5 PETS for privacy

1. 1Protecting Netflix ratings from other users using privacy settings can be seen as social privacy.

   1. ATrue
   2. BFalse
2. 2Using TLS to encryption ur communications with the email server can be seen as anti-surveillance privacy.

   1. ATrue
   2. BFalse
3. 3Privacy as control aims to hide as much info as possible from the provider.

   1. ATrue
   2. BFalse
4. 4A privacy technology that requires more than one entity to decrypt data follows the confidentiality paradigm.

   1. ATrue
   2. BFalse
5. 5Privacy is a security problem.

   1. ATrue
   2. BFalse
6. 6PETS = P[blank_start]rivacy[blank_end] E[blank_start]nhancing[blank_end] T[blank_start]echnologie[blank_end]s
   2 dimensions:
   1) Privacy p[blank_start]aradigm[blank_end] : how privacy is defined
   2) A[blank_start]dversarial[blank_end] models: who defines the problem & what are the goals

   1. Arivacy
   2. Bnhancing
   3. Cechnologie
   4. Daradigm
   5. Edversarial
7. 7Privacy paradigm 1: Privacy as c[blank_start]onfidentiality[blank_end]
   Privacy's definition in this paradigm:
   1) [blank_start]Minimize[blank_end] data disclosure : data [blank_start]encryption[blank_end] in transit / storage, [blank_start]homomorphic[blank_end] encryption to compute when encrypted, [blank_start]obfuscate[blank_end] (perturb, generalise)
   2) [blank_start]Distributed[blank_end] trust: secret sharing, [blank_start]multiparty[blank_end] computation
   3) Rely/ require [blank_start]open[blank_end] source

   1. Aonfidentiality
   2. BMinimize
   3. Cencryption
   4. Dhomomorphic
   5. Eobfuscate
   6. FDistributed
   7. Gmultiparty
   8. Hopen
8. 8Privacy paradigm 2: Privacy as c[blank_start]ontrol[blank_end]
   Privacy in this paradigm:
   1) User [blank_start]participation[blank_end]: let user decide how data will be shared
   > privacy [blank_start]settings[blank_end], privacy [blank_start]policy[blank_end] languages
   2) T[blank_start]ransparen[blank_end]cy and a[blank_start]ccountabilit[blank_end]y: let user know how data is used, and who is responsible
   > secure [blank_start]log[blank_end]ging
   3) Organisational c[blank_start]omplianc[blank_end]e: General data protection regulation (GDPR)

   1. Aontrol
   2. Bparticipation
   3. Csettings
   4. Dpolicy
   5. Eransparen
   6. Fccountabilit
   7. Glog
   8. Homplianc
9. 9Privacy paradigm 3: Privacy as P[blank_start]ractice[blank_end]
   Privacy definition in this paradigm & its PETS:
   1) Improve user [blank_start]agency[blank_end]: help user ne[blank_start]gotiat[blank_end]e privacy
   2) Aid [blank_start]decision[blank_end] making ad [blank_start]transparency[blank_end] of social impact: help users u[blank_start]nderstan[blank_end]d consequences of their actions
   3) Privacy as [blank_start]collective[blank_end] practice: help identify b[blank_start]es[blank_end]t practice for a group of users
   
   PETS for all 3:
   > c[blank_start]ontextu[blank_end]al feedback (privacy mirrors, preview)
   > recommenders for configuration
   > privacy [blank_start]nudge[blank_end]s

   

   1. Aractice
   2. Bagency
   3. Cdecision
   4. Dtransparency
   5. Ecollective
   6. Fnderstan
   7. Ges
   8. Hgotiat
   9. Iontextu
   10. Jnudge
10. 10Adversarial model 1: PETS for [blank_start]social[blank_end] privacy
    Concern: privacy problem is defined by [blank_start]users[blank_end]
    eg “My parents discovered I'm gay” / “My boss knows I am looking for other job”
    
    Goal: Do not [blank_start]surprise[blank_end] the user
    1) support [blank_start]decision[blank_end] making
    2) help [blank_start]identify[blank_end] actions that [blank_start]impact[blank_end]
    using: [blank_start]contextual[blank_end] feedback, easy defaults, privacy [blank_start]nudge[blank_end]s
    
    Limitation: only protects from other users, not the trusted [blank_start]service[blank_end] [blank_start]provider[blank_end]!

