## Enhancing privacy through negative representations of data (2004)

Citations: | 12 - 9 self |

### BibTeX

@TECHREPORT{Esponda04enhancingprivacy,

author = {Fernando Esponda and Stephanie Forrest and Paul Helman},

title = {Enhancing privacy through negative representations of data},

institution = {},

year = {2004}

}

### Years of Citing Articles

### OpenURL

### Abstract

The paper introduces the concept of a negative database, in which a set of records DB is represented by its complement set. That is, all the records not in DB are represented, and DB itself is not explicitly stored. After introducing the concept, several results are given regarding the feasibility of such a scheme and its potential for enhancing privacy. It is shown that a database consisting of n, l-bit records can be represented negatively using only O(ln) records. It is also shown that membership queries for DB can be processed against the negative representation in time no worse than linear in its size and that reconstructing the database DB represented by a negative database NDB given as input is an NP-hard problem when time complexity is measured as a function of the size of NDB.

### Citations

1174 |
Probabilistic encryption
- Goldwasser, Micali
- 1984
(Show Context)
Citation Context ...ally has a unique representation. Representing data negatively, as described here, permits a message to be encoded in several ways and one is chosen randomly (an idea used in probabilistic encryption =-=[21, 6]-=-). In privacy-preserving data mining, the goal is to protect the confidentiality of individual data while still supporting certain data-mining operations, for example, the computation of aggregate sta... |

784 |
Applied Cryptography: Protocols, Algorithms, and Source Code
- Schneier
- 1996
(Show Context)
Citation Context ... privacy-preserving data-mining, query restriction and negative data. An obvious starting point for protecting sensitive data is the large body of work on cryptographic methods, e.g., as described in =-=[27]-=-. Some researchers have investigated how to combine cryptographic methods with databases [18, 17, 5], for example, by encrypting each record with its own key. Cryptography, however, is intended to con... |

502 | Cryptography and Data Security
- Denning
- 1982
(Show Context)
Citation Context ...ving data mining, the goal is to protect the confidentiality of individual data while still supporting certain data-mining operations, for example, the computation of aggregate statistical properties =-=[3, 2, 1, 12, 14, 29, 28]-=-. In one example of this approach (ref. [3]), relevant statistical distributions are preserved, but the detals of individual records are obscured. Our method is almost the reverse of this approach, in... |

319 | On the Design and Quantification of Privacy Preserving Data mining algorithms
- Agrawal, Aggarwal
- 2002
(Show Context)
Citation Context ...ving data mining, the goal is to protect the confidentiality of individual data while still supporting certain data-mining operations, for example, the computation of aggregate statistical properties =-=[3, 2, 1, 12, 14, 29, 28]-=-. In one example of this approach (ref. [3]), relevant statistical distributions are preserved, but the detals of individual records are obscured. Our method is almost the reverse of this approach, in... |

315 | Security-control methods for statistical databases: a comparative study
- Adam, Worthmann
- 1989
(Show Context)
Citation Context ...ving data mining, the goal is to protect the confidentiality of individual data while still supporting certain data-mining operations, for example, the computation of aggregate statistical properties =-=[3, 2, 1, 12, 14, 29, 28]-=-. In one example of this approach (ref. [3]), relevant statistical distributions are preserved, but the detals of individual records are obscured. Our method is almost the reverse of this approach, in... |

313 | Universal One-way Hash Functions and Their Applications. [STOC
- Naor, Yung
- 1989
(Show Context)
Citation Context ...tive databases there is a large body of literature that addresses this issue due to its isomorphism with the satisfability (SAT) problem [24, 11]. Of particular relevance, then, are one-way functions =-=[20, 25]-=-—functions that are easy to compute but hard to reverse and one-way accumulators [4, 9] which are essentially one-way hash functions with the property of being commutative. One key distinction between... |

168 | Dynamic accumulators and application to efficient revocation of anonymous credentials
- Camenisch, Lysyanskaya
(Show Context)
Citation Context ...somorphism with the satisfability (SAT) problem [24, 11]. Of particular relevance, then, are one-way functions [20, 25]—functions that are easy to compute but hard to reverse and one-way accumulators =-=[4, 9]-=- which are essentially one-way hash functions with the property of being commutative. One key distinction between these existing methods and the negative database is that the output of a one-way funct... |

