Anonymisation

Question 1

What is the definition of privacy according to Westin and Kasem-Madani?

Answer 1

The right of individuals to protect their personal lives and matters from the outside world and to determine which information about themselves should be known to others.

Question 2

What is the difference between personal data and anonymous data according to the GDPR?

Answer 2

Personal data is any information relating to an identified or identifiable natural person, while anonymous data is data rendered anonymous in such a manner that the data subject is not or no longer identifiable.

Question 3

What are the three types of attributes that can be used to identify or link individuals in a data set?

Answer 3

(Direct) identifiers, quasi-identifiers, and sensitive attributes.

Question 4

What are the three types of privacy threats that can occur when releasing a data set?

Answer 4

Membership disclosure, attribute disclosure, and identity disclosure.

Question 5

What is the difference between a semi-honest and a malicious attacker in the context of MPC?

Answer 5

A semi-honest attacker follows the protocol but tries to learn additional information, while a malicious attacker can arbitrarily deviate from the protocol.

Question 6

What is 1-out-of-2 oblivious transfer and why is it important for MPC?

Answer 6

1-out-of-2 oblivious transfer is a protocol that allows a sender to send one of two messages to a receiver, who can choose which message to receive without the sender knowing which one was chosen. It is important for MPC because it enables private inputs to be used in computations.

Question 7

What is a garbled circuit and how is it used in MPC?

Answer 7

A garbled circuit is a cryptographic protocol that allows two parties to jointly compute a function over their inputs while keeping those inputs private. It is used in MPC as a way to compute functions without revealing any information other than the output.

Question 8

What is the point and permute technique in the context of garbled circuits?

Answer 8

The point and permute technique is a method used to reduce the computation and communication cost of garbled circuits. It involves adding a sorting bit to each wire label and sorting the garbled table by the sorting bits, so that the evaluator can find the correct row to decrypt faster.

Question 9

What is the garbled row reduction technique in the context of garbled circuits?

Answer 9

The garbled row reduction technique is a method used to reduce the size of garbled circuits. It involves choosing the output label in the top row of the garbled table to be a zero-bitstring, so that the top row does not need to be transmitted to the evaluator, saving bandwidth.

Question 10

What is the free XOR technique in the context of garbled circuits?

Answer 10

The free XOR technique is a method used to reduce the computation and communication cost of garbled circuits. It involves altering the generation of wire labels such that XOR gates can be evaluated without a garbled table, saving computation and communication.

Question 11

What is the half gates technique in the context of garbled circuits?

Answer 11

The half gates technique is a method used to reduce the computation and communication cost of garbled circuits. It involves representing AND gates with two half gates, where each party knows one input each, and reducing the number of ciphertexts in each half gate to one, saving encryption and decryption costs.

Question 12

What is the Fairplay library and what is it used for?

Answer 12

The Fairplay library is a Java implementation of garbled circuits without performance improvements. It is used to define and compile circuits, and to execute them with different OT variants.

Question 13

What is the Bristol fashion format and what is it used for?

Answer 13

The Bristol fashion format is another format for representing garbled circuits. It is used to define and compile circuits, and to execute them with different OT variants.

Question 14

What is the JIGG library and what is it used for?

Answer 14

The JIGG library is another Java implementation of garbled circuits with performance improvements. It is used to run circuits with a server, a garbler, and an evaluator.

Question 15

What is the difference between the Fairplay library and the JIGG library?

Answer 15

The main difference between the Fairplay library and the JIGG library is that the Fairplay library does not include performance improvements for garbled circuits, while the JIGG library does.

Question 16

What is the difference between the SHDL format and the Bristol fashion format?

Answer 16

The main difference between the SHDL format and the Bristol fashion format is that the SHDL format is used in the Fairplay library, while the Bristol fashion format is used in the JIGG library.

Question 17

What is the difference between k-anonymity, l-diversity, and t-closeness?

Answer 17

K-anonymity requires that every unique combination of values of quasi-identifiers must occur for at least k individuals in a released data set. L-diversity requires that each equivalence class must have at least l ‘well-represented’ values for each sensitive attribute. T-closeness requires that the distributions of sensitive attributes in each equivalence class and the whole table differ by at most t.

Question 18

What is the unsorted matching attack and how can it be prevented?

Answer 18

The unsorted matching attack is a privacy breach that exploits the order of the entries in published data sets and links different published k-anonymous versions of the original table, destroying the k-anonymity. It can be prevented by mixing the entries of a table before publishment.

Question 19

What is the complementary release attack and how can it be prevented?

Answer 19

The complementary release attack is a privacy breach that exploits the fact that every published table may reveal additional information, even when fulfilling k-anonymity, and that the linked (combined) table may destroy the k-anonymity. It can be prevented by considering all attributes of a published table as a quasi-identifier or base publishments on the same table.

Question 20

What is the difference between perturbative methods, non-perturbative methods, and synthetic data in the context of anonymization?

Answer 20

Perturbative methods alter the attribute values in a way that the new data set may contain erroneous information. Non-perturbative methods replace attribute values with less specific than incorrect values. Synthetic data replace attribute values by artificially created values, which may be based on a model of the real data.