    1. Aservice
    2. Bprovider
    3. Ccontextual
    4. Dnudge
    5. Esocial
    6. Fusers
    7. Gsurprise
    8. Hdecision
    9. Iidentify
    10. Jimpact

11. 11Adversarial model 2 - PETS for [blank_start]institutional[blank_end] privacy
    Concerns - Privacy problem defined by [blank_start]legislation[blank_end]
    -> Data should not be collected without user [blank_start]consent[blank_end] or processed for illegitimate uses
    -> Data should be secured: co[blank_start]rrect[blank_end], in[blank_start]tegrity[blank_end], de[blank_start]letion[blank_end]
    ** different countries has different definitions of private data
    
    Goals: [blank_start]Compliance[blank_end] with data protection principles
    > [blank_start]informed[blank_end] consent
    > purpose [blank_start]limitation[blank_end]
    > subject [blank_start]access[blank_end] rights
    > data [blank_start]minimisation[blank_end]
    > Preserving the [blank_start]security[blank_end] of data: Auditability and accountability
    
    Limitations: Assumes:
    1) collection and processing by organisations is ne[blank_start]cessary[blank_end]
    > organisations are trusted & honest
    > no technical protection of data
    2) Focus on [blank_start]limitation[blank_end] of misuses, not on [blank_start]collection[blank_end]
    3) Tr[blank_start]ansparen[blank_end]cy

    

    1. Ainstitutional
    2. Blegislation
    3. Cconsent
    4. Drrect
    5. Etegrity
    6. Fletion
    7. GCompliance
    8. Hinformed
    9. Ilimitation
    10. Jaccess
    11. Kminimisation
    12. Lsecurity
    13. Mcessary
    14. Nlimitation
    15. Ocollection
    16. Pansparen
12. 12Adversarial model 3: PETS for [blank_start]anti[blank_end]-[blank_start]surveillance[blank_end] privacy
    Concerns: the privacy problem is defined by [blank_start]security[blank_end] [blank_start]expert[blank_end]s
    > The adversary is [blank_start]anybody[blank_end]
    > Concerned about censorship, freedom of speech, surveillance
    
    Goal: Minimize..
    1) [blank_start]disclosure[blank_end] of implicit / explicit personal [blank_start]inf[blank_end]o to anyone
    >> end-to-[blank_start]end[blank_end] [blank_start]encryption[blank_end]
    >> anonymity: t[blank_start]o[blank_end]r, mixnets
    >> obfuscation: [blank_start]differential[blank_end] privacy, hiding, generalization
    2) [blank_start]trust[blank_end] to others
    > [blank_start]private[blank_end] information retrieval
    > anonymous aut[blank_start]hentication[blank_end]
    > blind [blank_start]signature[blank_end]s
    > [blank_start]multiparty[blank_end] computation
    
    Limitations:
    1) Privacy-preserving designs are n[blank_start]arrow[blank_end] – difficult to create “general purpose privacy”
    > Difficult to combine
    > Difficult to ev[blank_start]olve[blank_end] - agile paradigm
    
    2) [blank_start]Usa[blank_end]bility problems both for developers and users
    > how to code this?
    > [blank_start]perfo[blank_end]rmance down
    
    3) Lack of [blank_start]incenti[blank_end]ves:
    > industry will lose the data
    > Governments: national security, fraud detection...

    1. Aanti
    2. Bsurveillance
    3. Csecurity
    4. Dexpert
    5. Eanybody
    6. Fdisclosure
    7. Ginf
    8. Hend
    9. Iencryption
    10. Jo
    11. Kdifferential
    12. Ltrust
    13. Mprivate
    14. Nhentication
    15. Osignature
    16. Pmultiparty
    17. Qarrow
    18. Rolve
    19. SUsa
    20. Tperfo
    21. Uincenti
13. 13The application gathers the recommendations from the students and then: lets other students see each other recommendations, and lets the bars see the student recommendations so that they can offer discounts to students that give good ratings.
    
    Adversaries: [blank_start]other[blank_end] [blank_start]student[blank_end]s, [blank_start]owner[blank_end]s, [blank_start]cloud[blank_end] provider
    They can see: [blank_start]every[blank_end]thing

    1. Acloud
    2. Bowner
    3. Cother
    4. Dstudent
    5. Eevery
14. 14The application gathers the recommendations from the students and then: lets other students and the restaurant owners see the average rating for a restaurant.
    
