Census0004.dvi

DIGITAL GOVERNMENT PROJECT
Progress Report: April 24, 2000
The Research Team
NISS: Alan Karr, Ashish Sanil, Jaeyong Lee [, James
CMU: Adrian Dobro, George Duncan, Stephen
LANL: Sallie Keller–McNulty
MCNC: Bonnie Parrish, Karen Litwin, Syam Sun-
Build a Web-based query system that
1. Is dynamic and history-dependent 2. Dispenses statistical analyses rather than 3. Uses statistical technology to preserve con- Implement the system on “live” Federal agency
Understand how the system is used and performs
Evaluate disclosure risk models and risk reduction
strategies at realistic scales, using the systemas testbed Summary of Progress to Date
• Algorithms for geographic (or other) aggregation (Sanil, • Statistical implications of aggregation (Lee, Sanil, Karr)• Prototype table server design (Karr, Sanil, Hilden–Minton)• QHDB schema for table server (Sanil, Karr, Hilden–Minton)• NASS prototype under construction (Karr, Lee, Sanil, • Scalability of methods to compute bounds (Fienberg, • Bayesian framework for confidentiality protection (Dun- • Confidentiality Reading Group, involving NISS, RTI, other • Interactions with other DG projects (Columbia, UNC) Table Server Prototype
Data: Sample Census data set with
• 8 (after trimming) categorical variables: Age, Education, Employer type, Marital status, Query: Sub-table of full 8-way table
Response: Requested sub-table (FTP, character dis-
play, visualization) or statement that it cannot Problem Conceptualization
New Release Movement of Frontier
Risk Criteria
• Predictive capability for sensitive variable • Accuracy of IPF reconstruction of full table • [Accuracy of LP bounds on cell entries] System Design
• Visualization as a means of risk reduction • Visual interfaces incorporating association Fienberg/Dobro
Progress: Formal results on bounds for tables and their rela-
tionship to log-linear model and graphical structures.
New theorems for the "decomposable case" and ex-tensions that reduce the bounding problem to smallerdimensional components.
With Duncan, exploration of formal structures requiredto weight the tradeoff between disclosure risk and so-cietal gains from data release, using a formal Bayesianinformation theoretic approach.
Current Challenges: Scaling up the results so that they are
computationally feasible for actual government sur-vey settings.
Products: Papers (PNAS); code to be incorporated in table
Duncan/Keller–McNulty
Progress: Initial steps toward formal Bayesian decision–the-
oretic framework for confidentiality protection throughdisclosure limitation. The framework explicitly incorpo-rates disclosure risk and data utility. It also permits thecomparison of disclosure limitation through matrix mask-ing and generation of synthetic data.
With Fienberg, exploration of formal structures requiredto weight the tradeoff between disclosure risk and so-cietal gains from data release, using a formal Bayesianinformation theoretic approach.
Current Challenges: Formally analyze the impact on dis-
closure risk and data utility of data swapping. Bet-ter understand synthetic data as a disclosure limitationtool. Develop associated procedures for disclosure riskestimation and disclosure limitation that scale.
Products: New algorithms. Review paper on confidential-
ity and disclosure limitation, to be published in the In-ternational Encyclopedia of the Social and BehavioralSciences (Duncan).
The Next Six Months
• Complete NASS prototype; write associated • Functional table server prototype with dynamic risk estimation and visualizations. Major scal- ability questions will remain.
• Initial concepts of query, risk, response for re- • [Initial consideration of longitudinal data]

Source: http://nisla05.niss.org/dg/presentations/census0004.pdf

optfunction.com

SECTION A: HISTORY 8. Does exposure to perfumes, insecticides,fabric shop odors and other chemicals provoke. For each “yes” answer in Section A, circle Optimum Function: the point score for that question . Total Dysbiosis Questionnaire your score and record it in the box at theend of the section. Then move to sections9. Are your symptoms worse on damp, muggyThis questionnair

gsite.univ-provence.fr

Copyright 2008 by the American Psychological AssociationNative Language Influences on Word Recognition in a Second Language:Kristin Lemho¨fer, Ton Dijkstra, Herbert Schriefers,Centre National de la Recherche ScientifiqueMany studies have reported that word recognition in a second language (L2) is affected by the nativelanguage (L1). However, little is known about the role of the specific lang

Copyright © 2010-2018 Pharmacy Drugs Pdf