Financing guarantee data retrieval method based on cloud platform

By calculating the similarity between the search word feature vector and the keyword index vector on the cloud platform and generating the recommended degree of ciphertext file, the problem that ciphertext search technology on the cloud platform can only support single keyword retrieval, and multiple keywords are realized Fuzzy search ensures the security of financial data and the efficiency of retrieval services.

CN118964709AInactive Publication Date: 2024-11-15RIZHAO CITY FINANCE GUARANTEE CO LTD
View PDF 10 Cites 0 Cited by

Patent Information

Application Number
CN202411448515.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-10-17
Publication Date
2024-11-15
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

The existing ciphertext search technology can only support single keyword retrieval on the cloud platform, which cannot meet the complex and diverse needs of users. Most of them only allow precise keyword query and cannot handle keyword similarity or spelling errors. .

Method used

By receiving the search word feature vector group, the similarity between the search word feature vector and each keyword index vector in the database is calculated, the ciphertext file recommendation degree is generated, and the order and recommendation are sorted and recommended based on this, supporting multi-keyword fuzzy retrieval.

Benefits of technology

It significantly reduces the risk of data leakage and attacks, ensures the security of financial data to the greatest extent, and meets the complex and diverse search needs of users, and provides efficient and reliable search services.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118964709A_ABST
    Figure CN118964709A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of information retrieval, in particular to a financing guarantee data retrieval method based on a cloud platform, which is used for a server and comprises the following steps: receiving a retrieval word feature vector group; respectively calculating the similarity between the search word feature vector group and each keyword index vector in a database to obtain a plurality of similarities; calculating a plurality of ciphertext file recommendation degrees based on the similarity corresponding to each keyword index vector and each index value; and sending a ciphertext file retrieval result to the retrieval end sending the retrieval word feature vector group based on the plurality of ciphertext file recommendation degrees. According to the method, the data file and the retrieval keyword are only processed in a plaintext form at the client, and the server data is encrypted, so that the leakage risk is effectively reduced, and the financing guarantee data security is guaranteed.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of information retrieval, and in particular to a financing guarantee data retrieval method based on a cloud platform. Background Art

[0002] There are a large number of small and micro enterprises in my country, and the financing data generated every day is growing exponentially, which leads to the continuous expansion of data storage on the financing service center platform. To meet this challenge, the financing service center platform generally adopts distributed cloud storage technology.

[0003] Distributed cloud storage technology has become an ideal choice for storing massive financial data due to its advantages such as large storage capacity, high reliability, high scalability, high performance and data security. In particular, distributed cloud storage technology performs well for financial information that requires high confidentiality. However, encrypted data loses some of the characteristics of plaintext, which poses new challenges to data retrieval.

[0004] Although existing ciphertext retrieval technologies can ensure the accuracy of retrieval results while ensuring the efficiency and security of the retrieval process, these technologies currently only support single keyword retrieval. As the scale of data continues to expand, the single keyword retrieval method can no longer meet the increasingly complex and diverse needs of users. At the same time, most of the existing ciphertext retrieval technologies only allow precise keyword queries (i.e. Boolean searches), and the retrieval keywords submitted by users cannot include keyword similarity issues or character spelling issues. Therefore, there is an urgent need for a ciphertext retrieval that meets user needs for financing guarantee data on cloud platforms. Summary of the invention

[0005] In order to solve the above technical problems, the purpose of the present invention is to provide a financing guarantee data retrieval method based on a cloud platform to solve the above problems. In order to achieve the above purpose, the technical solution adopted by the present invention is as follows: In the first aspect, the present application provides a financing guarantee data retrieval method based on a cloud platform, which is used on a server side. The method includes: receiving a search term feature vector group, wherein the search term feature vector group is obtained by vectorizing and encrypting multiple search keywords; respectively calculating the similarity between the search term feature vector group and each keyword index vector in the database to obtain multiple similarities, wherein the keyword index vector is encrypted by the data upload end and uploaded to the server side and then stored in the database, and each keyword index vector corresponds to a ciphertext file; based on the similarity corresponding to each keyword index vector and each index value, multiple ciphertext file recommendation degrees are calculated, wherein the index value is calculated by the number of times an index in the keyword index vector appears in the database; based on the multiple ciphertext file recommendation degrees, the ciphertext file retrieval results are sent to the retrieval end that sent the search term feature vector group.

