Investigation Apparatus for Table Data Similarity Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In posterior investigations for information leakage, the large volume of feature data from table data leads to prolonged investigation times due to the difficulty in treating and comparing keyword-based feature data, especially when dealing with table data that has varying attributes and records.
Innovation Solution
An investigation apparatus that analyzes the uniqueness of values in each attribute of a table, reduces the number of values by a predetermined ratio for attributes with lower uniqueness, and outputs features for both the investigation original and target tables, allowing for efficient comparison and similarity investigation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If feature data of table data is recorded for posterior investigation, then the ability to specify leaked original is improved, but the investigation time becomes long due to huge data volume
Solution Approach 1:
The patent extracts only the necessary feature data from table data for investigation purposes. Instead of recording and comparing entire table datasets, the system extracts specific feature elements (such as unique identifiers, key attribute combinations) that are sufficient for identifying leaked originals, thereby reducing the volume of data to be processed while maintaining investigation effectiveness
Solution Approach 2:
The patent transforms the investigation approach by changing parameters of feature data representation. It converts detailed table data into condensed feature representations with controlled granularity, adjusting the level of detail retained based on investigation needs. This parameter transformation reduces data volume while preserving the ability to identify similar tables
2Measurement precision
If all values in table attributes are retained for feature data, then the accuracy of similarity detection is improved, but the data amount becomes huge
Solution Approach 1:
The patent applies local quality by treating different attributes of table data differently based on their investigative value. Instead of uniformly processing all attributes, the system identifies and retains features from attributes that are most discriminatory for detecting similar tables, while reducing or eliminating features from less important attributes. This selective approach maintains detection accuracy for critical features while reducing overall data volume
Data Source
AI summary
An investigation apparatus according to an embodiment includes a processor that executes a process including: analyzing uniqueness of a plurality of values included in an attribute for each attribute of a table; reducing a plurality of values included in an attribute of which the analyzed uniqueness is lower than a predetermined value by a predetermined ratio with respect to an investigation original table and outputting a feature of the investigation original table; reducing a plurality of values included in an attribute of which the analyzed uniqueness is higher than a predetermined value by a predetermined ratio with respect to an investigation target table and outputting a feature of the investigation target table; and investigating similarity of the investigation target table to the investigation original table by comparing the feature of the investigation original table with the feature of the investigation target table.