    Adversaries: [blank_start]other[blank_end] [blank_start]student[blank_end]s, [blank_start]owner[blank_end]s, [blank_start]cloud[blank_end] provider
    They can see: [blank_start]student[blank_end]s and cloud [blank_start]owner[blank_end]s cannot learn students behavior anymore; The Cloud [blank_start]provider[blank_end] still can see everything and learn student behavior and preferences

    1. Aother
    2. Bstudent
    3. Cowner
    4. Dcloud
    5. Estudent
    6. Fowner
    7. Gprovider
15. 15What kind of privacy for each:
    > You want to protect the students social network (who is friends with whom) from the students they do not know
    PET for [blank_start]social[blank_end] privacy (access control) => Privacy as [blank_start]control[blank_end]
    
    > You do not want the bar owners to learn which bar each student has visited, only the aggregates
    PET for [blank_start]social[blank_end] privacy (access control) => Privacy as [blank_start]control[blank_end]
    
    > You do not want the cloud provider to learn what students connect
    [blank_start]Anti-surveillance[blank_end] PET for [blank_start]social[blank_end] privacy (anonymous communications) => Privacy as [blank_start]confidentiality[blank_end]
    
    > You want the students to be able to evaluate how much other application users know about them
    PET for [blank_start]social[blank_end] privacy (privacy mirrors) => Privacy as [blank_start]practice[blank_end]
    

    1. Asocial
    2. Bcontrol
    3. Ccontrol
    4. Dsocial
    5. EAnti-surveillance
    6. Fsocial
    7. Gconfidentiality
    8. Hsocial
    9. Ipractice
16. 16Technical goals of anti-surveillance PETs: [blank_start]pseudonymity[blank_end]
    eg. persistent random id, hashed id, emails, pseudo-id, quasi-id
    

    

    1. Apseudonymity
17. 17Technical goals of anti-surveillance PETs: [blank_start]anonymity[blank_end]
    > decouple [blank_start]action[blank_end] from [blank_start]identity[blank_end]!
    > hiding: who is...
    ....the reader of a web page, the person accessing a service
    ...the sender of an email, the writer of a text
    ...the person to whom an entry in a database relates
    ...the person present in a physical location

    

    1. Aaction
    2. Bidentity
    3. Canonymity
18. 18Technical goals of anti-surveillance PETs: [blank_start]unlinkability[blank_end]
    > decouple [blank_start]2 actions[blank_end] from [blank_start]1 user[blank_end]!
    > hiding: Two...
    ... anonymous letters written by the same person
    ... web page visits by the same user
    ... entries in a databases related to the same person
    ... two people related by a friendship link
    ... same person spotted in two locations

    1. Aunlinkability
    2. B2 actions
    3. C1 user
19. 19Technical goals of anti-surveillance PETs: [blank_start]unobservability[blank_end]
    > decouple [blank_start]observation[blank_end] from [blank_start]action existence[blank_end]!
    > hiding:
    ...whether someone is accessing a web page
    ...whether a message is being sent
    ...whether an entry in a database corresponds to a real person
    ...whether someone or no one is in a given location

    1. Aunobservability
    2. Bobservation
    3. Caction existence
20. 20Technical goals of anti-surveillance PETs: [blank_start]plausible[blank_end] [blank_start]deniability[blank_end]
    > decouple [blank_start]observation[blank_end] from [blank_start]true action[blank_end]!
    > Not possible to prove ...
    ... that a person has hidden information in a computer
    ... that someone has the combination of a safe
    ... that a person has been in a place at a certain point in time
    ... that a database record belongs to a person

    1. Aplausible
    2. Bdeniability
    3. Ctrue action
    4. Dobservation
21. 21However, ______ is breakable using a probabilistic analysis

    1. Aconfidentiality
    2. Bpseudonimity
    3. Canonimity
    4. Dunlinkability
    5. Eunobservability
    6. Fplausible deniability
22. 22Systematic privacy evaluation:

    

    1. Aidentify
    2. Baction A
    3. Caction B
    4. Dreal action
    5. Efake
    6. Freal
    7. Gobfuscated
23. 23Systematic privacy evaluation
    1) Model the privacy-preserving mechanism as a [blank_start]probabilis[blank_end]tic transformation
    > What is the probability that, given an input the privacy mechanism returns a given output
    2) Determine what the [blank_start]adversary[blank_end] will see
    > Threat model: who is the adversary? what is the “ob[blank_start]servation[blank_end]”? what is her p[blank_start]rior[blank_end] knowledge?
    3) “Invert” the mechanism as the adversary would do
    > Always assume the adversary knows the me[blank_start]chanism[blank_end] and would try to undo its effect
    4) Evaluate property after in[blank_start]version[blank_end]
    > This is the real pr[blank_start]obability[blank_end] the adversary can compute
    5) Measure
    > Not trivial!