[0006] In combination with the first aspect, in a possible implementation method, the similarity between the search term feature vector group and each keyword index vector in the database is calculated respectively to obtain multiple similarities, including: separately counting the search term frequency, index term frequency and file frequency of each target feature vector, the search term frequency is the term frequency of the target feature vector in the search term feature vector group, the index term frequency is the term frequency of the target feature vector in the keyword index vector, the file frequency is the number of ciphertext files containing the target feature vector, and the target feature vector is one of the search term feature vector group; the similarity between the search term feature vector group and one of the keyword index vectors is calculated based on the search term frequency, index term frequency and file frequency of each target feature vector.

[0007] In combination with the first aspect, in a possible implementation method, a ciphertext file retrieval result is sent to a retrieval end of the search term feature vector group based on multiple ciphertext file recommendation degrees, including: screening and obtaining multiple ciphertext files according to the relationship between a preset sorting threshold and the ciphertext file recommendation degree; sorting the multiple ciphertext files according to the ciphertext file recommendation degree to obtain a first sorting result; in the first sorting result, sorting the ciphertext files with the same ciphertext file recommendation degree according to the ciphertext file creation time to obtain a second sorting result; and sending the ciphertext file directory sorted according to the second sorting result to the retrieval end.

[0008] In combination with the first aspect, in a possible implementation manner, the encryption method of the keyword index vector is symmetric encryption.

[0009] In the second aspect, the present application provides a financing guarantee data retrieval method based on a cloud platform, which is used for a data upload end, and the method includes: extracting keywords from a data file to obtain a keyword phrase vector; generating a plaintext keyword index structure based on the positional relationship of the keyword phrase vector in the data file; encrypting the plaintext keyword index structure and the data file to obtain a keyword index vector and a ciphertext file; sending the keyword index vector and the ciphertext file to a server end, and the keyword index vector is used to calculate the ciphertext file recommendation degree with a search term feature vector group, and the search term feature vector group is sent by the search end.

[0010] In combination with the second aspect, in a possible implementation method, the plaintext keyword index structure and the data file are encrypted to obtain a keyword index vector and a ciphertext file, including: generating a random number; generating an encryption key based on a reversible matrix encryption method and the random number; and encrypting the plaintext keyword index structure and the data file based on the encryption key to obtain a keyword index vector and a ciphertext file.

[0011] In combination with the second aspect, in a possible implementation, the method for calculating the recommendation degree of the ciphertext file includes: calculating the similarity between the search term feature vector group and the keyword index vector; calculating the recommendation degree of the ciphertext file based on the similarity corresponding to the keyword index vector and each index value, wherein the index value is calculated by the number of times an index in the keyword index vector appears in the database.

[0012] In the third aspect, the present application provides a financing guarantee data retrieval method based on a cloud platform, which is used for the retrieval end, and the method includes: obtaining retrieval keywords; preprocessing the retrieval keywords to obtain a retrieval term feature vector group; sending the retrieval term feature vector group to the server end, and the retrieval term feature vector group is used to trigger the server end to calculate the ciphertext file recommendation degree based on the keyword index vector and the retrieval term feature vector group, and the keyword index vector is encrypted by the data upload end and uploaded to the server end.

[0013] In combination with the second aspect, in a possible implementation, the search keyword is preprocessed to obtain a search word feature vector group, including: vectorizing the search keyword to obtain a search keyword vector; encrypting the search keyword vector to obtain a search word feature vector group.

[0014] In combination with the second aspect, in a possible implementation method, the server calculates the recommendation degree of the ciphertext file based on the keyword index vector and the search term feature vector group, including: calculating the similarity between the search term feature vector group and the keyword index vector; calculating the recommendation degree of the ciphertext file based on the similarity corresponding to the keyword index vector and each index value, wherein the index value is calculated by the number of times an index in the keyword index vector appears in the database.

