A fuzzy search method for protecting the privacy of the querying party

Abstract

The present invention belongs to the field of information security technology, and discloses a fuzzy search method for protecting the privacy of the query party, comprising: the query party encrypts the query keyword, and after encryption, the query keyword is sent to the data party; the data party uses a fuzzy matching algorithm to process the encrypted query keyword; establishes an encrypted data index; returns the encrypted search results; decrypts the search results and displays them. This method can securely retrieve information without leaking the user's query content. The present invention uses homomorphic encryption technology and fuzzy matching algorithms to achieve high-performance fuzzy search. The query party's search keywords are always in an encrypted state, ensuring the security of user privacy. Through specially designed data indexes and encrypted search result return mechanisms, both the efficiency of the search and the security of the data content are guaranteed. The present invention solves the limitations of traditional privacy information retrieval technology in processing fuzzy queries, and improves the application flexibility and practical application scope of the search system.

Description

Technical Field

[0001] The present invention belongs to the field of information security technology, and in particular relates to a fuzzy search method for protecting the privacy of a querying party. Background Art

[0002] Privacy Information Retrieval (PIR) is an important branch of information security that focuses on protecting the privacy of users when performing data queries. The main goal of PIR is to securely retrieve information without revealing the content of the user's query. Traditional technologies mainly support precise queries, that is, users must know exactly what information they want to retrieve, which is impractical in many practical application scenarios. The core of these technologies is to ensure the security of data and the privacy of user queries, avoiding the leakage of sensitive information during data transmission or processing. Such as medical records, financial information, and personal identity data. The main goal of PIR is to securely retrieve information without revealing the content of the user's query. This is particularly important in fields such as medical, financial, and personal data.

[0003] Fuzzy search methods offer a significant extension to traditional PIR techniques. This approach not only protects user privacy but also allows for more flexible queries, even when users are uncertain about the information they seek. In many cases, users may only partially remember information or wish to explore related information. Fuzzy search allows for this imprecise querying, better serving users' actual needs. A key challenge in designing fuzzy search methods is balancing flexibility with high performance and privacy protection. This can involve complex algorithms and security protocols.

[0004] In recent years, research has begun exploring the integration of machine learning techniques with PIR. These efforts attempt to improve the search efficiency and accuracy of PIR systems while protecting user privacy. Furthermore, as data volumes increase, the demand for fuzzy matching also grows. These techniques allow users to search based on partial or approximate keywords while maintaining data privacy. Recent research focuses on improving the efficiency of PIR systems. For example, researchers have explored different data indexing structures and query optimization algorithms to reduce query time and increase data retrieval speed.

[0005] Therefore, a significant technical challenge arises: how to improve the PIR system's ability to support fuzzy queries without sacrificing user privacy. This requires the development of new algorithms and methods that can not only handle inaccurate or partial keyword queries, but also maintain data security and response efficiency. Summary of the Invention

[0006] The main goal of this invention is to propose a novel fuzzy search method to address the technical issues of traditional privacy information retrieval (PIR) technology, which suffers from insufficient performance and efficiency when processing fuzzy or partial match queries. To achieve this goal, the specific technical solutions of this invention are as follows:

[0007] A fuzzy search method for protecting the privacy of a querying party comprises the following steps:

[0008] Step 1: The query party encrypts the query keyword and sends it to the data party.

[0009] Step 2: The data provider uses a fuzzy matching algorithm to process the encrypted query keywords;

[0010] Step 3: Create an encrypted data index;

[0011] Step 4: Return the encrypted search results;

[0012] Step 5: Decrypt the search results and display them.

[0013] Furthermore, in step 1, the querying party uses homomorphic encryption technology to encrypt the query keyword.

[0014] Furthermore, in step 2, the data provider estimates the similarity between the encrypted data item and the encrypted query keyword by designing a homomorphic encryption fuzzy matching scoring system. The steps for designing the homomorphic encryption fuzzy matching scoring system are as follows:

[0015] Step 2.1. Select a homomorphic encryption model based on the learning error problem.