147 | Hiding informations and signatures in trapdoor knapsacks
- Merkle, Hellman
- 1978
(Show Context)
Citation Context ... in which we want to support some queries efficiently but not reveal the entire database. Cryptosytems founded on N P-complete problems [16] have been explored such as the Merkle-Hellman cryptosystem =-=[23]-=-, which is based on the general knapsack problem. These systems rely on a series of tricks 8Randomize c-clause algorithm Let wi denote a i bit prefix and Wi a set of i length patterns. 1. i ← ⌈log2(l... |

143 | Logic and Databases: a Deductive Approach
- Gallaire, Minker, et al.
- 1984
(Show Context)
Citation Context ...full privileges from inspecting individual records to retrieve information. The term “negative data” sounds similar to our method, but is actually quite different. The deductive database model (e.g., =-=[19]-=- presents an excellent survey of the foundations of the model) supports in the intensional database (IDB) the negative representation of data. The objectives, mechanisms, and consequences 9here are q... |

114 | One-way accumulators: a decentralized alternative to digital signatures
- Benaloh, Mare
- 1993
(Show Context)
Citation Context ...somorphism with the satisfability (SAT) problem [24, 11]. Of particular relevance, then, are one-way functions [20, 25]—functions that are easy to compute but hard to reverse and one-way accumulators =-=[4, 9]-=- which are essentially one-way hash functions with the property of being commutative. One key distinction between these existing methods and the negative database is that the output of a one-way funct... |

113 | Finding Hard Instances of the Satisfiability Problem: A Survey
- Cook, Mitchell
- 1997
(Show Context)
Citation Context ...t indeed produce hard to reverse instances. In the case of negative databases there is a large body of literature that addresses this issue due to its isomorphism with the satisfability (SAT) problem =-=[24, 11]-=-. Of particular relevance, then, are one-way functions [20, 25]—functions that are easy to compute but hard to reverse and one-way accumulators [4, 9] which are essentially one-way hash functions with... |

112 |
An efficient probabilistic public-key encryption scheme which hides all partial information
- Blum, Goldwasser
- 1985
(Show Context)
Citation Context ...ally has a unique representation. Representing data negatively, as described here, permits a message to be encoded in several ways and one is chosen randomly (an idea used in probabilistic encryption =-=[21, 6]-=-). In privacy-preserving data mining, the goal is to protect the confidentiality of individual data while still supporting certain data-mining operations, for example, the computation of aggregate sta... |

82 | Secure databases: protection against user influence
- Dobkin, Jones, et al.
- 1979
(Show Context)
Citation Context |

49 | The rise and fall of knapsack cryptosystems
- Odlyzko
(Show Context)
Citation Context ...nd its capable of producing every possible c-clause. to concel the existence of a “trapdoor” that permits retrieving the hidden information. However, almost all knapsack cyptosystems have been broken =-=[26]-=-, and it has been shown [7, 8] that in general if breaking a cryptosystem is N P-hard then N P=co-N P, a point addressed in Section 4. If a scheme based on a N P-hard result, such as the one proposed ... |

41 | A formal framework for positive and negative detection schemes,” July 2002, draft version. 170
- Esponda, Forrest, et al.
(Show Context)
Citation Context ...he database fragments. We would like to study this property in more detail. Finally, we are interested in inexact representations. The NDB representation is closely related to partial match detection =-=[15]-=- which has many applications in anomaly detection. We are interested in studying how those methods might be combined with NDB either for designing an adaptive query mechanism or for approximate databa... |

38 |
A modified random perturbation method for database security
- Tendick, Matloff
- 1994
(Show Context)
Citation Context |

27 |
Inference controls for statistical databases
- Denning, Schlorer
- 1983
(Show Context)
Citation Context ... obscured. Our method is almost the reverse of this approach, in that we support efficient membership queries but higher-level queries may be expensive. Our method is also related to query restriction=-=[22, 10, 12, 13, 28]-=-, where the query language is designed to support only the desired classes of queries. Although query restriction controls access to the data by outside users, it cannot protect an insider with full p... |

