Search Query Obfuscation via Character String Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing search technologies using public search engines risk disclosing confidential search terms, and previous methods either yield irrelevant results or incur high processing loads, making it difficult to efficiently acquire required information without revealing secret information.
Innovation Solution
A search apparatus and method that generates valid and noise character strings from input search words, combining them to create queries that are transmitted to a public search engine, allowing for a second local search to filter results and maintain secrecy while improving search accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If search terms are transmitted to a public search engine, then search results can be obtained, but confidential search terms are disclosed to the public
Solution Approach 1:
The search term is divided into multiple character strings of different lengths (e.g., 2-character, 3-character, 4-character strings from a 5-character search term). This segmentation allows the search to be performed without transmitting the complete original search term, thus protecting confidentiality while maintaining search effectiveness through multiple partial matches.
Solution Approach 2:
Character strings of various lengths serve as intermediaries between the confidential search term and the public search engine. Instead of directly transmitting the secret search term, the system transmits derived character strings that act as mediators, enabling the search engine to return relevant results without exposing the original confidential term.
2Loss of information
If search conditions are concealed or obscured, then search term secrecy is maintained, but very few relevant search results are identified
Solution Approach 1:
By segmenting the search term into multiple overlapping character strings (2-char, 3-char, 4-char), the system generates multiple search queries that collectively retrieve more results than a single obscured query would produce, thereby maintaining both secrecy and result quantity.
Solution Approach 2:
The system combines multiple search results from different character string queries to produce a comprehensive final result set. This merging approach ensures that even though each individual query uses obscured terms, the aggregated results provide sufficient quantity and relevance.
3Loss of information
If search conditions are changed by deletion or synonym replacement, then some secrecy is maintained, but an enormous amount of irrelevant search results are acquired
Solution Approach 1:
The segmentation approach preserves the original characters from the search term rather than replacing them with synonyms or deleting them. This ensures that the character strings remain precise and relevant to the original search intent, avoiding the generation of irrelevant results while still protecting the complete search term.
4Loss of information
If encryption is used to conceal search conditions, then search term secrecy is maintained, but processing load becomes excessively high
Solution Approach 1:
Instead of using computationally expensive encryption algorithms, the system employs a simple, low-cost transformation method (extracting character strings of different lengths) that requires minimal processing power. This disposable-like approach to obfuscation achieves secrecy without the high processing load of traditional encryption.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Input of a desired search word consisting of three or more characters is received, and two or more arbitrary consecutive characters are extracted from the search word. Then, multiple valid character strings each consisting of the thus extracted characters are generated. On the other hand, an arbitrary word is selected as a noise word from a dictionary appropriate to a search engine to be used, and two or more arbitrary consecutive characters are selected from the noise word. Then, a noise character string consisting of the thus extracted characters is generated. A query is generated by combining the valid character strings and the noise character strings, and is transmitted to the search engine. A second search is performed by use of the search word over a search result transmitted back from the search engine in reply to the query, and a final search result of the second search is displayed.