[0016] Step 2.2: Encrypt and configure parameters of the feature vector generated by the homomorphic encryption model, including key size and error rate;

[0017] For each element in the private data D that the query party wants to query, calculate its feature vector f → (D), eigenvector f → (D) includes word frequency and context;

[0018] Apply the homomorphic encryption model with learning error problem to obtain the encrypted feature vector Enc(f → (D)), ensure that the encrypted feature vector Enc(f → (D)) preserves mathematical properties that can be used in subsequent homomorphic operations;

[0019] Step 2.3: Build index and query encryption:

[0020] Use B-tree or hash table data structures to index encrypted feature vectors;

[0021] When querying, the query keyword Q is converted into an encrypted feature vector Enc(f → (Q));

[0022] Step 2.4, homomorphic encryption similarity calculation:

[0023] Design a homomorphic similarity function H, which can be used to calculate Enc(f → (D)) and Enc(f → (Q)) is operated to generate the encrypted similarity score Enc(S);

[0024] Step 2.5: Implement the scoring system:

[0025] The encryption threshold Enc(T) is introduced to filter the scores. Only the scores Enc(S) above this threshold will be considered.

[0026] Applying the homomorphic property in the learning error problem model to linearly combine and compare encrypted scores;

[0027] Step 2.6, sorting of encryption results:

[0028] A sorting protocol is constructed using homomorphic properties, which enables the sorting protocol to sort the encrypted scores without decryption;

[0029] Step 2.7: Decryption and display of encryption results:

[0030] The querying party uses its private key to decrypt the returned encrypted scores to obtain the final similarity score list; the decrypted results are displayed and sorted from high to low by score.

[0031] Furthermore, step 3 specifically includes:

[0032] Step 3.1: Extract a fixed set of features from each item in the database;

[0033] Step 3.2: Encrypt each feature vector using the linear homomorphic encryption scheme, and obtain the homomorphic encrypted feature vector Enc(f → (D)) can be used for subsequent encryption operations;

[0034] Step 3.3: Add multi-dimensional data processing based on the KD tree structure, and modify it to support encrypted data processing capabilities; use a linear homomorphic encryption scheme based on the learning error problem to encrypt the feature vector of each data item in the database;

[0035] Select an encrypted vector from the encrypted feature vector set as the root node, recursively select the split point according to the fixed or predefined dimension order, divide the data into two parts, and construct subtrees for each part. For each encrypted feature vector, determine the subtree it should be inserted into based on the comparison result between its value in the selected dimension and the node, and recursively continue until a suitable insertion position is found;

[0036] Step 3.4: Insert the homomorphically encrypted feature vector of each data item into the index structure.

[0037] Step 3.5: Compare the query vector and the encrypted vector in the index through the algorithm;

[0038] Step 3.6: When the encrypted query vector is received, the index structure is traversed to find the most similar data item. The search is completed by comparing the encrypted query vector with the encrypted feature vector of the data item in the index structure.

[0039] Step 3.7: Sort the results by homomorphic encryption based on the searched encryption similarity scores.

[0040] Furthermore, in step 4, the sorted encrypted results are sent to the querying party, and the querying party decrypts the results using the private key to obtain the similarity score and data item.

[0041] Furthermore, in step 5, after receiving the encrypted search results, the querying party uses the private key to decrypt the encrypted search results.

[0042] The present invention has the following advantages:

[0043] High privacy protection: Through full encryption processing, user queries are ensured not to be leaked, and the server is unable to access unencrypted user data.

[0044] Efficient data processing: Utilizes optimized encryption algorithms and index structures to significantly improve the speed and accuracy of processing fuzzy or partial match queries.

[0045] Flexibility and scalability: Supports various types of fuzzy queries, including misspellings and synonym processing, and is suitable for a variety of business and scientific research scenarios.

[0046] The present invention not only improves the efficiency and effectiveness of data retrieval, but also provides strong protection for the privacy of the querying party, and promotes the application and development of privacy protection technology in the field of data retrieval. The present invention can retrieve information securely without disclosing the user's query content. The present invention mainly realizes high-performance fuzzy search through homomorphic encryption technology and fuzzy matching algorithm. In the present invention, the search keywords of the querying party are always in an encrypted state, ensuring the security of user privacy. At the same time, through specially designed data indexing and encrypted search result return mechanism, both the efficiency of the search and the security of the data content are guaranteed. In addition, the present invention also solves the limitations of traditional privacy information retrieval technology in processing fuzzy queries, and improves the application flexibility and practical application scope of the search system. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 Schematic diagram of the method of the present invention. DETAILED DESCRIPTION

[0048] In order to better understand the purpose, technical solutions and functions of the present invention, Figure 1 , the present invention is described in further detail. However, the present invention may be implemented in a variety of different ways as defined and covered by the claims. The accompanying drawings, which constitute a part of this invention, are provided to provide a further understanding of the invention. The exemplary embodiments of the invention and their descriptions are used to explain the invention and do not constitute undue limitations of the invention.

