API sensitive data identification method and system based on multiple DBSCAN and Apriori algorithms

Abstract

The invention discloses an API sensitive data identification method and system based on multiple DBSCAN and Apriori algorithms, and relates to the technical field of API sensitive data identification, and the method comprises the steps: collecting data, and constructing an API sensitive data set based on data sequence correlation; constructing an API sensitivity quantification model according to the data set; an API sensitivity clustering model based on multiple DBSCAN is constructed; and constructing an API sensitive attribute automatic identification model based on an Apriori algorithm to carry out sensitive data identification. The API sensitive data identification method based on the multiple DBSCAN and Apriori algorithms has the advantages that API sensitive data identification can be effectively achieved, a privacy protection method can be continuously improved in the technical aspect, and the privacy leakage degree is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of API sensitive data identification, and specifically to an API sensitive data identification method and system using multiple DBSCAN and Apriori algorithms. Background Art

[0002] With the deepening of the digital transformation of the power grid and the gradual deepening of business cloudification, the number of application programming interfaces (APIs) continues to grow, and the amount of data transmitted through APIs is also growing rapidly. API sensitive data identification can ensure the continuous monitoring of power grid privacy data, commercial data and other sensitive data, effectively protect the data security of power grid companies, and promote the safe and reliable development of the power grid.

[0003] Information technology has provided strong support for the development of modern society. The vigorous development of big data has brought a new starting point for the digital transformation of the modern power industry. Massive data is widely used in all aspects of power production. With the collection of various data platforms and information systems, more and more sensitive data related to power grid privacy information are gathered. While it brings convenience to power grid production, it is also easy to involve privacy leakage. The privacy of the power grid faces huge risks. Moreover, with the continuous improvement of data analysis technology, the mining of data value is becoming more and more favorable. It is easy to analyze sensitive information from some seemingly safe data. In order to better protect personal privacy information, countries around the world have introduced many laws, regulations and policies on information security protection. However, simply enforcing legal regulations is not enough to effectively protect the privacy security of the power grid. It should also continuously improve privacy protection methods in terms of technology to reduce the degree of privacy leakage. Therefore, it is of great theoretical significance and practical value to identify sensitive data of the power grid as one of the core technologies for studying power grid security. Summary of the invention

[0004] In view of the above-mentioned problems, the present invention is proposed.

[0005] Therefore, the technical problem solved by the present invention is: how to improve the privacy protection method and reduce the degree of privacy leakage.

[0006] To solve the above technical problems, the present invention provides the following technical solutions: a method for identifying API sensitive data using multiple DBSCAN and Apriori algorithms, comprising:

[0007] Collect data to build API sensitive datasets based on data sequence correlation;

[0008] Build a quantitative model of API sensitivity based on the data set;

[0009] Construct an API sensitivity clustering model based on multiple DBSCAN;

[0010] Build an automatic identification model of API sensitive attributes based on the Apriori algorithm to identify sensitive data.

[0011] As a preferred solution of the API sensitive data identification method of multiple DBSCAN and Apriori algorithms described in the present invention, wherein: the API sensitive data set based on data sequence correlation includes selecting API sensitive information data sequence and averaging each data sequence element group layer;

[0012] Calculate the correlation coefficient of API sensitive information data sequence and establish API sensitive data set.

[0013] As a preferred solution of the API sensitive data identification method of multiple DBSCAN and Apriori algorithms described in the present invention, wherein: the construction of the API sensitivity quantification model includes, the API sensitivity quantification indicators include information entropy and maximum discrete entropy;

[0014] Using information entropy and maximum discrete entropy as quantitative indicators, the attribute sensitivity of the API dataset is calculated and the data sensitivity is evaluated.