[0015] The present invention has the following beneficial effects: The present invention records and processes all data file plaintexts and search keyword plaintexts only on the client during the entire search process, whether the data upload end is uploading ciphertext files or the search end is performing search operations. On the server side, all data exists in an encrypted form. The above method can significantly reduce the risk of data leakage and attacks, thereby ensuring the security of financial data to the greatest extent. At the same time, in the search process, the keyword constructs a keyword feature vector after encryption and sends it to the server side; after receiving the keyword feature vector, the server side calculates the ciphertext file recommendation degree between the keyword feature vector and the keyword index vector without performing a decryption operation, and finally sends the search to the client according to the ciphertext file recommendation degree. Therefore, in the above process, there is no need to follow the existing ciphertext search technology, which only allows precise keyword queries and single keyword search methods, which greatly meets the search needs. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0017] Figure 1 A flowchart of a financing guarantee data retrieval method based on a cloud platform provided in Embodiment 1 of the present invention; Figure 2 A schematic flow chart of step S2 provided in Example 1 of the present invention; Figure 3 A schematic flow chart of step S4 provided in Example 1 of the present invention; Figure 4 A flowchart of a financing guarantee data retrieval method based on a cloud platform provided in Embodiment 2 of the present invention; Figure 5 A schematic diagram of the process of step 3 provided in embodiment 2 of the present invention; Figure 6 A schematic diagram of the process of step 4 provided in embodiment 2 of the present invention; Figure 7 A schematic diagram of a process of a financing guarantee data retrieval method based on a cloud platform provided in Embodiment 3 of the present invention; Figure 8 A schematic flow chart of step A2 provided in Embodiment 3 of the present invention; Fig. 9This is a flow chart of step A3 provided in Embodiment 3 of the present invention. DETAILED DESCRIPTION

[0018] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the specific implementation method, structure, features and effects of a financing guarantee data retrieval method based on a cloud platform proposed by the present invention in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0019] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0020] Distributed cloud storage has the advantages of large storage capacity, high reliability, high scalability, high performance and data security, so it is a good data storage method for massive amounts of highly confidential financial file data. However, encrypted financial file data loses some of the characteristics of plain text, making it difficult to use data efficiently. In addition, in a cloud environment, users upload and share data, and financial file data such as financing guarantee data is highly unique, so users can only retrieve specific data files in specific environments.

[0021] Although existing ciphertext retrieval technologies can ensure the efficiency and security of the retrieval process, as well as the accuracy of the retrieval results, these solutions still have some shortcomings. First, existing ciphertext retrieval technologies only support single-keyword retrieval. As the scale of data continues to expand, limited keywords cannot accurately describe user needs, so ciphertext retrieval needs to support multi-keyword retrieval. Second, most existing ciphertext retrieval technologies only allow precise keyword queries (i.e. Boolean searches). The search keywords submitted by users cannot include keyword similarity issues or character spelling issues, which may lead to incorrect search results.

[0022] Embodiment 1: The specific scheme of the cloud platform-based financing guarantee data retrieval method for the server provided by the present invention is described in detail below with reference to the accompanying drawings.

[0023] See also Figure 1 , which shows a flow chart of a financing guarantee data retrieval method based on a cloud platform provided by an embodiment of the present invention, such as Figure 1 As shown in this embodiment, steps S1 to S4 are included.

[0024] S1. Receive a search term feature vector group, where the search term feature vector group is obtained by vectorizing and encrypting a plurality of search keywords.

[0025] In this embodiment, the search term feature vector group is sent from the search end to the server end of the cloud platform. The search end refers to the device used by the user when searching for information, such as a mobile phone, a computer, or a communication machine. These devices can communicate with the server end of the cloud platform to achieve information retrieval and processing. For those skilled in the art, it can be understood that the search end is actually a device authorized by the cloud platform, and these devices can effectively improve the security of the ciphertext files stored in the cloud service.

[0026] Among them, the search keywords mentioned may include but are not limited to "guarantor", "creditor", "financing amount" and other words related to financing guarantees. The keywords may be determined by the specific implementation scenario and no specific restrictions are made in this embodiment.

[0027] Specifically, in this embodiment, the generation process of the search term feature vector group is as follows: First, the search end will vectorize the search keywords entered by the user, and then encrypt the vectorized data. The vectorization process can be implemented by a variety of existing methods, such as the Bag of Words (BOW), GloVe (Global Vectors for Word Representation) or Bidirectional Encoder Representations from Transformers (BERT). In this embodiment, we do not make specific restrictions on these methods.