25 |
A note on the complexity of cryptography
- Brassard
- 1979
(Show Context)
Citation Context ...every possible c-clause. to concel the existence of a “trapdoor” that permits retrieving the hidden information. However, almost all knapsack cyptosystems have been broken [26], and it has been shown =-=[7, 8]-=- that in general if breaking a cryptosystem is N P-hard then N P=co-N P, a point addressed in Section 4. If a scheme based on a N P-hard result, such as the one proposed here, is to be used in a priva... |

21 |
A Database Encryption Scheme which Allows the Computation of Statistics using Encrypted Data
- Blakley, Meadows
(Show Context)
Citation Context ...nt for protecting sensitive data is the large body of work on cryptographic methods, e.g., as described in [27]. Some researchers have investigated how to combine cryptographic methods with databases =-=[18, 17, 5]-=-, for example, by encrypting each record with its own key. Cryptography, however, is intended to conceal all information about the encrypted data, and it is therefore not conducive to situations in wh... |

21 |
Security problems on inference control for sum, max, and min queries
- Chin
- 1986
(Show Context)
Citation Context ... obscured. Our method is almost the reverse of this approach, in that we support efficient membership queries but higher-level queries may be expensive. Our method is also related to query restriction=-=[22, 10, 12, 13, 28]-=-, where the query language is designed to support only the desired classes of queries. Although query restriction controls access to the data by outside users, it cannot protect an insider with full p... |

18 | Cryptographic Protection of Databases and Software," Distributed Computing and Cryptography
- Feigenbaum, Liverman, et al.
- 1991
(Show Context)
Citation Context ...nt for protecting sensitive data is the large body of work on cryptographic methods, e.g., as described in [27]. Some researchers have investigated how to combine cryptographic methods with databases =-=[18, 17, 5]-=-, for example, by encrypting each record with its own key. Cryptography, however, is intended to conceal all information about the encrypted data, and it is therefore not conducive to situations in wh... |

15 | Cryptographic Protection of Membership Lists
- Feigenbaum, Liberman, et al.
- 1992
(Show Context)
Citation Context ...nt for protecting sensitive data is the large body of work on cryptographic methods, e.g., as described in [27]. Some researchers have investigated how to combine cryptographic methods with databases =-=[18, 17, 5]-=-, for example, by encrypting each record with its own key. Cryptography, however, is intended to conceal all information about the encrypted data, and it is therefore not conducive to situations in wh... |

12 |
Cryptography and NP-Completeness
- Even, Yacobi
- 1980
(Show Context)
Citation Context ... encrypted data, and it is therefore not conducive to situations in which we want to support some queries efficiently but not reveal the entire database. Cryptosytems founded on N P-complete problems =-=[16]-=- have been explored such as the Merkle-Hellman cryptosystem [23], which is based on the general knapsack problem. These systems rely on a series of tricks 8Randomize c-clause algorithm Let wi denote ... |

12 |
Translucent Databases
- Wayner
- 2002
(Show Context)
Citation Context |

9 |
Inference control via query restriction vs. data modification: a perspective
- Matloff
- 1988
(Show Context)
Citation Context ... obscured. Our method is almost the reverse of this approach, in that we support efficient membership queries but higher-level queries may be expensive. Our method is also related to query restriction=-=[22, 10, 12, 13, 28]-=-, where the query language is designed to support only the desired classes of queries. Although query restriction controls access to the data by outside users, it cannot protect an insider with full p... |

8 |
Problem solving: Hardness and easiness - hard and easy distributions of SAT problems
- Mitchell, Selman, et al.
- 1992
(Show Context)
Citation Context ...t indeed produce hard to reverse instances. In the case of negative databases there is a large body of literature that addresses this issue due to its isomorphism with the satisfability (SAT) problem =-=[24, 11]-=-. Of particular relevance, then, are one-way functions [20, 25]—functions that are easy to compute but hard to reverse and one-way accumulators [4, 9] which are essentially one-way hash functions with... |

3 |
J.E.: A note on cryptography and NP∩ coNP-P
- Brassard, Fortune, et al.
- 1978
(Show Context)
Citation Context ...every possible c-clause. to concel the existence of a “trapdoor” that permits retrieving the hidden information. However, almost all knapsack cyptosystems have been broken [26], and it has been shown =-=[7, 8]-=- that in general if breaking a cryptosystem is N P-hard then N P=co-N P, a point addressed in Section 4. If a scheme based on a N P-hard result, such as the one proposed here, is to be used in a priva... |