[0015] As a preferred solution of the API sensitive data identification method of multiple DBSCAN and Apriori algorithms described in the present invention, wherein: the construction of the API sensitivity clustering model based on multiple DBSCAN includes: the API sensitivity clustering samples include a sensitive attribute set and a suspected sensitive attribute set; the API sensitive data set is subjected to dimensionality reduction processing to obtain two-dimensional data points representing API sensitive information;

[0016] The maximum global parameter radius is input to execute the DBSCAN algorithm on the data set multiple times to extract data point information, and the final clustering results are the sensitive attribute set and the suspected sensitive attribute set.

[0017] As a preferred solution of the API sensitive data identification method of multiple DBSCAN and Apriori algorithms described in the present invention, wherein: the construction of an automatic identification model for API sensitive attributes based on the Apriori algorithm includes mining the association relationship between suspected sensitive attributes and sensitive attributes through the Apriori algorithm, using support and confidence to evaluate the association relationship, and accurately identifying API sensitive data.

[0018] As a preferred solution of the API sensitive data identification method of multiple DBSCAN and Apriori algorithm described in the present invention, wherein: the construction of the API sensitivity clustering model based on multiple DBSCAN also includes randomly selecting k data points from the data set P = {x1, x1, ..., xm} to form initial k mean vectors, recorded as {μ1, μ2, ..., μ k};

[0019] Input the value of the parameter Minpts, draw the k-distance curve, find all the effective distance mutation points in the graph, and determine the values ​​of m Eps;

[0020] Enter the maximum parameter Eps to execute the first DBSCAN algorithm on the data set P; the clustering calculation with the maximum parameter Eps will remove the overly discrete points in the data set P as noise points, and merge some point clusters with close distances; the clusters after clustering with the maximum parameter Eps include both individually identified point clusters and unidentified point clusters that are merged;

[0021] Input the remaining parameters Eps to execute the i-th (i=2,3,…,m-1) DBSCAN algorithm; the clustering calculation with a smaller parameter Eps will re-identify the clusters that were not identified and merged in the previous step, and further extract the feature information of the data points;

[0022] The clustering results of each time are combined to obtain the final clustering result, which is divided into the sensitive attribute set SA and the suspected sensitive attribute set PSA.

[0023] As a preferred solution of the API sensitive data identification method of multiple DBSCAN and Apriori algorithms described in the present invention, the method includes: using an association rule algorithm to mine strong association rules between attributes, optimizing the identification effect of sensitive attributes, and realizing automatic identification of API sensitive attributes;

[0024] Define association rules. If the set of transaction items in the database is L = {l1,l2,…,l m}, the association rule formula is expressed as:

[0025]

[0026] in, And x1 and x2 represent two disjoint subsets of transaction items, x1∩x2=φ; Apriori algorithm is used to mine the association relationship between sensitive attribute set SA and suspected sensitive attribute set PSA, and the association rule expression is used To mark each association rule, the classification processing of suspected sensitive attribute sets is realized according to the number of successfully established association rules between two attribute sets. The attributes in the suspected sensitive attribute set that have a strong correlation with the sensitive attributes are accurately divided into the sensitive attribute set, and the data set is divided into the sensitive attribute set SA and the non-sensitive attribute set NSA.

[0027] Another object of the present invention is to provide an API sensitive data identification system using multiple DBSCAN and Apriori algorithms, which can solve the problem of how to improve the privacy protection method in current technology through one of the solutions.

[0028] As a preferred solution of the API sensitive data identification system of multiple DBSCAN and Apriori algorithms described in the present invention, it includes: a first construction module for constructing an API sensitive data set based on data sequence correlation;

[0029] The second building module is used to build an API sensitivity quantification model, wherein the API sensitivity quantification indicators include information entropy and maximum discrete entropy;

[0030] The third construction module is used to construct an API sensitivity clustering model based on multiple DBSCAN, wherein the API sensitivity clustering samples include a sensitive attribute set and a suspected sensitive attribute set;

[0031] The fourth building block is used to build an automatic identification model for API sensitive attributes based on the Apriori algorithm.

[0032] A computer device includes a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of a method for identifying API sensitive data using multiple DBSCAN and Apriori algorithms.

