Bloom Filter Data Leakage Prevention Using Mapping Functions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current data leakage prevention methods rely on heuristics and pattern matching, leading to high false positives and false negatives, and fail to detect sensitive data in different contexts, due to their rigidity and high memory requirements.
Innovation Solution
A computer-implemented method using a Bloom filter constructed from representations of sensitive data, where candidate data is mapped using a mapping function to perform membership queries, ensuring no false negatives and reducing memory usage while providing security and anonymity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If complete records of all sensitive data are stored for accurate data leakage detection, then detection accuracy is improved, but memory requirements become unreasonably high
Solution Approach 1:
The patent creates a Bloom filter data structure that contains representations (hashes) of sensitive data patterns rather than the actual sensitive data itself. This copying approach allows the system to maintain detection accuracy while using minimal memory, as the Bloom filter stores only compact hash representations instead of complete data records.
Solution Approach 2:
The patent transforms sensitive data into a different parameter representation through hashing and Bloom filter construction. By changing the data from its original form to a compact bit array representation, the system achieves both memory efficiency and detection capability, resolving the contradiction between storage size and detection accuracy.
2Quantity of substance
If heuristic-based approaches are used for data leakage prevention, then memory usage is reduced, but false positives and mis-categorisation increase
Solution Approach 1:
Instead of using heuristic approximations, the patent copies the actual sensitive data patterns into the Bloom filter structure. This allows for precise pattern matching while maintaining low memory usage, eliminating false positives caused by heuristic mis-categorisation of data such as confusing artificial test data with genuine sensitive data.
3Quantity of substance
If generalised properties about sensitive data are stored rather than real data, then memory requirements are reduced, but the system fails to spot sensitive data in different contexts
Solution Approach 1:
The Bloom filter data structure serves multiple functions: it stores exact sensitive data patterns, enables detection across different contexts and formats, and maintains memory efficiency. The mapping function transforms various data representations into a universal format that the Bloom filter can detect regardless of context, making the system both memory-efficient and context-adaptable.
4Ease of manufacture
If rigid rule-based systems are used for data detection, then implementation simplicity is improved, but flexibility to detect sensitive data in different formats is reduced
Solution Approach 1:
The patent employs a mapping function that transforms sensitive data from various formats and contexts into a standardized representation suitable for Bloom filter storage. This parameter transformation approach maintains implementation simplicity while achieving high format flexibility, as the mapping function handles various data types and the Bloom filter detects patterns regardless of their original context or format.
Data Source
AI summary
This disclosure is in the field of data leakage prevention and relates to computer-implemented methods for identifying data leakage and for enabling identification of data leakage, and data processing apparatus, computer programs, and computer readable storage media for performing the same. There are provided computer-implemented methods of constructing and querying a first Bloom filter, formed from a plurality of representations of sensitive data, wherein the plurality of representations have been generated from underlying sensitive data using a mapping function.