[0049] Step 1: Encrypt the query keyword.

[0050] First, the querying party needs to encrypt its query keywords to ensure that the user's search intent is not exposed during transmission. Homomorphic encryption technology is used. This technology allows the data party to operate and calculate on encrypted data without decrypting the data. The specific content of homomorphic encryption is as follows:

[0051] (1) Design a lightweight fully homomorphic encryption algorithm, named "Probabilistic Homomorphic Encryption Fuzzy Matching Algorithm" (PHEFMA), which provides data providers with greater flexibility in processing ciphertext data. The following is the design framework:

[0052] (2) Key Generation (KeyGen):

[0053] Generate two large prime numbers p and q using a secure pseudorandom number generator (PRNG).

[0054] Calculate n = p × q, where n is part of the public key.

[0055] Appropriate noise distribution parameters are selected to ensure the security and homomorphic properties of encryption.

[0056] (3) Encryption (Enc):

[0057] Represent the query keyword m as an integer.

[0058] Select random noise r and calculate the ciphertext c = (m + 2r) mod n.

[0059] The ciphertext c is a number modulo n that hides the message m.

[0060] (4) Homomorphic operation:

[0061] The algorithm is designed to allow addition and multiplication operations on the ciphertext c.

[0062] Addition: c1+c2modn corresponds to the addition of the plaintext.

[0063] Multiplication: c1×c2modn corresponds to the multiplication of the plaintext.

[0064] Step 2: Design of fuzzy matching algorithm.

[0065] After encryption, the query keywords are sent to the data provider. The data provider uses a lightweight, fully homomorphic encryption fuzzy matching algorithm to process these encrypted query keywords to support fuzzy matching. A homomorphic encryption fuzzy matching scoring system is designed to estimate the similarity between the encrypted data items and the encrypted query keywords. The technical details of the homomorphic encryption fuzzy matching scoring system are as follows:

[0066] (1) Basic homomorphic encryption model selection

[0067] Based on the consideration of balancing security and efficiency, a homomorphic encryption model based on the learning error problem (LWE) is selected.

[0068] After selecting LWE as the homomorphic encryption model, parameters (including key size and error rate) will be adjusted for feature vector encryption to ensure that when encrypting the feature vector, neither the security of the encrypted data nor the mathematical operation capability after encryption is compromised.

[0069] (2) Feature vector encryption

[0070] For each element in the data item D, calculate its eigenvector f → (D), including word frequency, context, etc.

[0071] Apply the LWE homomorphic encryption algorithm to obtain Enc(f → (D)), ensuring that these cryptographic features preserve mathematical properties that can be used in subsequent homomorphic operations.

[0072] (3) Building indexes and query encryption

[0073] Use data structures such as B-trees or hash tables to index encrypted feature vectors and optimize retrieval performance.

[0074] When querying, the query keyword Q is also converted into an encrypted feature vector Enc(f → (Q)).

[0075] (4) Homomorphic encryption similarity calculation

[0076] Design a homomorphic similarity function H, which can be used to calculate the similarity between two encrypted vectors Enc(f → (D)) and Enc(f → (Q)) is operated to generate the encrypted similarity score Enc(S).

[0077] A homomorphic version of the vector dot product is used, which can be performed directly on the ciphertext due to its linear property.

[0078] (5) Implementation of the scoring system

[0079] An encryption threshold Enc(T) is introduced to filter the scores, and only scores Enc(S) above this threshold will be considered.

[0080] The homomorphic property in the LWE model is used to perform linear combination and comparison of encrypted scores.

[0081] (6) Sorting of Encrypted Results

[0082] A secure sorting protocol is constructed using homomorphic properties, which enables it to sort encrypted scores without decryption.

[0083] (7) Decryption and display of encryption results

[0084] The querying party uses its private key to decrypt the returned encrypted scores and obtain the final similarity score list.