[0033] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of a method for identifying API sensitive data using multiple DBSCAN and Apriori algorithms.

[0034] Beneficial effects of the present invention: The API sensitive data identification method of multiple DBSCAN and Apriori algorithms provided by the present invention can effectively realize API sensitive data identification, help to continuously improve the privacy protection method in terms of technology, and reduce the degree of privacy leakage. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0036] Figure 1 An overall flow chart of a method for identifying API sensitive data using multiple DBSCAN and Apriori algorithms provided in the first embodiment of the present invention.

[0037] Figure 2 A specific step diagram of an API sensitive attribute automatic identification model based on the Apriori algorithm in a method for identifying API sensitive data using multiple DBSCAN and Apriori algorithms provided in the second embodiment of the present invention. DETAILED DESCRIPTION

[0038] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0039] Example 1, reference Figure 1-2 , as an embodiment of the present invention, provides a method for identifying API sensitive data using multiple DBSCAN and Apriori algorithms, including:

[0040] S1: Collect data to build an API sensitive dataset based on data sequence correlation.

[0041] Furthermore, the API sensitive data set based on data sequence correlation includes selecting API sensitive information data sequences and averaging each data sequence element group layer.

[0042] Furthermore, the correlation coefficient of the API sensitive information data sequence is calculated to establish an API sensitive data set.

[0043] Furthermore, the present invention uses a data sequence correlation algorithm to extract sensitive information from the massive data of the power grid API, and designs a data sequence that meets the correlation of sensitive information. Taking the extracted API image information as an example, to extract sensitive information from the image, it is first necessary to select the data sequence of sensitive information. After an image undergoes a two-dimensional discrete Fourier transform, the image will be decomposed into multiple sub-images. Most of the image information is concentrated in the near sub-image of the image.

[0044] Furthermore, at the corresponding point position of the image, the probability coefficient of the existence of sensitive information in the close sub-image is larger, and the probability coefficient of the existence of sensitive information in the detail sub-images in three different directions is smaller and closer to the 0 component. The corresponding point coefficients between the detail sub-images in three different directions are close to each other, and the three element groups of each data sequence are averaged. The correlation between different data sequences is measured using the correlation coefficient, and its calculation formula is:

[0045]

[0046] In the above formula, K represents the correlation coefficient; i represents the length of the data sequence; E and F represent two different data sequences respectively; Represents the mean mean(E) of the data sequence E; Represents the mean value mean(F) of the data sequence F. The correlation coefficient K of the two data sequences is obtained by calculation, and the correlation coefficient is used to represent the linear relationship statistics of the two different data sequences. When extracting sensitive information, the value of the correlation coefficient K is between [-1,1].

[0047] It should be noted that the above calculations show that the correlation between two different data sequences is additively invariant and multiplicatively invariant, and the spatial scaling of the data sequence will not affect the correlation of the data sequence itself. Based on this feature, the sensitive information data sequence is used to extract API sensitive information by extracting the data sequence, thereby constructing an API sensitive data set.

[0048] S2: Build an API sensitivity quantification model based on the dataset.

[0049] Furthermore, constructing an API sensitivity quantification model includes that API sensitivity quantification indicators include information entropy and maximum discrete entropy.

[0050] Furthermore, information entropy and maximum discrete entropy are used as quantitative indicators to calculate the attribute sensitivity of the API dataset and evaluate the data sensitivity.

