Database Index Tokenization for Data Security

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing database management systems (DBMS) face challenges in ensuring adequate protection of indexed data, as conventional index structures can reveal sensitive information, making it difficult to protect pseudonymized data sets from reconstruction by unauthorized parties.

Innovation Solution

A computer-implemented method for creating an index that tokenizes field values from multiple field types, generating tokens regardless of their origin, and storing them in a searchable index structure without referencing their original field types, thereby obscuring the semantic meaning of the data.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Speed

If conventional index structures are used to store field values, then searching efficiency is improved, but data security deteriorates because sensitive information can be reconstructed from the index

Engineering Contradiction:
Improvesearching efficiencyVSAvoiddata security risk
Core Design Contradiction:
SpeedVSObject-affected harmful factors

Solution Approach 1:

The patent applies segmentation by dividing field values into multiple tokens based on different field types. Instead of storing complete field values in the index, the system segments them into type-specific tokens (first tokens from first field type, second tokens from second field type, etc.). This segmentation prevents unauthorized reconstruction of original data while maintaining search capability through token matching.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent implements local quality by storing different types of tokens in different locations within the index structure. First tokens, second tokens, and tokens from other field types are stored in distinct regions of the index. This local differentiation ensures that even if the index is accessed, the semantic meaning of individual tokens cannot be determined, thereby protecting data security while enabling efficient searching.

Inventive Principle:
Principle #3Local quality

2Object-affected harmful factors

If field values are tokenized and stored in the index, then data protection is improved, but the complexity of index creation increases

Engineering Contradiction:
Improvedata protectionVSAvoidindex creation complexity
Core Design Contradiction:
Object-affected harmful factorsVSDevice complexity

Solution Approach 1:

The patent applies universality by creating a unified index structure that handles multiple field types through a single tokenization process. The same index structure and storage mechanism is used for all field types, with tokens being differentiated only by their storage location rather than requiring separate index structures for each field type. This reduces overall system complexity while maintaining strong data protection.

Inventive Principle:
Principle #6Universality (Multi-functionality)

3Measurement precision

If tokens are stored with their original field type information, then search accuracy is improved, but data confidentiality deteriorates

Engineering Contradiction:
Improvesearch accuracyVSAvoiddata confidentiality
Core Design Contradiction:
Measurement precisionVSLoss of information

Solution Approach 1:

The patent resolves this contradiction by moving the field type differentiation from the token value dimension to the spatial dimension of the index structure. Instead of encoding field type information within the token itself (which would compromise confidentiality), the system stores tokens from different field types in different locations within the index. Search accuracy is maintained through token matching, while confidentiality is preserved because the location-based differentiation does not reveal semantic meaning to unauthorized users.

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

Data Source

PatentEP3552140B1Database index comprising multiple fields
Publication Date: 2020.11.11 BUNDESDRUCKEREI GMBH
  • EP3552140B1 patent drawingFigure 1~2
  • EP3552140B1 patent drawingFigure 3~4
  • EP3552140B1 patent drawingFigure 5

AI summary

The invention relates to a computer-implemented method for creating an index (154). The method comprises: – providing (170) a DBMS (160) configured to store data records (DR1 -DR7) each structured as a set comprising multiple field values, wherein the field values are each stored in a field, wherein the fields of each of the data records belong to at least two different field types (F1 - F7), – generating (172) first tokens (250) from first field values of multiple data records, wherein the first field values are stored in first instances of the fields, wherein the first fields belong to a first field type (F1); – generating (174) second tokens (252, 254) from second field values of the multiple data records, wherein the second field values are stored in second instances of the fields, wherein the second fields belong to a second field type (F2, F5); – generating (176) a searchable index (154) from at least the first and second tokens, wherein the location of the storage of the first and second tokens in the structure of the index is independent of which of the fields the tokens come from.