The ARGUS Programming of the SDC-venture.


95 views
Uploaded on:
Category: Funny / Jokes
Description
The ARGUS Programming of the SDC-venture Anco Hundepool Measurements Netherlands Washington, August 1999 Factual Exposure Control the harmony between the requirement for (to an ever increasing extent) data and the security of the respondents Factual Revelation Control
Transcripts
Slide 1

The ARGUS Software of the SDC-venture Anco Hundepool Statistics Netherlands Washington, August 1999

Slide 2

Statistical Disclosure Control the harmony between the requirement for (an increasing amount) data and the respondents\' security

Slide 3

Statistical Disclosure Control Need for nitty gritty smaller scale information documents electronic productions processing force of clients Need for more point by point tables But....!!!

Slide 4

Statistical Disclosure Control Protection of security of respondents persons, undertakings, establishments Respondents must have the capacity to trust Statistical Offices! Dangers: Intruders/programmers Accidental acknowledgment Advanced record linkage systems

Slide 5

Statistical Disclosure Control Produce ‘safe’ datafiles and tables Apply information change methods Preserve as much data Implemented in ARGUS!

Slide 6

Framework of advancement of ARGUS SDC extend incompletely sponsored by EU (fourth Framework) Co-operation between The Netherlands, Italy (+Spain) and UK

Slide 7

General points of SDC venture Methodological exploration in SDC microdata, tables concerning insights, OR geological information (general) SDC Software improvement microdata ( m - ARGUS) tables ( t - ARGUS)

Slide 8

SDC venture individuals Netherlands CBS (ARGUS) TU-Eindhoven (OR for microdata) Italy Istat (with Univ. of Rome) (Research/testing) CPR-Padova (with Univ. Tenerife) (OR for even information)

Slide 9

SDC venture individuals UK ONS (information) Univ. Manchester (with Univ. of Southampton) (Research on SARs) Univ. Of Leeds (Geographical information)

Slide 10

Main programming created in SDC-venture m - ARGUS (CBS and TUE) smaller scale information t - ARGUS (CBS and CPR) plain information

Slide 11

Ideas of m - ARGUS Intruder uses data of distinguishing variables (e.g. district, sex, age, instruction, occupation) to distinguish records. This prompts the delicate data

Slide 12

m - ARGUS Levels of security open utilization documents (PUF) smaller scale records for specialists (MUC) colleges, contract and so forth safe-setting

Slide 13

Ideas of m - ARGUS a rundown of blends of distinguishing variables must checked discover worth mixes that are perilous e.g. |a x b x c| <= limit edge relies on upon level of assurance Public utilization documents Micro information for specialists (contract)

Slide 14

Ideas of m - ARGUS dispose of the perilous blends by worldwide recoding (age - > agegroup, area - > region) neighborhood concealment (attributing missings) intuitively/consequently with least data misfortune (entropy)

Slide 15

m - ARGUS For microdata Developed in Borland C++ Windows-95/98 Version 3.0 last SDC-adaptation intelligent/programmed worldwide recoding programmed nearby concealment

Slide 16

Features of m - ARGUS can deal with huge microdata records just tables got from microdata are being utilized adaptable worldwide recoding alternatives for programmed blend of worldwide recoding and nearby concealment (TU Eindhoven)

Slide 17

Addit. elements of m - ARGUS Micro-collection Top/Bottom coding Rounding

Slide 18

m - ARGUS metadata microdata Generate tables Recoding plans Global recoding Local concealment Micro conglomeration Top/base coding Rounding Report metadata microdata

Slide 19

m - ARGUS info information Data: Fixed configuration ASCII Metadata Name Position Missing qualities (2) Identification level Hierarchical coding Codelist (pick.)

Slide 20

Using m - ARGUS perusing information document producing tables apply worldwide recodes nearby concealment create safe record create report

Slide 21

t - ARGUS

Slide 22

Ideas of t - ARGUS recognizable proof of delicate cells utilizing e.g. predominance principle in any event n (e.g. 2) supporters to a phone whole of biggest 3 givers >= 75% (one vast patron could recalculate the commitment of its rival) simple part

Slide 23

Ideas of t - ARGUS Eliminate/secure touchy cells (crucial step) by applying SDC procedures table update cell concealment adjusting intelligently and/or consequently with least data misfortune (e.g. cell weights)

Slide 24

Ideas of t - ARGUS cell concealment in tables with marginals distinguish essential delicate cells ensure essential cells by stifling extra (auxiliary) cells to avoid recalculation (to some estimate) with negligible data misfortune (CPR)

Slide 25

t - ARGUS 3-D tables intelligent table upgrade essential & optional cell concealment enhancement schedules for programmed cell concealment adjusting

Slide 26

t - ARGUS metadata microdata classification codelists overhaul adjusting concealment report Safe table

Slide 27

Features of t - ARGUS Initial gone through microdata Determine likewise best k per cell - >sensitive cells Table update conceivable without backpedaling to microdata Uses methods for optional cell concealment utilizing condition of-the advancement calculations (CPR) Prepared for connected tables

Slide 28

t - ARGUS Data: settled arrangement ASCII Meta information: Variable name Start. position Field length Status

Slide 29

t-ARGUS Apply worldwide recoding Protect document with optional concealment Rounding Safe table as ASCII or .WK1 (in addition to report)

Slide 30

t - ARGUS Version 2.0 last SDC-adaptation obliges business OR-solver (Xpress by Dash, UK, 600 GBP)

Slide 31

Future/CASC Computational Aspects of Statistical Confidentiality New European venture proposition (2000-2002) Extending ARGUS New research Additional joint USA/EU-venture?

Slide 32

CASC-m Concentration on business/financial information microaggregation PRAM Noise-expansion/veiling

Slide 33

CASC-t Hierarchical tables Linked tables Optimal arrangement vz. heuristics Different info positions

Slide 34

CASC-group Statistics Netherlands Istat (Italy) ONS, Univ. Southampton, Manchester, London, Plymouth (UK) Bundesambt, IAB (Germany) Stat. Catalunya, Univ Tenerife (Spain)

Slide 35

Contact Anco Hundepool Statistics Netherlands PO box 4000 2200 JM Voorburg The Netherlands email ahnl@krypton.vb.cbs.nl fax: +31 70 3375990 telephone: +31 70 3375038

Slide 36

.:tslid

Recommended
View more...