[0051] Furthermore, a quantitative model of API sensitivity is constructed. The present invention is oriented to a massive API sensitive data set. The data records are processed after uniform sampling, and an API sensitivity quantitative model based on information entropy is proposed to improve recognition efficiency. The mathematical expectation of the self-information of each discrete message of the source (i.e., the probability-weighted statistical semi-mean) is the semi-mean information of the source, which is called the information entropy of the source, also known as source entropy or Shannon entropy, referred to as entropy, denoted as H(X). Its calculation formula is as follows:

[0052]

[0053] In the formula, 0≤p(a i )≤1, Represents the probability of each discrete message occurring. The maximum discrete entropy theorem states that when the source X contains n different discrete messages, the source entropy H(X) has H(X)≤log2n, and H(X)=log2n if and only if the probability of each message in X occurring is equal. According to the definition of the entropy of the attribute in the original data set and its maximum discrete entropy, the attribute sensitivity can be expressed as:

[0054]

[0055] Furthermore, the value range of attribute sensitivity is SVi ∈(0,1), the smaller the value, the more sensitive the attribute. If the attribute entropy value is larger, the closer the distance between it and the maximum discrete entropy is, the smaller the sensitivity value is, indicating that the attribute is more sensitive, otherwise, it is insensitive. Attribute entropy is an indicator to measure the degree of disorder of the attribute. The larger the entropy, the more disordered the attribute is.

[0056] It should be noted that the maximum discrete entropy is the evaluation index with the greatest attribute uncertainty. If such attributes are captured by malicious attackers, the value of obtaining the attributes is very high, indicating that such attributes are extremely sensitive. The sensitivity of the attribute can be determined based on the distance between the attribute entropy and the maximum discrete entropy. The closer the distance between them, the more sensitive the attribute is, and vice versa. Usually, all data records of the attributes in the data set are unique, so the probability of their values ​​appearing is equal, and the uncertainty is high. The larger the entropy of such attributes, the smaller the distance between them and the maximum discrete entropy, and the more sensitive the attribute is.

[0057] S3: Build an API sensitivity clustering model based on multiple DBSCAN.

[0058] Furthermore, an API sensitivity clustering model based on multiple DBSCAN is constructed, including that the API sensitivity clustering samples include a sensitive attribute set and a suspected sensitive attribute set; and the API sensitive data set is subjected to dimensionality reduction processing to obtain two-dimensional data points representing API sensitive information.

[0059] Furthermore, the maximum global parameter radius is input to execute the DBSCAN algorithm on the data set multiple times to extract the data point information, and the final clustering results are obtained as a sensitive attribute set and a suspected sensitive attribute set.

[0060] Furthermore, an API sensitivity clustering model based on multiple DBSCANs is constructed. Cluster analysis, as a common unsupervised learning task in machine learning, aims to reveal the intrinsic connections between data variables in unknown sample sets, laying a good foundation for accurate data analysis.

[0061] Furthermore, the API sensitivity clustering process based on multiple DBSCANs proposed in the present invention is as follows:

[0062] Step 1: Randomly select k data points from the data set P = {x1, x1, ..., xm} to form the initial k mean vectors, denoted as {μ1, μ2, ..., μ k}.

[0063] Step 2: Input the value of the parameter Minpts, draw the k-distance curve, find all the effective distance mutation points in the graph, and determine the values ​​of m Eps.

[0064] Step 3, input the maximum parameter Eps to perform the first DBSCAN algorithm on the data set P. The clustering calculation with the maximum parameter Eps will remove the overly discrete points in the data set P as noise points, and merge some point clusters with close distances. The clusters after the maximum parameter Eps clustering include both individually identified point clusters and unidentified point clusters that are merged.

[0065] Step 4: Input the remaining parameters Eps to execute the i-th (i=2, 3, ..., m-1) DBSCAN algorithm. The clustering calculation with a smaller parameter Eps will re-identify the clusters that were not identified and merged in the previous step, and further extract the feature information of the data points.

[0066] Step 5: Merge the clustering results of each time to obtain the final clustering result, which is divided into a sensitive attribute set SA and a suspected sensitive attribute set PSA.