[0028] As for the encryption process, in this embodiment, you can choose to use symmetric encryption or asymmetric encryption methods. Considering that in this embodiment, the data upload end and the retrieval end need to share a key to complete the extraction of data retrieval results, in this embodiment, a symmetric encryption method is more likely to be used. It should be noted that in this embodiment, the data upload end refers to the owner of the ciphertext file, and the retrieval end is the user of the ciphertext file. Regarding the specific data processing method of the search term feature vector group, please refer to the content described in Example 3, which will not be repeated in this embodiment.

[0029] S2. Calculate the similarity between the search term feature vector group and each keyword index vector in the database respectively to obtain multiple similarities. The keyword index vector is encrypted by the data upload end and uploaded to the server end and then stored in the database. Each keyword index vector corresponds to a ciphertext file.

[0030] Among them, the keyword index vectors mentioned in this embodiment are obtained by extracting keywords from the ciphertext file using a keyword extraction algorithm to obtain keyword phrases, and then vectorizing and encrypting them. The keyword extraction algorithm can be a TF-IDF algorithm or an LDA model (Latent Dirichlet Allocation). In addition, the encryption mentioned in this step is the same as the encryption method mentioned in step S1, specifically, both are symmetric encryption.

[0031] In this embodiment, the keyword index vector involved is to first extract the keyword phrase from the ciphertext file through the keyword extraction algorithm, then vectorize these keywords, and convert the vectorized keywords into index vectors through the encryption algorithm. In this process, the keyword extraction algorithm can adopt a variety of different methods, such as: term frequency-inverse document frequency (TF-IDF) algorithm and latent Dirichlet Allocation (LDA) model. In this embodiment, the encryption method mentioned is the same as the encryption method mentioned in step S1. Specifically, the encryption method used by the keyword index vector is also symmetric encryption. Symmetric encryption is a common encryption technology that uses the same key to encrypt and decrypt data. It is a prior art and will not be repeated in this embodiment. In this way, the confidentiality and integrity of the keyword index vector during generation and use can be ensured, thereby providing a strong guarantee for the security and reliability of the entire retrieval process.

[0032] To clarify the similarity calculation method in this step, please refer to Figure 2 , Figure 2 The specific calculation process of similarity is shown in: S21. Count the search term frequency, index term frequency and file frequency of each target feature vector respectively, wherein the search term frequency is the term frequency of the target feature vector in the search term feature vector group, the index term frequency is the term frequency of the target feature vector in the keyword index vector, the file frequency is the number of ciphertext files containing the target feature vector, and the target feature vector is one of the search term feature vector group.

[0033] S22. Calculate the similarity between the search term feature vector group and one of the keyword index vectors according to the search term frequency, index term frequency and file frequency of each target feature vector.

[0034] Specifically, the similarity calculation formula is as follows: in, Represents the search term feature vector group With keyword index vector similarity; Indicates search keywords Total number of keywords in the Indicates keywords In the search term feature vector group Frequency of search terms in ; Indicates keywords In the keyword index vector The frequency of index terms in Not zero; Indicates that it contains keywords The number of ciphertext files, is not zero. It is understandable that when the keyword In the keyword index vector When the frequency of the index term in is zero, we can is 100 to meet different application environments; Indicates that it contains keywords When the number of ciphertext files is zero, The value is 100 to meet different application environments. Meanwhile, other values ​​greater than 10 may also be selected by those skilled in the art, and this is not limited in this embodiment.

[0035] In this embodiment, a keyword is considered In the search term feature vector group and keyword index vector The frequency of occurrence is close to The closer it is to 1, the more likely it is that the search term feature vector group and keyword index vector The more similar they are, the higher the match degree is, and if they contain keywords The fewer the number of ciphertext files, The larger the value, the more likely the keyword For the search term feature vector group and keyword index vector The higher the uniqueness of is, the higher the matching degree is. Therefore, in the above calculation formula, this step is based on Combined with the description of the search term feature vector group A keyword in The similarity is finally accumulated and summed to obtain the search term feature vector group With keyword index vector The similarity.

