
and he mi iga ion measu es (e.g. supp ession) mus be applied. The idea is cap u ed by
he seman ic model:
• Class LowCoun is associa ed o class MinimumTh esholdCheck using
dp :hasRiskAssessmen .
• MinimumTh esholdCheck u he links o he class MinimumTh eshold, which holds
a li e al alue p o ided by he TRE.
• The class LowCoun is also connec ed o he dp :Likelihood using
dp :hasLikelihood o ep esen he likelihood o he isk.
• The isk LowCoun is mi iga ed by Noise, Rounding, o Supp ession ia
dp :isMi iga ionMeasu e
The o he wo isks p esen a e modelled in he same way.
No e ha we use he o ganisa ional measu e Requi edZe oCheck o p o ide a
mechanism o TREs o s a e ha class disclosu e is no an issue o hei da a.
The g aphical ep esen a ion o he schema sugges s ha he isk posed by LowCoun in
equencies da a can be 1) ecognized, 2) assessed, and 3) mi iga ed ia known SDC
echniques. Simila ly, o he isks like ClassDisclosu e and Di e encing, hei likelihood,
and hei mi iga ion measu es pe aining o he F equencies s a ba n can be in e p e ed
om Figu e 7.
Figu e 8 Subclass hie a chy o F equencies
4.2. S a ba n: Posi ion
Figu e 9 ep esen s he schema o he s a ba n “Posi ion”. Figu e 10 shows all he
subclass hie a chy o Posi ion: Qua ile, Box plo and so on.
Figu e 9 S a ba n: Posi ion
Figu e 10 Subclass hie a chy o Posi ion
4.9. S a ba n: Haza dSu i al Tables
Figu e 22 ep esen s he schema o he s a ba n “Haza d Su i al Tables”. Figu e 23
shows all he subclasses o Haza dSu i alTables: Kaplan_Mie e, Haza dTables and
Su i alTables.
Figu e 22 S a ba n: Haza d Su i al Tables
Figu e 23 Subclass hie a chy o Haza d Su i al Tables
4.10. S a ba n: Gini Coe icien
Figu e 24 ep esen s he schema o he s a ba n “Gini Coe icien ”. Figu e 25 shows
GiniCoe icien has a subclass GiniCu es.
Figu e 24 S a ba n: Gini Coe icien
Figu e 25 Subclass hie a chy o Gini Coe icien
4.11. S a ba n: Co ela ion Coe icien s
Figu e 26 ep esen s he schema o he s a ba n “Co ela ion Coe icien s”. Table 1
p esen s di e se ypes o Co ela ion Coe icien s, each is ep esen ed as a subclass o
he Co ela ionCoe icien s class.
Table 1 Types o Co ela ion Coe icien
ANCOVA
Mul inomial logi
Bina y Logis ic Reg ession
Mul iple eg ession
Canonical Co ela ion
Mul i a ia e analysis o a iance
Con as Coe icien s
Odds Ra ios (es ima ed)
Co ela ion Coe icien s
Omnibus es s o model coe icien s
Co a ia es
Panel da a models
Gene al linea model
Pa ial co ela ion coe icien s
Kendall’s ank
Phi coe icien
Ke nel Es ima es
P obi
Linea Reg ession Coe icien s
S anda dised eg ession coe icien s
Logis ic eg ession
S uc u al equa ion modelling
Logi
Th ee-s age leas squa es
Log-linea o highe o de ables
Two-s age leas squa es
Longi udinal es ima ion
Two-way analysis o a iance
MANCOVA
Ze o-o de co ela ion
Figu e 26 S a ba n: Co ela ion Coe icien s
4.12. S a ba n: S a is ical Hypo hesis Tes
Figu e 27 ep esen s he schema o he s a ba n “S a is ical Hypo hesis Tes ”. Table 2
p esen s di e en ypes o S a is ical Hypo hesis Tes s and each is ep esen ed as a
subclass o he S a is icalHypo hesisTes .
