CN-122027129-A - Privacy set intersection processing method under dynamic data scene

CN122027129ACN 122027129 ACN122027129 ACN 122027129ACN-122027129-A

Abstract

The invention belongs to the technical field of private data sharing processing, and particularly relates to a private collection intersection processing method under a dynamic data scene. The method comprises the steps of initializing a secret key, constructing a query, transmitting the privacy and decrypting the result, respectively generating public and private keys required by careless transmission by a requester and a server, encrypting all query data by the public keys, calculating query hashes by the requester based on local hash characteristic values, constructing a query request, encrypting and sending a query set to the server, encrypting and matching the server and the requester after the requester and the server realize linear combination of the privacy, receiving ciphertext data returned by the server, decrypting by using shared information and the private keys, and finally restoring the matched hash characteristic set. The method and the device are used for solving efficiency bottleneck and security risk in a dynamic data scene, realizing matching through careless transmission and privacy interaction, protecting data privacy, supporting large-scale data set searching, reducing communication and calculation expenditure and improving query efficiency.

Inventors

HU XIANJUN
ZHANG CHUNLEI
YUAN JINGMIN
NIU JINGHUA
NAI HE

Assignees

中国人民解放军海军工程大学

Dates

Publication Date: 20260512
Application Date: 20260130

Claims (6)

1. A method for processing privacy set intersection under a dynamic data scene is characterized by comprising four basic steps of key initialization, query construction, privacy transmission and result decryption; key initialization, namely, a requester and a server respectively generate public and private keys required by careless transmission and encrypt all inquiry data by using the public keys, wherein the requester generates a key pair , In order for the requester to be a public key, Generating a key pair for a requester private key by a server Wherein In order to be a server public key, The request party and the server encrypt the query data and the response data respectively, and establish a basic communication channel; query construction refers to the calculation of a query hash by a requestor based on local hash feature values: Constructing and encrypting the query request Then the requesting party sends the query set Feeding the server; The privacy transmission is that after the requester and the server realize the linear combination of privacy, the server and the requester are encrypted and matched, whether the query data are matched with the local data is judged, the intersection of the matching characteristics is calculated with the server, and the intersection of the matching characteristics is output; Result decryption, namely that the requester receives ciphertext data returned by the server And decrypting by using the shared information and the private key generated by the VOLE, and finally restoring the matched hash feature set.
2. The method for processing privacy set intersection under dynamic data scene according to claim 1, wherein the privacy transmission specifically comprises encryption matching and feature intersection calculation; the encryption matching means that the server and the requesting party encrypt the hash feature respectively, the requesting party encrypts the hash feature by using the public key of the requesting party, and the server encrypts the corresponding feature by using the public key of the requesting party; The feature intersection calculation comprises that a server processes encrypted hash features submitted by a requester and calculates intersections with corresponding data in a database of the requester and the server, wherein the requester and the server encrypt the features after hashing the features respectively to prevent original data from being leaked in the calculation process, random vectors are generated through a VOLE technology, the requester and the server calculate intersections of matching features through linear combination of the random vectors, calculated intersection information is returned to the requester only in an encrypted form, and the requester uses a private key to decrypt encryption results to obtain final intersection features.
3. The method for processing privacy set in dynamic data scene according to claim 2, wherein the encryption matching specifically comprises the steps of inquiring data by a requester And encrypted shared random vector Combining, generating an encrypted query request Wherein, the method comprises the steps of, Is an encrypted query request sent by the requestor to the server, Is a random vector generated by the requesting party, Is query data, after receiving the query request, the server uses local data Encryption request to requestor And matching is carried out, whether query data are matched or not is judged, the server encrypts and compresses the matching result and sends the result back to the requester, and the requester uses the private key to decrypt to obtain a final result.
4. The method for processing privacy set in dynamic data scene as set forth in claim 1, wherein the shared random vector is generated by randomly generating a shared matrix Constructing a matrix of possible output associations by a requestor Matrix associated by possible outputs of server And (2) and Wherein , Extracting matrix by the requesting party Several rows in Composition and matrix Correlated submatrices And solving the random row vector by a linear equation solver Line vector Satisfy the following requirements The requester and the server perform VOLE operation and distribute vectors to the requester 、 Distributing vectors to servers And scalar quantity Wherein The request is sent to the server The server passes through an inadvertent pseudo-random function Computing key ; The method is characterized by comprising a random predictor, wherein a requester completes pseudo-random function solving through inner product linear operation.
5. The method for processing privacy set intersection in dynamic data scene according to claim 2, wherein the feature intersection calculation further comprises a step for implementing a privacy data segmentation process, specifically: for data interaction end set composed of requesting party and server Defining data interaction end Corresponding private data set ; Refers to a private data set The first of (3) Individual private data, then the private data intersection is represented as ; In the key initialization process, for each data transmitting terminal And a data receiving terminal Random key seed for combination allocation ; Using pseudo-random functions Determining index value of data interaction end Corresponding private data shares Expressed as Obtaining a random key , Refers to the sending of a request party And receiving the requesting party First on a structured key network A personal key fragment; In the communication process, a transmitting end generates a control coefficient And a prime number much greater than 1 And send it to receiving end, the sending end calculates And send to the receiving end, wherein Is a random number and Receiving end calculation And send to the sending end, wherein Is a random number and Transmitting end calculation Obtaining a session key Receiving end calculation Obtaining a session key Establishing an encryption channel through the session key; For each private data set Private data fragments in (a) The sender calculates the pseudo-random value corresponding to the privacy data segment through the session key, and the pseudo-random value is spliced to obtain the whole data, specifically, for the sender Calculation and receiving terminal Shared first Personal key fragment Corresponding private data shares And splicing all the private data shares to obtain whole data.
6. The method for processing privacy set intersection in dynamic data scenario according to claim 2, wherein the result decryption further comprises a step for implementing a decentralized multi-party high-speed intersection process, specifically: For any two requesters a and b, the private data sets are defined as respectively And ; In the communication process, a server generates control coefficients And a prime number much greater than 1 And send it to all requesters; selecting random numbers with secrecy by requester a And Calculation of After which the private key is used For a pair of After encryption, a digital signature is obtained I=1, 2 and sends it to the requestor b; Requester b obtains digital signature After that, the public key sent by the requester a Verifying the digital signature, terminating transmission if the verification fails, and selecting a random number by the requester b if the verification is successful And , Thereafter using the private key For a pair of After encryption, a digital signature is obtained I=1, 2, and digitally sign , , Sending to the requester a; requester a obtains digital signature After that, the public key sent by the requester b Verifying the digital signature, and terminating transmission if the verification fails, and establishing connection if the verification succeeds; After the connection is successful, the requester a calculates , After that use Generating a set of random numbers Using For its private data set Performing pseudo-random function calculation to obtain And will To the server while the requestor b uses Generating a set of random numbers Random number set Is obtained after the sequence of (1) is disordered , After that use For its private data set Performing pseudo-random function calculation to obtain Will be Medium element as key value set selection Obtain data set from first M elements of (C) Will be The method comprises the steps of performing inadvertent key value storage coding and then sending to a server; the server obtains the data sent by the requester a and the requester b to Storing coded objects for inadvertent key values for transmission by requester a1 In (a) and (b) Inadvertent key value storage decoding for keys The decoding result is then Returning to the requester a; The requester a obtains the decoding result After that, calculate Calculation result At the same time as the private data sets are respectively And Is a complex of the two.