[0036] S3. Calculate multiple ciphertext file recommendation degrees based on the similarity corresponding to each of the keyword index vectors and each index value, where the index value is calculated based on the number of times one of the keyword index vectors appears in the database.

[0037] In this embodiment, it is considered that there may be multiple financing guarantee data of the same guarantor with different creditors in the financing guarantee data; similarly, the same creditor may have financing guarantee agreements with multiple different guarantors. That is, the keywords used in the search are likely to be a many-to-many relationship. Therefore, in this embodiment, a method is proposed to first calculate the ciphertext file The degree to which each index represents the ciphertext file itself, or the value of the ciphertext file index. Specifically, the calculation formula of the index value in this embodiment is as follows: in, Indicates a ciphertext file Index The index value of Indicates that the database contains indexes the number of files; is a natural constant; Indicates a ciphertext file The total number of indices in the keyword index vector.

[0038] In this embodiment, a ciphertext file is considered The value of all included indexes is the same, that is, the importance of each index is Then, use the database containing the index The number of files to represent the index For encrypted files That is, if the same index appears in many files If so, then this index The higher the versatility of , the weaker its ability to represent a single ciphertext file. Therefore, in this step, a negative exponential function is used. To express it. In turn, the purpose of accurately describing the value of each index is achieved.

[0039] The calculation formula for the recommendation degree of the ciphertext file is as follows: in, Indicates a ciphertext file and the search term feature vector group The recommendation degree of the ciphertext files between them; Indicates a ciphertext file The total number of indexes in the keyword index vector in ; Represents the search term feature vector group With ciphertext file Keyword index vector The similarity between Indicates a ciphertext file Index the value of Represents an index Belongs to the keyword index vector ; Represents the normalization function.

[0040] The above calculation formula can be divided into two parts; the first part Indicates a ciphertext file The average value of the index value in a ciphertext file When there are multiple indexes, only the matching degree of one of the indexes is used to describe the index and the entire ciphertext file. The reliability of the match is not high. At the same time, the ciphertext file Contains only a few indexes, then from the search term feature vector group From the perspective of the proportion of the entire index, it means that the search term feature vector group The higher the probability of matching with its index, the smaller the The greater the combination , the greater the match, the more likely it is to be used to describe the ciphertext file. and the search term feature vector group Then the second part of the similarity As a weight value, it can better reflect the ciphertext file and the search term feature vector group The mutual relationship between them.

[0041] In this embodiment, based on the search term feature vector group sent by the search end, this embodiment completes the calculation process of the ciphertext file recommendation degree for each ciphertext file. In other words, by calculating the ciphertext file recommendation degree, the goal of fuzzy retrieval is achieved. Furthermore, those skilled in the art can meet the search requirements of the search end based on a method combining a preset sorting threshold with the ciphertext file recommendation degree. For details, see step S4, which describes in detail how to combine the search term feature vector group with the ciphertext file recommendation degree of the preset sorting threshold to achieve the fuzzy retrieval process.

[0042] S4. Sending a ciphertext file retrieval result to a retrieval terminal that sends the retrieval term feature vector group based on the plurality of ciphertext file recommendation degrees.

[0043] To clarify the process of sending encrypted file retrieval results, please refer to Figure 3, the figure shows that this step also includes steps S41-S44.

[0044] S41. Filter and obtain a plurality of ciphertext files according to a relationship between a preset sorting threshold and the recommendation degree of the ciphertext files.

[0045] In the present embodiment, the preset sorting threshold is set to 0.8. However, for those skilled in the art, other values ​​may also be selected as the preset sorting threshold according to specific circumstances. The present embodiment does not impose restrictions on the specific value of the preset sorting threshold. Specifically, in the present embodiment, the recommendation degree of the ciphertext file is compared with the preset sorting threshold to screen out those ciphertext files whose recommendation degree is lower than the preset sorting threshold. This is intended to facilitate further sorting of these ciphertext files in subsequent steps. In addition, if the recommendation degree of all ciphertext files is lower than the preset sorting threshold, the server will take corresponding measures. These measures may include returning a prompt message that there is no search result, or prompting the user to re-enter the search term to ensure the correct execution of the search process. This type of processing method is a conventional technical means in this field and will not be repeated in this embodiment.

[0046] S42: Sort the plurality of ciphertext files according to the ciphertext file recommendation degrees to obtain a first sorting result.