Table 2 Types o S a is ical Hypo hesis Tes
Adjus ed R Squa ed
Eigen alues – Sc ee plo s
Mauchly's sphe ici y es
Analysis o a iance
E a Squa ed
McNema 's es
ANOVA
F iedman es
Nagelke ke R Squa ed
Ba le ’s Tes o Sphe ici y
Homogenei y o Reg ession
Pai ed - es s
Box’s Tes o Equali y o
Co a iance Ma ices
Homogenei y o Va iance
Pa allel analysis
Chi-Squa ed es
Homogenei y o Va iance-
Co a iance ma ices
Pa ial e a squa ed
Coch an’s Q es
Hosme -Lemeshow es
Pea son’s p oduc -momen
co ela ion coe icien
Coe icien o de e mina ion
Independen - es s
Pea son’s
Cohen’s d
Kaise –Meye –Olkin es
P incipal componen analysis
Cohen’s kappa coe icien
Kappa Measu e o
Ag eemen
Pseudo-R-squa ed
Con idence in e als
Kolmogo o -Smi no es
R Squa ed
Cox & Snell R Squa ed
K uskal-Wallis es
Spea man’s ank co ela ion
coe icien
C ame ’s V
Lambda
Tukey’s hones y signi ican
di e ence es
Disc iminan Func ion
Analysis
Le ene’s es
Wilcoxon signed ank es
Disc iminan Validi y
Mann-Whi ney U es
Wilks’s Lambda
Figu e 27 S a ba n: S a is ical Hypo hesis Tes s
5. Key Bene i s o he Schema
• Fo malise S a is ical Ou pu s, P i acy Risks, and mi iga ions:
The schema p o ides a s uc u ed, machine- eadable amewo k o he
s a ba n axonomy ha seman ically models isks ela ed o s a is ical ou pu s
and ecommends sui able mi iga ions adhe ing o W3C s anda ds.
• Facili a es Au oma ed Reasoning:
The schema acili a es au oma ed easoning o de e mine which isks apply o
gi en s a is ical ou pu s because he concep s and ela ionships a e explici ly
de ined using seman ics and connec ed o s anda dised ocabula ies e.g. he
Da a P i acy Vocabula y (DPV).
• Ensu es Concep ual Consis ency and In e ope abili y:
The schema ex ends he DPV amewo k o he domain o S a is ical Disclosu e
Con ol, s anda dised de ini ions, and p omo es in e ope abili y.
6. Conclusion
The schema p o ides a o mal amewo k o easoning abou s a is ical ou pu s in
e ms o hei p i acy isks. I can acili a e au oma ed ools o in e which isks apply
and guides h ough applicable isk mi iga ions. I c ea es consis ency in how s a is ical
disclosu e con ol concep s a e desc ibed and linked wi h DPV concep s. This
me ada a schema o he s a ba n se e as a ounda ion o s a is ical disclosu e
con ol and can suppo complex ou pu s uc u es such as AI models.
7. Re e ences
Es e es, Bea iz, Dela am Golpayegani, Geo g P. K og, Ha sh a dhan J. Pandi , Julian
Flake , and Paul Ryan . 2025. Da a P i acy Vocabula y. 16 Ma ch.
h ps://w3c.gi hub.io/dp /2.1/dp /.
G een, Elizabe h, Felix Ri chie, and Paul Whi e. 2024. “The s a ba n: A New Model o
Ou pu S a is ical Disclosu e Con ol.” In e na ional Con e ence on P i acy in
S a is ical Da abases. Sp inge Na u e Swi ze land. 284-293.
G i i hs, E,.G eci, G., Ko o sios, Y., Pa ke , S., Sco , J., Welp on, R., Wol e s A., and
Woods, C. (2019) Handbook on S a is ical Disclosu e Con ol o Ou pu s Secu e
Da a Access P o essionals h ps://secu eda ag oup.o g/wp-
con en /uploads/2019/10/sdc-handbook- 1.0.pd
J. Pandi , Ha sh a dhan, Bea iz Es e es, Geo g P. K og, Paul Ryan, Dela am
Golpayegani, and Julian Flake. 2024. “Da a P i acy Vocabula y (DPV) – Ve sion
2.0.” In In e na ional Seman ic Web Con e ence. Sp inge Na u e Swi ze land.
171-193.
Ri chie, Felix, Elizabe h G een, Jim Smi h, Amy Tilb ook, and Paul Whi e. 2023. “The
SACRO guide o s a is ical ou pu checking (Ve sion 1).”
doi:h ps://zenodo.o g/ eco ds/10282526.
8. Appendix: Classes and p ope ies in he Seman ic
Me ada a Schema o S a is ical Risks and Mi iga ions
Table 3 Classes in he Seman ic Me ada a Schema o S a is ical Risks and Mi iga ions
Classes
Supe class
Desc ip ion
S a ba n
Owl:Thing
I is he base class o he axonomy
ep esen ing domain-speci ic
concep ela ed o s a is ical
disclosu e con ol classi ica ion
amewo k.
LowCoun
dp :Risk
Rep esen s a disclosu e isk whe e
a ibu es can be in e ed om he
da a.
ClassDisclosu e
dp :Risk
Rep esen s disclosu e isk whe e
disclosu e happens a a
class/ca ego y le el.
Di e encing
dp :Risk
Rep esen ing disclosu e isk om
compa ing linked ables o ind
di e ences.
LowDOF
dp :Risk
Rep esen s disclosu e isk
associa ed wi h less a iabili y in
da a.