[0067] It should be noted that the density-based spatial clustering of applications with noise (DBSCAN) algorithm is a typical density-based clustering algorithm. It uses two global parameters, radius (Eps) and the minimum number of samples in the cluster (Minpts), to identify sample points with higher density in the data set and divide them into clusters. The algorithm first marks the samples in the data set as unprocessed objects, calculates the set of points with a radius of Eps for each sample point, and marks the points with more than or equal to Minpts in the set as core points. The remaining points are marked as boundary points if they are in the core point area, and continue to be added to the cluster until there are no points that meet the conditions. If a point in the data does not meet the requirements of either the core point or the boundary point, the point is marked as a noise point. The DBSCAN algorithm clusters all the points in the simulation data by repeating the above process. The advantage of this algorithm is that the number of clusters can be directly obtained based on fixed parameters. Unlike clustering algorithms based on distance functions, the extension and expansion characteristics of the DBSCAN algorithm during calculation can identify clusters of arbitrary shapes in noisy data.

[0068] S4: Build an automated API sensitive attribute recognition model based on the Apriori algorithm to identify sensitive data.

[0069] Furthermore, an automatic identification model of API sensitive attributes based on the Apriori algorithm is constructed. In order to prevent attackers from probabilistically inferring sensitive attribute information through suspected sensitive attribute information to a certain extent, the present invention further considers the association relationship between suspected sensitive attributes and sensitive attributes, and further uses the association rule algorithm to mine strong association rules between attributes, optimize the identification effect of sensitive attributes, and realize automatic identification of API sensitive attributes. Association rules refer to that if the set of transaction items in the database is L = {l1,l2,…,l m}, the association rules between them are available This kind of implication is expressed in the form of And x1 and x2 are two disjoint subsets of transaction items, that is: x1∩x2=φ.

[0070] Furthermore, the Apriori algorithm is used to mine the association between the sensitive attribute set SA and the suspected sensitive attribute set PSA, and the association rule expression is used To mark each association rule, the classification processing of suspected sensitive attribute sets is realized according to the number of successfully established association rules between two attribute sets. The attributes in the suspected sensitive attribute set that have a strong correlation with the sensitive attributes are accurately divided into the sensitive attribute set, and the data set is divided into the sensitive attribute set SA and the non-sensitive attribute set NSA.

[0071] Furthermore, assuming that in the transaction database formed by the original data set, the association relationship between the suspected sensitive attribute PSAi and the sensitive attribute SAj is mined through the Apriori algorithm, and after filtering the association rules that do not meet the requirements, the association rule between the suspected sensitive attribute set PSA and the sensitive attribute set SA is obtained as follows:

[0072] Set the confidence level to Confidence = 0.7. According to the above association rules, if the number of association rules that can be established between the suspected sensitive attribute PSA1 and the sensitive attribute accounts for 1 / 3 of the total number of association rules, it means that the probability of inferring the sensitive attribute through the suspected sensitive attribute PSA1 is high, and it is classified as a sensitive attribute.

[0073] Furthermore, if there are multiple association rules between attributes, such as: Support and confidence are the key factors to measure the strength of association rules. Support is an indicator to measure the importance of association rules, and confidence is an indicator to measure the accuracy of association rules. Observe the size of support and confidence in multiple existing association rules, and take the association rule with high support and confidence as the first choice. Use this association rule to calculate the conditional change entropy of attribute x2. This association rule indicates that there is a strong correlation with attribute x2.

[0074] Furthermore, according to the above definition, if the following association rule can be established between the suspected sensitive attribute and the sensitive attribute SA1: Assume that there is It can be concluded that there is a strong correlation between the suspected sensitive attribute PSA1 and the sensitive attribute SA1.

[0075] Furthermore, assume that an association rule can be established between attribute x1 and attribute x2, that is After the association rules are established, the sensitivity of the attribute may change due to the influence of the association relationship. The calculation formula for the conditional change entropy of attribute x1 is as follows:

[0076] H Z (x1) = H(x1) + H(x1 / x2)

[0077] In the formula, H(x1) represents the original attribute entropy of attribute x1, and H(x1 / x2) represents the conditional entropy after the attribute x1 and attribute x2 establish an association rule. Finally, the sensitivity of attribute x1 after being affected by the association relationship can be expressed as:

[0078]