    1. Aprobabilis
    2. Badversary
    3. Cservation
    4. Drior
    5. Echanism
    6. Fversion
    7. Gobability
24. 24Measuring privacy:
    > [blank_start]Correctness[blank_end]: given the distribution, what is the probability that the adversary does not guess correctly?
    > [blank_start]Uncertainty[blank_end]: given the distribution, what is the uncertainty of the adversary on the answer?
    > [blank_start]Accuracy[blank_end]: given the distribution, and a guess, how certain is the adversary of his inference?
    > [blank_start]Differential[blank_end] [blank_start]privacy[blank_end]-based: given the observation what is the bound on the information the adversary can gain

    1. ACorrectness
    2. BUncertainty
    3. CAccuracy
    4. DDifferential
    5. Eprivacy
25. 25What privacy properties are being achieved?
    A. A user accesses a web three days in a row, and the web cannot recognize it is the same user

    1. AAnonimity
    2. BUnlinkability
    3. CUnobservability
    4. DPlausible deniability
    5. EPseudonymity
26. 26What privacy properties are being achieved?
    B. A user has a program in its computer that publishes Tweets automatically. Twitter cannot prove that a given Tweet was made by the user

    1. AAnonymity
    2. BUnlinkability
    3. CUnobservability
    4. DPlausible deniability
    5. EPseudonimity
27. 27 what privacy properties are being achieved?
    C. A user uses her university account for accessing university services and her Gmail account to access Netflix.

    1. AUnlikability
    2. BUnobservability
    3. CPseudonimity
    4. DAnonymity
    5. EPlausible deniability
28. 28what privacy properties are being achieved?
    A user accesses a newspaper website using a proxy, the newspaper can only see the proxy
    IP address if she is the only user behind the proxy.

    1. APseudonymity
    2. BAnonymity
    3. CUnlinkability
    4. DUnobservability
    5. EPlausible deniability
29. 29what privacy properties are being achieved?
    D. A user accesses a newspaper website using a proxy, the newspaper can only see the proxy
    IP address if there are several users behind the proxy.

    1. APseudonymity
    2. BPlausible deniability
    3. CUnlinkability
    4. DUnobservability
    5. EAnonymity
30. 30Which of the 2 system provides better unlinkability between a student and a comment?
    A. a system that always mixes the feedback from two students before presenting it to the professor (i.e., comment from student Bob appears sent by student Alice and vice versa)
    B. a system that with 50% of the time forwards the comment of a student and 50% of the time forwards a fake comment
    

    1. AA
    2. BB
31. 31PETS for data anonymization:
    Produce a dataset that preserves the utility of the original dataset without leaking information about individuals

    

    1. Aquasi-identifiers
    2. Blink
    3. Canonymity
32. 32What property/properties Alice enjoys
    - if she uses one-time credentials to users login in a website (imagine a random number that can only be shown once, like a ticket to the cinema)
    - with respect to the website? [blank_start]Anonymity[blank_end], [blank_start]Unlinkability[blank_end]
    - with respect to the credential issuer? If the credential is obtained in person: [blank_start]none[blank_end]. If the credential is
    obtained online, [blank_start]pseudonymity[blank_end].
    - regarding email receivers, if every time she wants to send an email she ask one of our friends to send it for us
    - if each friend sends at most one email - [blank_start]Anonymity[blank_end]
    - if friends can send more than one email - [blank_start]Pseudonymity[blank_end]

    1. AAnonymity
    2. BUnlinkability
    3. Cnone
    4. Dpseudonymity
    5. EAnonymity
    6. FPseudonymity

Description

- Information security (5 PETS for privacy) Quiz on 5 PETS for privacy, created by Yawen Hou on 10/06/2019.

Author

Yawen Hou

Language

English

Views

3

Tags

information security5 pets for privacy-

Rating

No ratings yet

Rate this quiz

Rate this quiz

More on Privacy Risks and Safeguards