[0085] Display the decrypted results, sorted by score from high to low.

[0086] The scoring system is designed based on a series of probabilistic models and encrypted data features. The design of this algorithm is the key to the system, which needs to identify similar data items without decrypting the query keyword.

[0087] Step 3: Create an encrypted data index

[0088] To improve search efficiency, a special index is created for the data provider's database. This index, based on the encrypted form of the data item, allows the system to quickly retrieve results that fuzzily match the encrypted query keyword. This indexing method ensures both search efficiency and the security of the database content.

[0089] The specific process involved in encrypting data indexes is as follows:

[0090] (1) Data Feature Extraction

[0091] A fixed set of features, such as term frequency (TF) and inverse document frequency (IDF), is extracted from each item in the database.

[0092] (2) Homomorphic Encryption of Feature Vectors

[0093] Each feature vector is encrypted using a linear homomorphic encryption scheme (an encryption scheme using LWE).

[0094] The homomorphic encryption feature vector Enc(f → (D)) can be used for subsequent encryption operations.

[0095] (3) Index Structure Establishment

[0096] (3.1) Index structure design and feature vector encryption

[0097] The KD-tree structure is modified to support multidimensional data processing and encrypted data. The feature vector of each data item in the database is encrypted using a linear homomorphic encryption scheme based on LWE. This ensures that all data remains encrypted during indexing and searching, ensuring data security. Because encrypted data cannot be directly compared in size, a fixed dimension order or split dimensions based on previously observed statistics can be selected to facilitate data segmentation during index building and querying.

[0098] (3.2) Index creation

[0099] An encrypted vector is selected from the set of encrypted feature vectors as the root node. Split points are recursively selected in a fixed or predefined order of dimensions, dividing the data into two parts and constructing subtrees for each. For each encrypted feature vector, the subtree to be inserted is determined based on the comparison of its value in the selected dimension with the node (using the designed encrypted data comparison logic). This recursive process continues until a suitable insertion location is found.

[0100] (4) Index Population

[0101] Insert the homomorphically encrypted feature vector of each data item into the index structure.

[0102] The tree structure is optimized according to the dimension of the encrypted feature vector to ensure search efficiency.

[0103] (5) Implementation of Homomorphic Encrypted Fuzzy Matching Algorithm

[0104] Implement an algorithm to compare the query vector with the encrypted vector in the index.

[0105] This algorithm can be a homomorphically encrypted dot product operation, which allows computing a similarity score on the ciphertext.

[0106] (6) Index Searching

[0107] When an encrypted query vector is received, the index structure is traversed to find the most similar data item.

[0108] Searching is done by comparing the encrypted query vector with the encrypted feature vector of the data item in the index structure.

[0109] (7) Homomorphic Encrypted Result Processing

[0110] The results are sorted by homomorphic encryption based on the encrypted similarity scores found in the search.

[0111] A comparison operation implemented using homomorphic encryption technology determines which encrypted results best match the query.

[0112] Step 4: Encrypt the search results returned:

[0113] The sorted encrypted results are sent to the querying party, who decrypts the results using the private key to obtain the actual similarity score and data item.

[0114] When matching data is found, it is encrypted and sent back to the querying party. This step ensures that even if the search results are intercepted during transmission, they cannot be understood or used by unauthorized parties.

[0115] Step 5: Decryption and display of results:

[0116] After receiving the encrypted search results, the querying party uses its private key to decrypt them. The decrypted data is the information the end user needs. This step ensures that the end user can securely receive and view the search results.

[0117] Through the above scheme, the present invention realizes a search system that supports fuzzy search while protecting user privacy. The user's search keywords are encrypted from beginning to end, protecting the user's private information from being leaked. At the same time, through the special algorithm and data structure design, the efficiency and accuracy of the search are guaranteed.

[0118] This invention is a novel fuzzy search method that protects the privacy of the querying party. By leveraging encryption technology, specifically homomorphic encryption, it allows users to perform fuzzy searches even when the query keywords are incomplete or partially matched. This involves the application of homomorphic encryption, fuzzy matching algorithms, encrypted data indexing, and encrypted result transmission.