[0079] It should be noted that if in the above association rules, association rules can be established between sensitive attribute SA1 and attributes PSA1, PSA2, and PSA3 in the suspected sensitive attribute set, then SA1 is no longer an independent attribute, and its sensitivity will also be affected by the association relationship. In summary, the specific steps of the API sensitive attribute automatic identification model based on the Apriori algorithm can be designed as follows: Figure 2 Furthermore, the construction of an automatic identification model for API sensitive attributes based on the Apriori algorithm includes mining the association between suspected sensitive attributes and sensitive attributes through the Apriori algorithm, using support and confidence to evaluate the association, and accurately identifying API sensitive data.

[0080] Example 2: The following is an embodiment of the present invention, which provides an API sensitive data identification method using multiple DBSCAN and Apriori algorithms. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiments.

[0081] This example aims to verify the innovation and advantages of the API sensitive data identification method based on data sequence correlation. The experimental object is the API call log of a certain enterprise, which contains a large amount of API request data interacting with users. The experimental data set collects relevant information of different API calls, establishes a quantitative model for evaluating API sensitivity, and further identifies sensitive data through multiple DBSCAN clustering and Apriori algorithm.

[0082] Data Collection and Preprocessing,The experimental dataset is extracted from an enterprise system containing millions of,API call logs.,The data fields include API name, call parameters, call time, user identity information,,call results, etc. First, the data was cleaned, including removing duplicate data, formatting data,,denoising, etc., to ensure the integrity and consistency of the input data.

[0083] API sensitivity quantitative model construction,In order to accurately evaluate the sensitivity of each API call, this experiment built an API sensitivity quantitative model.,The measurement model is based on multiple factors, such as the frequency of API calls, the type of sensitive information involved in the call parameters (such as,ID card number, bank card number, etc.), the sensitivity of the call results (such as transaction success or failure), and assigns each API call a sensitivity score.

[0084] Based on multiple DBSCAN API sensitivity clustering, the Apriori algorithm is used to mine association rules between API calls based on the DBSCAN clustering results to automatically identify sensitive data. Through the mining of frequent itemsets, the association relationship between APIs is revealed, especially the pattern of sensitive information calls. The association rules mined by the Apriori algorithm provide a reliable basis for subsequent sensitive data identification.

[0085] The experiment was conducted in a server environment with a 32-core CPU and 128GB memory, ensuring the performance requirements for large-scale API log data processing. The experiment used Python and related libraries (such as SciPy and MLxtend) to implement DBSCAN clustering and Apriori algorithms.

[0086] Table 1 Experimental data table

[0087]

[0088] It can be seen from the experimental data that the API sensitive data identification method of the present invention has significant advantages in many aspects. First, the sensitivity score of each API call is quantified by constructing an API sensitivity quantification model. This process not only improves the operability of the data, but also provides an important basis for subsequent clustering and association rule mining. From the data in the table, it can be found that the sensitivity scores of API1, API2, API4 and API5 are higher, while the sensitivity scores of API3 and API6 are lower, which is consistent with their performance in DBSCAN clustering and Apriori association rules.

[0089] Secondly, DBSCAN clustering effectively divides API calls into different clusters according to their sensitivity. API1, API2, API4 and API5 are divided into the same cluster, indicating that they have similar sensitivity, while API3 and API6 are divided into another cluster. The rationality of this clustering was verified in the subsequent Apriori association rule mining. Through the Apriori algorithm, it was found that there is a frequent calling relationship between API1, API2 and API4, which further supports the sensitive data involved in these API calls. Through sensitive attribute identification, the method of the present invention can automatically identify which API calls involve sensitive data. The table shows that API1, API2, API4 and API5 are all identified as APIs involving sensitive data, while API3 and API6 have no identified sensitive attributes. This shows that the method of the present invention can accurately identify sensitive data in API calls with an extremely low false negative rate.

[0090] Example 3, reference Figure 1 , 2 , as an embodiment of the present invention, provides an API sensitive data identification system using multiple DBSCAN and Apriori algorithms, including: a first building module for building an API sensitive data set based on data sequence correlation.