Dominance
dp :Risk
Rep esen s a disclosu e isk ha
a ises when a small numbe o
con ibu o s accoun o a la ge
p opo ion o a cell’s o al alue.
Auxilia yIn o
dp :Risk
Rep esen s a disclosu e isk when
ex e nal o backg ound knowledge
can be combined wi h eleased
da a o e-iden i y indi iduals o
in e con iden ial a ibu es.
Implici Tables
dp :Risk
Rep esen s a disclosu e isk when
sensi i e in o ma ion can be
in e ed indi ec ly by compa ing
ela ed ables o sub ables.
Noise
dp :RiskMi i
ga ionMeasu
e
Rep esen s echnique o mi iga e
disclosu e isks ha in ol es
adding andom a ia ion o da a o
mask indi idual alues.
Rounding
dp :RiskMi i
ga ionMeasu
e
Rep esen s echnique o mi iga e
disclosu e isks by modi ying
nume ical alues (e.g., coun s o
o als) o he nea es speci ied
base.
Supp ession
dp :RiskMi i
ga ionMeasu
e
Rep esen s echnique o mi iga e
disclosu e isks by emo ing
speci ic da a alues wi h high
disclosu e isk.
Ou lie Remo al
dp :RiskMi i
ga ionMeasu
e
Rep esen s echnique o mi iga e
disclosu e isks by excluding
ex eme o unique da a poin s.
Agg ega ion
dp :RiskMi i
ga ionMeasu
e
Rep esen s echnique o mi iga e
disclosu e isks by combining
indi idual da a poin s in o b oade
ca ego ies o g oups.
MinimumTh esholdCheck
dp :RiskAss
essmen
Rep esen s a check equi ed o
Minimum h eshold.
P esenceO LinkedTablesCheck
dp :RiskAss
essmen
Rep esen s a check equi ed o
p esence o linked ables.
P sesenceO Ze osCheck
dp :RiskAss
essmen
Rep esen s a check equi ed o
p esence o ze os.
Requi edZe oCheck
dp :RiskAss
essmen
Rep esen s a check equi ed o
assessing he isk o class
disclosu e whe he he TRE
speci ies P esence o Ze os o be
checked o no . Some o ganiza ions
iew ze o alues as a disclosu e isk
equi ing a Ze oCheck, while o he s
do no , so i is impo an o assess
whe he a Ze oCheck is equi ed in
each con ex .
PQCheck
dp :RiskAss
essmen
Rep esen s a check ha e alua es
s a is ical ou pu s using a PQ Tes .
NKCheck
dp :RiskAss
essmen
Rep esen s a check ha applies he
NK Tes .
MinimumDOFCheck
dp :RiskAss
essmen
Rep esen s a check ha e alua es
whe he he deg ees o eedom in
s a is ical ou pu s mee he
p ede ined MinimumDOF
h eshold.
S a ba nDa aCheck
dp :RiskAss
essmen
Rep esen s a check ha e i ies
whe he he da a displayed in a
S a ba n is ele an and app op ia e
o elease.
MinimumTh eshold
dp :O ganis
a ionMeasu
e
Rep esen s an o ganisa ion
measu e o minimum h eshold
designed o educe disclosu e isk
by se ing a minimum coun o
alue ha da a cells mus mee o
be published. I is associa ed wi h a
decimal alue.
P esenceO Ze o
dp :O ganis
a ionMeasu
e
Rep esen s an o ganisa ion
measu e ha lags o moni o s
whe he ze o alues occu in da a
cells, as he p esence o ze os can
inc ease disclosu e isk, I is
associa ed wi h a Boolean alue.
P esenceO LinkedTables
dp :O ganis
a ionMeasu
e
Rep esen s an o ganisa ion
measu e ha iden i ies o moni o s
when mul iple ela ed ables o
da ase s a e linked, as such
connec ions can inc ease
disclosu e isk. I is associa ed wi h
a Boolean alue.
Ze oCheck
dp :O ganis
a ionMeasu
e
Rep esen s an o ganisa ion
measu e om TREs i p esence o
ze o is checked and is associa ed
wi h a Boolean alue.
PRa io
dp :O ganis
a ionMeasu
e
Rep esen s an o ganisa ion
measu e ha e alua es disclosu e
isk using a s a is ical es , whe e a
p- alue indica es whe he he da a
mee accep able h esholds o
sa e elease. I is associa ed wi h
decimal alue.
NValue
dp :O ganis
a ionMeasu
e
Rep esen s an o ganisa ion
measu e used in he NK es o
ep esen he minimum numbe o
con ibu o s equi ed in a cell o
conside i non-disclosi e.
KValue
dp :O ganis
a ionMeasu
e
KValue is an o ganisa ional
measu e used in he NK es o
speci y he maximum allowable
con ibu ions.