[0047] S43. In the first sorting result, the ciphertext files with the same ciphertext file recommendation degree are sorted according to the creation time of the ciphertext files to obtain a second sorting result.

[0048] S44: Send the ciphertext file directory sorted according to the second sorting result to the search end.

[0049] In this embodiment, the above steps S41 to S44 only show one possible sorting and screening method for obtaining the ciphertext file directory. For those skilled in the art, the sorting and screening can also be performed according to other criteria, such as the size of the ciphertext file, the type of the uploaded object, etc. Therefore, this embodiment does not impose any restrictions on the specific method of sorting and screening. The technicians can flexibly select the most appropriate sorting and screening method according to the actual application scenario and requirements.

[0050] After completing the above steps, the retrieval process for financial data based on the server has been implemented in this embodiment. During the entire retrieval process, whether the data upload end is uploading the ciphertext file or the retrieval end is performing the retrieval operation, all the plaintext of the data file and the plaintext of the retrieval keyword are only recorded and processed on the client. On the server side, all data exists in encrypted form. This processing method can significantly reduce the risk of data leakage and attack, thereby ensuring the security of financial data to the greatest extent. Through the implementation based on the above method, the security of financial data in the entire retrieval process can be ensured, and at the same time, efficient and reliable retrieval services are provided to users.

[0051] Embodiment 2: See also Figure 4 , which shows a flow chart of a financing guarantee data retrieval method based on a cloud platform provided by an embodiment of the present invention, wherein this embodiment is used for data uploading end, such as Figure 4 As shown in this embodiment, steps Step 1 to Step 4 are included.

[0052] Step 1: Extract keywords from the data file to obtain keyword phrase vectors.

[0053] Specifically, there are many different methods for extracting keywords from data files, such as the term frequency-inverse document frequency (TF-IDF) algorithm and the Latent Dirichlet Allocation (LDA) model.

[0054] Step 2: Generate a plaintext keyword index structure based on the position relationship of the keyword phrase vector in the data file.

[0055] Specifically, in this embodiment, the plaintext keyword index structure can be constructed in the form of a dictionary, and the frequency of each keyword is added to the plaintext keyword index structure in the dictionary form. The specific construction process is the prior art, and the specific process will not be repeated in this embodiment.

[0056] At the same time, this embodiment also provides a method for extracting keywords based on a Bloom extractor and converting them into plaintext keyword indexes, which is as follows: Step 21, firstly vectorize the data file to obtain a one-dimensional vector , .

[0057] Among them, vectorization can be performed using methods such as the Bag of Words (BOW), the GloVe (Global Vectors for Word Representation) or the Bidirectional Encoder Representations from Transformers (BERT), which will not be described in detail in this embodiment.

[0058] Step 22, based on one-dimensional vector Building a Bloom filter , and used as a plaintext keyword index structure.

[0059] in, Indicates the length of the Bloom filter; Represents the number of hash functions, that is, the hash functions include . Represents a one-dimensional vector Any element in , ,pass The hash function maps to a length of The Bloom filter process is used to obtain the data file Keyword index structure .

[0060] Step 3: Encrypt the plaintext keyword index structure and the data file to obtain a keyword index vector and a ciphertext file.

[0061] For clarity on encryption methods, see Figure 5 , the figure shows an encryption method, which specifically includes steps Step31 to Step33.

[0062] Step 31. Generate random numbers.

[0063] Step 32: Generate an encryption key based on the reversible matrix encryption method and the random number .

[0064] in, represents a random vector generated by random numbers, and Represents the encryption key generated by the reversible matrix encryption method based on random numbers. , and The specific generation method is prior art, and the specific process will not be described in detail in this embodiment.

[0065] Step 33: Encrypt the plaintext keyword index structure and the data file based on the encryption key to obtain a keyword index vector and a ciphertext file.

[0066] Step 4: Send the keyword index vector and the ciphertext file to the server. The keyword index vector is used to calculate the ciphertext file recommendation degree with the search term feature vector group. The search term feature vector group is sent by the search end.

[0067] Specifically, to clarify the calculation method of the recommendation degree of the ciphertext file, see Figure 6 , as shown in the figure, step Step4 also includes step Step41 and step Step42.