Question 21

What is the definition of IT security?

Answer 21

IT security is the field that deals with the protection of information and systems from unwanted access, changes, or destruction.

Question 22

What is the definition of personal data according to the GDPR?

Answer 22

Personal data is any information relating to an identified or identifiable natural person, such as name, address, identity number, etc.

Question 23

What is the definition of anonymization according to the GDPR?

Answer 23

Anonymization is the process of changing personal data in such a way that the affected person is not or no longer identifiable. Anonymized data is no longer subject to the General Data Protection Regulation (GDPR).

Question 24

What is secure multi-party computation (MPC)?

Answer 24

Secure multi-party computation (MPC) is a technique that allows multiple parties to compute a common function with secret inputs, without learning private information from other parties.

Question 25

What are garbled circuits?

Answer 25

Garbled circuits are a method of MPC that models the function as a boolean circuit and encrypts the output lines of the circuit elements with random labels.

Question 26

What is the difference between a semi-honest and a malicious attacker in the context of MPC?

Answer 26

A semi-honest attacker follows the protocol but tries to learn additional information, while a malicious attacker can arbitrarily deviate from the protocol.

Question 27

What is 1-out-of-2 oblivious transfer?

Answer 27

1-out-of-2 oblivious transfer is a protocol that allows a sender to send one of two messages to a receiver, who can choose which message to receive without the sender knowing which one was chosen.

Question 28

What is the point and permute technique in the context of garbled circuits?

Answer 28

The point and permute technique is a method used to reduce the computation and communication cost of garbled circuits. It involves adding a sorting bit to each wire label and sorting the garbled table by the sorting bits, so that the evaluator can find the correct row to decrypt faster.

Question 29

What is the garbled row reduction technique in the context of garbled circuits?

Answer 29

The garbled row reduction technique is a method used to reduce the size of garbled circuits. It involves choosing the output label in the top row of the garbled table to be a zero-bitstring, so that the top row does not need to be transmitted to the evaluator, saving bandwidth.

Question 30

What is the free XOR technique in the context of garbled circuits?

Answer 30

The free XOR technique is a method used to reduce the computation and communication cost of garbled circuits. It involves altering the generation of wire labels such that XOR gates can be evaluated without a garbled table, saving computation and communication.

Question 31

What is the half gates technique in the context of garbled circuits?

Answer 31

The half gates technique is a method used to reduce the computation and communication cost of garbled circuits. It involves representing AND gates with two half gates, where each party knows one input each, and reducing the number of ciphertexts in each half gate to one, saving encryption and decryption costs.

Question 32

What is the Fairplay library and what is it used for?

Answer 32

The Fairplay library is a Java implementation of garbled circuits without performance improvements. It is used to define and compile circuits, and to execute them with different OT variants.

Question 33

What is the Bristol fashion format and what is it used for?

Answer 33

The Bristol fashion format is another format for representing garbled circuits. It is used to define and compile circuits, and to execute them with different OT variants.

Question 34

What is the JIGG library and what is it used for?

Answer 34

The JIGG library is another Java implementation of garbled circuits with performance improvements. It is used to run circuits with a server, a garbler, and an evaluator.

Question 35

What is the difference between the Fairplay library and the JIGG library?

Answer 35

The main difference between the Fairplay library and the JIGG library is that the Fairplay library does not include performance improvements for garbled circuits, while the JIGG library does.

Question 36

What is the difference between the SHDL format and the Bristol fashion format?

Answer 36

The main difference between the SHDL format and the Bristol fashion format is that the SHDL format is used in the Fairplay library, while the Bristol fashion format is used in the JIGG library.

Question 37

What is the difference between k-anonymity, l-diversity, and t-closeness?

Answer 37

K-anonymity requires that every unique combination of values of quasi-identifiers must occur for at least k individuals in a released data set. L-diversity requires that each equivalence class must have at least l ‘well-represented’ values for each sensitive attribute. T-closeness requires that the distributions of sensitive attributes in each equivalence class and the whole table differ by at most t.

Question 38

What is the unsorted matching attack and how can it be prevented?

Answer 38

The unsorted matching attack is a privacy breach that exploits the order of the entries in published data sets and links different published k-anonymous versions of the original table, destroying the k-anonymity. It can be prevented by mixing the entries of a table before publishment.

Question 39

What is the complementary release attack and how can it be prevented?

Answer 39

The complementary release attack is a privacy breach that exploits the fact that every published table may reveal additional information, even when fulfilling k-anonymity, and that the linked (combined) table may destroy the k-anonymity. It can be prevented by considering all attributes of a published table as a quasi-identifier or base publishments on the same table.

Question 40

What is the difference between perturbative methods, non-perturbative methods, and synthetic data in the context of anonymization?

Answer 40

Perturbative methods alter the attribute values in a way that the new data set may contain erroneous information. Non-perturbative methods replace attribute values with less specific than incorrect values. Synthetic data replace attribute values by artificially created values, which may be based on a model of the real data.