MinimumDOF
dp :O ganis
a ionMeasu
e
Rep esen s an o ganisa ional
measu e ha se s he minimum
accep able deg ees o eedom o
s a is ical ou pu s.
S a ba nRele an ShowDa a
dp :O ganis
a ionMeasu
e
Rep esen s an o ganisa ional
measu e ha ensu es only
necessa y and ele an da a is
displayed.
F equencies
S a ba n
F equencies is one o he s a ba n.
F equency able, His og am,
PieCha , Sca e G aph,
Hea map, LineG aph, and so
on.
F equencies
Subclasses o F equencies.
S a is icalHypo hesisTes s
S a ba n
S a is ical hypo hesis es s is one o
he s a ba n.
Posi ion
S a ba n
Posi ion is one o he s a ba n.
Qua ile, Box plo and so on.
Posi ion
Subclasses o Posi ion.
Shape
S a ba n
Shape is one o he s a ba n.
Ku osis, S anda dDe ia ion,
and so on.
Shape
Subclasses o Shape.
Linea Agg ega ions
S a ba n
Linea Agg ega ions is one o he
s a ba n.
Mean, Sum, Ba G aph, and so
on
Linea
Agg ega ions
Subclasses o Linea Agg ega ions.
Mode
S a ba n
Mode is one o he s a ba n.
Clus e s
S a ba n
Clus e s is one o he s a ba n.
EndPoin s
S a ba n
End Poin s is one o he s a ba n.
Minimum, Maximum, and
Range
EndPoin s
Subclasses o EndPoin s.
NonLinea Concen a ionRa ios
S a ba n
Non-linea Concen a ion Ra ios is
one o he s a ba n.
He indhalHi schmanIndex
NonLinea C
oncen a ion
Ra ios
Subclass o
NonLinea Concen a ionRa ios.
Calcula edRa ios
S a ba n
Calcula ed Ra ios is one o he
s a ba n.
Calcula edAnalysis, OddRa ios,
and RiskRa ios
Calcula edR
a ios
Subclasses o Calcula edRa ios.
Co ela ionCoe icien s
S a ba n
Co ela ion Coe icien s is one o
he s a ba n.
Haza dSu i alTables
S a ba n
Haza d Su i al Tables is one o he
s a ba n.
Kaplan_Mie e, Haza dTables
and Su i alTables
Haza dSu i
alTables
Subclasses o
Haza dSu i alTables.
GiniCoe icien
S a ba n
Gini Coe icien is one o he
s a ba n.
GiniCu es
Ginicoe icie
n
Subclass o Ginicoe icien .
LinkedMul ile elTables
S a ba n
Linked Mul ile el Tables is one o
he s a ba n.
Table 4 P ope ies in he Seman ic Me ada a Schema o S a is ical Risks and Mi iga ions
P ope y
Type
Desc ip ion
dp :hasRisk
Objec
P ope y
Links S a ba n o Risks. Class-le el
es ic ions on dp :hasRisk a e applied. Fo
example, he class F equencies is de ined
wi h es ic ions on he p ope y
dp :hasRisk such ha i mus be
associa ed wi h a leas one o hese isk
classes: ClassDisclosu e, Di e encing, o
LowCoun .
dp :hasRiskAssessmen
Objec
P ope y
Links Risks o Risk Assessmen s o assess
i . Class-le el es ic ions on
dp :hasRiskAssessmen a e applied. Fo
example, he class LowCoun is de ined
wi h es ic ions on he p ope y
dp :hasRiskAssessmen such ha i mus
be associa ed wi h a leas one ins ance o
he class MinimumTh esholdCheck.
dp :hasO ganisa ionMeasu e
Objec
P ope y
Links Risk Assessmen s o O ganisa ion
Measu es. Class-le el es ic ions a e
applied o speci y he ype o measu e
ele an in each isk assessmen . Eg.
MinimumTh esholdCheck mus include an
o ganisa ional measu e o ype
MinimumTh eshold.
dp :isMi iga edbyMeasu e
Objec
P ope y
Links isks o hei mi iga ion measu es.
class-le el es ic ions on
dp :isMi iga edByMeasu e a e applied. Fo
example, o he isk class LowCoun , i
has ei he o he mi iga ion measu es o
apply: Supp ession, Noise o Rounding
dp :hasLikelihood
Objec
P ope y
Links Risk concep wi h Likelihood and Risk
Assessmen wi h Likelihood adhe ing o
he seman ics o DPV. Necessa y class-
le el es ic ions a e applied, and i is
con ex ualised wi hou ede ining global
domain and ange.
dp :hasRiskLe el
Objec
P ope y
Links isks o i s isk le els. Same as DPV.
hasDecimalValue
Da a
P ope y
Used in es ic ions o speci y ixed ypes o
p ope y alues (e.g. h esholds)
