Karim Eldefrawy

Cryptography, Cybersecurity, Privacy

Co-founder and CTO at Confidencial.io
2017-2021: SRI
2011-2016: HRL Laboratories
2006-2010: PhD@UC Irvine

Scientific curiosity

Scientific knowledge map · Paper #20

Blindfolded Data Search via Secure Pattern Matching

Karim Eldefrawy and Sky Faber

2013 · IEEE Computer, Volume 46, Number 12

  • Applied
  • Perspective

What does the paper try to establish?

Which secure pattern-matching techniques can support practical “blindfolded” searches in which a requester hides the query and a data holder reveals only matching information, and what tradeoffs separate the available constructions?

What is the proposed answer?

The article organizes secure search around application requirements, adversary strength, supported match types, rounds, computation, and communication. It compares integer-comparison, FFT, matrix-multiplication/5PM, and garbled-circuit approaches and argues that no single construction dominates; expressive functionality, malicious security, concrete efficiency, streaming, and deployable evidence remain open tradeoffs.

Abstract

Balancing security and privacy concerns with information sharing is a top priority for corporations, law enforcement agencies, governments, and other organizations. Secure pattern matching (SPM) addresses some of the challenges faced in sharing and searching private data.

Provenance: Transcribed from the complete introductory deck in the checked-in magazine PDF; the article does not contain a section labeled Abstract. Only line-break artifacts were normalized.

Six dimensions, kept separate

The chart summarizes documented evidence and process. It is not a correctness probability, confidence score, or ranking, and no composite score is calculated.

The visual spider chart requires JavaScript. The complete values and rationales follow in text.

LowMediumHighN/A = not assessed

A smaller value means less documented support for that dimension, not that the paper is false or unimportant.

Epistemic evidence Medium

The article offers a structured, source-cited comparison with explicit functionality, adversary, and complexity dimensions and worked examples. It is a perspective rather than new theorem or experiment, and the underlying results were not independently revalidated.

Search variants and adversary models Protocol comparison table Practical gaps and open directions
Auditability High

A complete author-hosted article is checked into the site with page count and SHA-256 identity, and a coauthor copy is public. The comparative claims and references are inspectable; no new executable artifact is expected for this perspective.

Blindfolded-search motivation and central concept Protocol comparison table Practical gaps and open directions
Production provenance Medium

Named authorship, author/coauthor copies, and the IEEE DOI establish baseline provenance. Contributor roles, drafting history, and selection criteria for surveyed systems are not separately documented.

Blindfolded-search motivation and central concept Official magazine publication record
External scrutiny Medium

Publication in IEEE Computer indicates editorial and venue scrutiny, but review reports, corrections, and independent validation of the synthesis were not located.

Official magazine publication record
Reception Low

The dated targeted search did not expose a stable attributable citation count. Under the rubric, zero located citations falls in the low band; this does not assert that the article has never been cited.

Citation search snapshot
Contribution significance Medium

The article makes secure pattern matching accessible through applications and a useful taxonomy, but it is a synthesis and does not itself establish new security or performance results.

Blindfolded-search motivation and central concept Protocol comparison table Practical gaps and open directions

Assessment: Ai draft author review pending · 2026-07-11 · rubric 0.2. These dimensions describe documented support and process, not truth, correctness, or a universal ranking. No composite score is calculated.

Hierarchical knowledge map

Collapse a branch for a top-level reading, or follow its source links and child nodes to audit the evidence and boundaries underneath it.

paper

Blindfolded data search

An application-oriented synthesis of secure pattern-matching techniques, their adversary models, and their practical tradeoffs.

Blindfolded-search motivation and central concept
  1. application group

    Why private matching matters

    illustrative

    Examples include law-enforcement search of passenger manifests and clinician search of patient genomes, where both query sensitivity and data-holder obligations constrain ordinary data sharing.

    Law-enforcement and medical/genomic examples
  2. method family

    Integer-comparison family

    prior work synthesis

    Hazay-Toft-style constructions encode patterns and candidate substrings as integers and compare them obliviously, supporting malicious security but incurring costs that vary by exact, wildcard, and approximate functionality.

    Integer-comparison protocols
  3. method family

    FFT family

    prior work synthesis

    Vergnaud's approach uses encrypted polynomial/FFT operations to reduce computation for generalized matching while relying on additive ElGamal and a malicious-security construction.

    FFT-based secure matching
  4. method family

    Matrix-multiplication family

    prior work synthesis

    5PM reduces exact, wildcard, and non-binary approximate matching to character-delay-vector and matrix operations, obtaining linear communication in n+m while retaining O(nm)-scale cryptographic computation.

    Matrix-multiplication and 5PM approach
  5. method family

    Garbled-circuit family

    prior work synthesis

    General secure computation provides flexible text processing, but the surveyed construction has limitations in malicious security, wildcard/substring support, and pattern-length privacy.

    Garbled-circuit text processing
  6. limitation group

    Open practical gaps

    source synthesis

    Only 5PM among the compared protocols had an accompanying implementation at publication time; concrete cross-system benchmarks, covert-security design, multidimensional queries, and constant-memory streaming search remained underdeveloped.

    Practical gaps and open directions
  7. limitation

    What the article does not establish

    editorial boundary

    The perspective does not independently prove the surveyed protocols, reproduce their experiments, or demonstrate an end-to-end deployment in either example domain.

    Protocol comparison table Practical gaps and open directions
  8. scrutiny

    External scrutiny and reception

    magazine reviewed

    The article appeared in IEEE Computer. A targeted citation search did not expose a stable attributable count during this audit, and citing contexts or teaching use were not reviewed.

    Official magazine publication record Citation search snapshot

Source index

Locators state the depth of the current audit. PDF page numbers, where present, are one-based file pages; metadata-, summary-, and abstract-bounded records explicitly identify their limitations.

  1. Blindfolded-search motivation and central concept Introductory deck and opening section, PDF pages 1-2
  2. Law-enforcement and medical/genomic examples The Need for Blindfolded Data Search, PDF pages 1-2
  3. Search variants and adversary models Background, PDF pages 2-3
  4. Protocol comparison table Table 1, PDF page 4
  5. Integer-comparison protocols Survey section, PDF pages 4-5
  6. FFT-based secure matching Survey section, PDF pages 5-6
  7. Matrix-multiplication and 5PM approach Survey section and Figures 3-4, PDF pages 6-7
  8. Garbled-circuit text processing Survey section, PDF page 8
  9. Practical gaps and open directions Closing discussion, PDF page 8
  10. Official magazine publication record IEEE Computer 46(12), pages 68-75, DOI record
  11. Citation search snapshot ResearchGate and targeted exact-title searches were checked 2026-07-11, but no stable per-paper count was exposed. This is not evidence of zero lifetime citations.