[0119] Homomorphic encryption allows computations to be performed on encrypted data through mathematical construction, ensuring that the decrypted ciphertext of the computation is identical to the plaintext. Homomorphic encryption allows operations such as sorting and matching to be performed on encrypted data without decryption, thus protecting the privacy of the query content.

[0120] Fuzzy matching algorithms: Combined with natural language processing technology, they analyze semantics and context to handle imprecise queries. Leveraging probabilistic models and text analysis techniques, the algorithm can identify and handle misspellings, synonyms, and semantically similar words.

[0121] For encrypted data indexing: We use specialized data structures, such as inverted indexes and B-trees, that are optimized to support efficient searches of encrypted data. We also create indexes to support fast fuzzy matching, allowing the system to quickly locate and return results that fuzzily match the query criteria.

[0122] Encrypted result transmission: Encryption protocols (such as TLS / SSL) are used to ensure the confidentiality and integrity of data transmitted over public networks. Search results are encrypted before being returned to users to protect the privacy and security of data during transmission.

[0123] The specific application of the method of the present invention is described below through a specific embodiment.

[0124] Application scenarios:

[0125] xx Network Security Data Management. In the xx network security environment, processing and managing large amounts of sensitive data is crucial. The fuzzy search method of this invention aims to improve the efficiency of data retrieval while ensuring data security and privacy protection.

[0126] System Configuration:

[0127] Encrypted communication network: Build a highly encrypted communication network to ensure that all data transmissions are encrypted with strict security protocols.

[0128] Central Control Server: A high-performance central server is used to receive, process, and respond to queries from within the organization. The server utilizes advanced fuzzy matching algorithms and integrates the latest encryption technology.

[0129] User terminal interface: The terminal device used by xx members is equipped with specially designed query software for generating and sending encrypted query requests.

[0130] Operation process:

[0131] Query Request Encryption: XX members initiate ambiguous query requests on the terminal device, such as partial information or incomplete identifiers for a specific security incident. This query request is encrypted by the terminal software before being sent.

[0132] Central server processing: Encrypted queries are transmitted over a secure network to a central server. The server processes these queries using specialized fuzzy matching algorithms, effectively locating relevant information within large amounts of sensitive data.

[0133] Result encryption and transmission: Once the matching data is located, the central server encrypts the result and sends it back to the requesting terminal via an encrypted communication network.

[0134] Terminal result decryption and display: The terminal device receives the encrypted search results, decrypts them and displays them to the operator to ensure the confidentiality and integrity of the information.

[0135] The present invention can produce the following beneficial technical effects in improving the existing privacy information retrieval (PIR) technology:

[0136] 1. Enhanced query flexibility and precision: By supporting fuzzy search, the present invention significantly improves query flexibility. Users can search based on imprecise keywords or fuzzy conditions, resulting in a wider range of relevant search results. This is particularly useful when processing complex or incomplete queries, improving overall search efficiency and user experience.

[0137] 2. Expanded Application Scope: The flexible search functionality of the present invention makes it applicable to a wider range of application scenarios. In particular, in areas that require in-depth data analysis and flexible information retrieval, such as business intelligence, medical information systems, and academic research, the present invention can provide richer data insights.

[0138] 3. Improved processing speed: Addressing the performance issues of existing technologies when processing large-scale data sets, the technical solution proposed in this invention can significantly improve data processing speed. This is particularly important in scenarios that require real-time data processing and rapid response.

[0139] 4. Maintaining high-level data privacy and security: While improving functionality, this invention still emphasizes data privacy and security protection. This means that even when performing more complex fuzzy searches, the user's query content and personal data are fully protected.

[0140] 5. Reduced data processing costs: By optimizing algorithms and data processing processes, the present invention can reduce the overall cost of data processing. This is particularly beneficial for organizations that need to process large amounts of data, effectively reducing the burden on hardware and computing resources.

[0141] Compared to existing technologies, this invention offers significant benefits in terms of enhanced search flexibility, faster processing, expanded application scope, data privacy protection, and reduced costs. These improvements not only optimize the user experience but also provide a more efficient and secure data processing solution for data-intensive industries.