[0068] Step 41, calculating the similarity between the search term feature vector group and the keyword index vector.

[0069] The similarity calculation method in this step may be specifically referred to step S22 of Embodiment 1, and will not be described in detail in this embodiment.

[0070] Step 42: The recommendation degree of the ciphertext file is calculated based on the similarity corresponding to the keyword index vector and each index value, wherein the index value is calculated by the number of times an index in the keyword index vector appears in the database.

[0071] The index value and the method for calculating the recommendation degree of the ciphertext file in this step may be specifically referred to step S23 of Embodiment 1, which will not be described in detail in this embodiment.

[0072] Embodiment 3: See also Figure 7 , which shows a flow chart of a financing guarantee data retrieval method based on a cloud platform provided by an embodiment of the present invention, wherein this embodiment is used at the retrieval end, such as Figure 7 As shown in this embodiment, steps A1 to A3 are included.

[0073] A1. Obtain search keywords.

[0074] It should be noted that the search keywords mentioned in this step are obtained by the user inputting a device authorized by the cloud platform.

[0075] A2. Preprocess the search keywords to obtain a search term feature vector group.

[0076] Specifically, see Figure 8 In this embodiment, the preprocessing process specifically further includes step A21 and step A22.

[0077] A21. Perform vectorization processing on the search keyword to obtain a search keyword vector.

[0078] In this step, vectorization can be implemented by a variety of existing methods, such as Bag of Words (BOW), GloVe (Global Vectors for Word Representation), or Bidirectional Encoder Representations from Transformers (BERT). In this embodiment, we do not make specific restrictions on these methods.

[0079] A22. Encrypt the search keyword vector to obtain a search word feature vector group.

[0080] The encryption mentioned in this step is symmetric encryption, and the encryption key used by the retrieval end is the same as the encryption key used by the data upload end, both of which are generated by a reversible matrix encryption algorithm. The specific expression of the search term feature vector group is , where the reversible matrix in the encryption key generated by the reversible matrix is The transposed matrix of Represents the vectorized search keywords.

[0081] A3. Send the search term feature vector group to the server, wherein the search term feature vector group is used to trigger the server to calculate the ciphertext file recommendation degree according to the keyword index vector and the search term feature vector group, wherein the keyword index vector is encrypted by the data upload end and uploaded to the server.

[0082] Specifically, see Fig. 9 In this embodiment, the preprocessing process specifically further includes step A31 and step A32.

[0083] A31. Calculate the similarity between the search term feature vector group and the keyword index vector.

[0084] The similarity calculation method in this step may be specifically referred to step S22 of Embodiment 1, and will not be described in detail in this embodiment.

[0085] A32. The recommendation degree of the ciphertext file is calculated based on the similarity corresponding to the keyword index vector and each index value, wherein the index value is calculated based on the number of times an index in the keyword index vector appears in the database.

[0086] The index value and the method for calculating the recommendation degree of the ciphertext file in this step may be specifically referred to step S23 of Embodiment 1, which will not be described in detail in this embodiment.

[0087] It should be noted that the sequence of the above embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0088] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

Claims

1. A financing guarantee data retrieval method based on a cloud platform, characterized in that: For the server, the method includes: receiving a search term feature vector group, wherein the search term feature vector group is obtained by vectorizing and encrypting a plurality of search keywords; Calculating the similarity between the search term feature vector group and each keyword index vector in the database respectively to obtain multiple similarities, wherein the keyword index vector is encrypted by the data upload end and uploaded to the server end and then stored in the database, and each keyword index vector corresponds to a ciphertext file; Calculate multiple ciphertext file recommendation degrees based on the similarity corresponding to each of the keyword index vectors and each index value, wherein the index value is calculated based on the number of times an index in the keyword index vector appears in the database; The retrieval result of the ciphertext file is sent to the retrieval end that sent the retrieval term feature vector group based on the plurality of ciphertext file recommendation degrees.