Description

Privacy set intersection processing method under dynamic data scene Technical Field The invention belongs to the technical field of private data sharing processing, and particularly relates to a private collection intersection processing method under a dynamic data scene. Background The privacy data intersection processing is widely applied to a plurality of fields such as academic integrity maintenance, data storage, anti-plagiarism detection, enterprise data duplication removal and the like, and has important practical significance. With the advent of the big data age, the data volume has been explosively increased, and the demand for privacy protection has become more remarkable. Therefore, the search has extremely important significance in both theoretical and practical application on the premise of ensuring the data security and realizing the high efficiency and the accuracy of intersection processing. The related technology or patent mainly comprises a 'FATE federal privacy set intersection method and system based on vector careless estimation', wherein in a federal learning framework FATE, vectors and scalar are generated through a vector careless estimation (VOLE) protocol, a careless pseudo-random function (OPRF) protocol is constructed to realize Privacy Set Intersection (PSI), the method is mainly used for solving the safety and efficiency problems in malicious scenes, the 'privacy request method based on unbalanced privacy set intersection' mainly aims at unbalanced data sets, target data sets are screened through specific bits of hash values, cuckoo hash and common hash buckets are combined, pseudo-random function values are transmitted in stages to realize privacy query, the scheme mainly aims at centralized processing of static data sets, and good application is difficult to be carried out in multiparty dynamic data scenes in dynamic environments. Disclosure of Invention The invention aims to provide a vector unintentional linear evaluation (VOLE) protocol-based efficient linear operation for batch data, which combines communication compression and active safety verification of Silent unintentional transmission (Silent OT), solves the efficiency bottleneck and safety risk of the traditional privacy aggregation intersection (PSI) technology in a dynamic data scene, can realize matching through unintentional transmission and privacy intersection, protects data privacy, supports large-scale data collection and re-search, reduces communication and calculation overhead and improves query efficiency. In order to achieve the above purpose, the present invention adopts the following technical scheme. A method for processing privacy set interaction under dynamic data scene comprises key initialization, query construction, privacy transmission and result decryption; key initialization, namely, a requester and a server respectively generate public and private keys required by careless transmission and encrypt all inquiry data by using the public keys, wherein the requester generates a key pair ,In order for the requester to be a public key,Generating a key pair for a requester private key by a serverWhereinIn order to be a server public key,The request party and the server encrypt the query data and the response data respectively, and establish a basic communication channel; query construction refers to the calculation of a query hash by a requestor based on local hash feature values: Constructing and encrypting the query request Then the requesting party sends the query setFeeding the server; The privacy transmission is that after the requester and the server realize the linear combination of privacy, the server and the requester are encrypted and matched, whether the query data are matched with the local data is judged, the intersection of the matching characteristics is calculated with the server, and the intersection of the matching characteristics is output; Result decryption, namely that the requester receives ciphertext data returned by the server And decrypting by using the shared information and the private key generated by the VOLE, and finally restoring the matched hash feature set. Further improving or optimizing the privacy set intersection processing method under the dynamic data scene, wherein the privacy transmission specifically comprises encryption matching and feature intersection calculation; the encryption matching means that the server and the requesting party encrypt the hash feature respectively, the requesting party encrypts the hash feature by using the public key of the requesting party, and the server encrypts the corresponding feature by using the public key of the requesting party; The feature intersection calculation comprises that a server processes encrypted hash features submitted by a requester and calculates intersections with corresponding data in a database of the requester and the server, wherein the requester and the server encrypt the features after hashing the