[0142] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A fuzzy search method for protecting the privacy of the querying party, characterized in that: The steps include: Step 1: The querying party uses homomorphic encryption technology to encrypt the query keyword. After encryption, the query keyword is sent to the data party. The homomorphic encryption technology includes: Key generation: Use a secure pseudo-random number generator to generate two large prime numbers p and q, calculate n = p × q, where n is part of the public key, and select appropriate noise distribution parameters; Encryption: Represent the query keyword m as an integer, select random noise r, and calculate the ciphertext c = (m + 2r) mod n. The ciphertext c is a number under the modulus n, which hides the message m. Homomorphic operation: Design the following algorithm to allow addition and multiplication operations on the ciphertext c: Addition: c1+c2modn, corresponding to the addition of plaintext; Multiplication: c1×c2modn, corresponding to the multiplication of plaintext; Step 2: The data party uses a fuzzy matching algorithm to process the encrypted query keyword. Specifically, the data party estimates the similarity between the encrypted data item and the encrypted query keyword by designing a homomorphic encryption fuzzy matching scoring system. The steps of designing the homomorphic encryption fuzzy matching scoring system are as follows: Step 2.

1. Select a homomorphic encryption model based on the learning error problem. Step 2.2: Encrypt and configure parameters of the feature vector generated by the homomorphic encryption model, including key size and error rate; For each element in the private data D that the query party wants to query, calculate its feature vector f → (D), eigenvector f → (D) includes word frequency and context; Apply the homomorphic encryption model with learning error problem to obtain the encrypted feature vector Enc(f → (D)), ensure that the encrypted feature vector Enc(f → (D)) preserves mathematical properties that can be used in subsequent homomorphic operations; Step 2.3: Build index and query encryption: Use B-tree or hash table data structures to index encrypted feature vectors; When querying, the query keyword Q is converted into an encrypted feature vector Enc(f → (Q)); Step 2.4, homomorphic encryption similarity calculation: Design a homomorphic similarity function H, which can be used to calculate Enc(f → (D)) and Enc(f → (Q)) is operated to generate the encrypted similarity score Enc(S); Step 2.5: Implement the scoring system: The encryption threshold Enc(T) is introduced to filter the scores. Only the scores Enc(S) above this threshold will be considered. Applying the homomorphic property in the learning error problem model to linearly combine and compare encrypted scores; Step 2.6, sorting of encryption results: A sorting protocol is constructed using homomorphic properties, which enables the sorting protocol to sort the encrypted scores without decryption; Step 2.7: Decryption and display of encryption results: The querying party uses its private key to decrypt the returned encrypted scores and obtain the final similarity score list; the decrypted results are displayed and sorted from high to low by score; Step 3: Create an encrypted data index, including: Step 3.1: Extract a fixed set of features from each item in the database; Step 3.2: Encrypt each feature vector using the linear homomorphic encryption scheme, and obtain the homomorphic encrypted feature vector Enc(f → (D)) can be used for subsequent encryption operations; Step 3.3: Add multi-dimensional data processing based on the KD tree structure, and modify it to support encrypted data processing capabilities; use a linear homomorphic encryption scheme based on the learning error problem to encrypt the feature vector of each data item in the database; Select an encrypted vector from the encrypted feature vector set as the root node, recursively select the split point according to the fixed or predefined dimension order, divide the data into two parts, and construct subtrees for each part. For each encrypted feature vector, determine the subtree it should be inserted into based on the comparison result between its value in the selected dimension and the node, and recursively continue until a suitable insertion position is found; Step 3.4: Insert the homomorphically encrypted feature vector of each data item into the index structure and optimize the tree structure based on the dimension of the encrypted feature vector. Step 3.5: Compare the query vector and the encrypted vector in the index using a homomorphic encryption dot product operation, which can calculate a similarity score on the ciphertext. Step 3.6: When the encrypted query vector is received, the index structure is traversed to find the most similar data item. The search is completed by comparing the encrypted query vector with the encrypted feature vector of the data item in the index structure. Step 3.7: Homomorphically encrypt and sort the results based on the encrypted similarity scores found, and use a comparison operation implemented using homomorphic encryption technology to determine which encrypted results match the query. Step 4: Returning the encrypted search results, including: sending the sorted encrypted results to the querying party, which decrypts the results using the private key to obtain the similarity score and data items. When matching data is found, the data is encrypted and sent back to the querying party. Step 5: decrypting the search results and displaying them, including: after the querying party receives the encrypted search results, it uses the private key to decrypt the encrypted search results.

Images