2. The cloud platform-based financing guarantee data retrieval method according to claim 1 is characterized in that: The similarity between the search term feature vector group and each keyword index vector in the database is calculated respectively to obtain multiple similarities, including: Count the search term frequency, index term frequency and file frequency of each target feature vector respectively, wherein the search term frequency is the term frequency of the target feature vector in the search term feature vector group, the index term frequency is the term frequency of the target feature vector in the keyword index vector, and the file frequency is the number of ciphertext files containing the target feature vector, and the target feature vector is one of the search term feature vector group; The similarity between the search term feature vector group and one of the keyword index vectors is calculated based on the search term frequency, index term frequency and file frequency of each target feature vector.

3. The cloud platform-based financing guarantee data retrieval method according to claim 1 is characterized in that: Sending a ciphertext file retrieval result to a retrieval end of the retrieval term feature vector group based on the plurality of ciphertext file recommendation degrees includes: Screening and obtaining a plurality of ciphertext files according to a relationship between a preset sorting threshold and a recommendation degree of the ciphertext file; Sorting the plurality of ciphertext files according to the ciphertext file recommendation degrees to obtain a first sorting result; In the first sorting result, the ciphertext files with the same ciphertext file recommendation degree are sorted according to the creation time of the ciphertext files to obtain a second sorting result; The ciphertext file directory sorted according to the second sorting result is sent to the search end.

4. The cloud platform-based financing guarantee data retrieval method according to claim 1, characterized in that: The encryption method of the keyword index vector is symmetric encryption.

5. A financing guarantee data retrieval method based on a cloud platform, characterized in that: For data upload, the method includes: Extract keywords from data files to obtain keyword phrase vectors; Generate a plaintext keyword index structure according to the positional relationship of the keyword phrase vector in the data file; Encrypting the plaintext keyword index structure and the data file to obtain a keyword index vector and a ciphertext file; The keyword index vector and the ciphertext file are sent to the server. The keyword index vector is used to calculate the ciphertext file recommendation degree with the search term feature vector group. The search term feature vector group is sent by the search end.

6. The cloud platform-based financing guarantee data retrieval method according to claim 5 is characterized in that: Encrypting the plaintext keyword index structure and the data file to obtain a keyword index vector and a ciphertext file includes: Generate random numbers; Generate an encryption key based on a reversible matrix encryption method and the random number; The plaintext keyword index structure and the data file are encrypted based on the encryption key to obtain a keyword index vector and a ciphertext file.

7. The cloud platform-based financing guarantee data retrieval method according to claim 5, characterized in that: The method for calculating the recommendation degree of the ciphertext file includes: Calculating the similarity between the search term feature vector group and the keyword index vector; The recommendation degree of the encrypted file is calculated based on the similarity corresponding to the keyword index vector and each index value, and the index value is calculated by the number of times an index in the keyword index vector appears in the database.

8. A financing guarantee data retrieval method based on a cloud platform, characterized in that: For a search end, the method includes: Get search keywords; Preprocessing the search keywords to obtain a search term feature vector group; The search term feature vector group is sent to the server, and the search term feature vector group is used to trigger the server to calculate the ciphertext file recommendation degree according to the keyword index vector and the search term feature vector group, and the keyword index vector is obtained by uploading the keyword index vector to the server after being encrypted by the data upload end.

9. The cloud platform-based financing guarantee data retrieval method according to claim 8, characterized in that: Preprocess the search keywords to obtain a search term feature vector group, including: Performing vectorization processing on the search keyword to obtain a search keyword vector; The search keyword vector is encrypted to obtain a search word feature vector group.

10. The cloud platform-based financing guarantee data retrieval method according to claim 8, characterized in that: The server calculates the recommendation degree of the ciphertext file according to the keyword index vector and the search term feature vector group, including: Calculating the similarity between the search term feature vector group and the keyword index vector; The recommendation degree of the encrypted file is calculated based on the similarity corresponding to the keyword index vector and each index value, and the index value is calculated by the number of times an index in the keyword index vector appears in the database.

Citation Information

Patent Citations

  • Word sense based local file searching method

    CN101059806A

  • Multi-keyword plaintext and ciphertext retrieving method and device oriented to cloud storage

    CN103593476A

  • Keyword based evaluation expert intelligent search and recommendation method

    CN103605665A

  • Fuzzy multi-keyword retrieval method of encrypted data in cloud environment

    CN106326360A

  • Lightweight outsourcing file multi-keyword retrieval method and system with privacy protection function

    CN110727951A