[0091] The second building module is used to build an API sensitivity quantification model, wherein the API sensitivity quantification indicators include information entropy and maximum discrete entropy.

[0092] The third construction module is used to construct an API sensitivity clustering model based on multiple DBSCANs, wherein the API sensitivity clustering samples include a sensitive attribute set and a suspected sensitive attribute set.

[0093] The fourth building block is used to build an automatic identification model for API sensitive attributes based on the Apriori algorithm.

[0094] If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods of each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0095] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0096] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering, or processing in another suitable manner as necessary, and then stored in a computer memory.

[0097] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc. It should be noted that the above embodiments are only used to illustrate the technical solution of the present invention and are not limited. Although the present invention is described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention, which should be included in the scope of the claims of the present invention.

[0098] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. API sensitive data identification method using multiple DBSCAN and Apriori algorithms, characterized by: include: Collect data to build API sensitive datasets based on data sequence correlation; Build an API sensitivity quantification model based on the data set; Construct an API sensitivity clustering model based on multiple DBSCAN; Build an automatic identification model of API sensitive attributes based on the Apriori algorithm to identify sensitive data.

2. The API sensitive data identification method of multiple DBSCAN and Apriori algorithms as claimed in claim 1, characterized in that: The API sensitive data set based on data sequence correlation includes selecting API sensitive information data sequence and averaging each data sequence element group layer; Calculate the correlation coefficient of API sensitive information data sequence and establish API sensitive data set.

3. The API sensitive data identification method of multiple DBSCAN and Apriori algorithms as claimed in claim 2, characterized in that: The constructing of the API sensitivity quantification model includes, the API sensitivity quantification indicators include information entropy and maximum discrete entropy; Using information entropy and maximum discrete entropy as quantitative indicators, the attribute sensitivity of the API dataset is calculated and the data sensitivity is evaluated.

4. The API sensitive data identification method of multiple DBSCAN and Apriori algorithms as claimed in claim 3, characterized in that: The construction of the API sensitivity clustering model based on multiple DBSCANs includes that the API sensitivity clustering samples include a sensitive attribute set and a suspected sensitive attribute set; Perform dimensionality reduction on the API sensitive data set to obtain two-dimensional data points representing API sensitive information; The maximum global parameter radius is input to execute the DBSCAN algorithm on the data set multiple times to extract data point information, and the final clustering results are the sensitive attribute set and the suspected sensitive attribute set.

5. The API sensitive data identification method of multiple DBSCAN and Apriori algorithms as claimed in claim 1, characterized in that: The construction of the API sensitive attribute automatic identification model based on the Apriori algorithm includes mining the association relationship between suspected sensitive attributes and sensitive attributes through the Apriori algorithm, using support and confidence to evaluate the association relationship, and accurately identifying API sensitive data.

6. The API sensitive data identification method of multiple DBSCAN and Apriori algorithms as claimed in claim 1, characterized in that: The construction of the API sensitivity clustering model based on multiple DBSCANs also includes randomly selecting k data points from the data set P = {x1, x1, ..., xm} to form initial k mean vectors, denoted as {μ1, μ2, ..., μ k }; Input the value of the parameter Minpts, draw the k-distance curve, find all the effective distance mutation points in the graph, and determine the values ​​of m Eps; Enter the maximum parameter Eps to perform the first DBSCAN algorithm on the data set P; the clustering calculation with the maximum parameter Eps will remove the overly discrete points in the data set P as noise points, and merge some point clusters with close distances; The clusters after clustering with the maximum parameter Eps include both individually identified point clusters and unidentified point clusters that are merged; Input the remaining parameters Eps to execute the i-th (i=2,3,…,m-1) DBSCAN algorithm; the clustering calculation with a smaller parameter Eps will re-identify the clusters that were not identified and merged in the previous step, and further extract the feature information of the data points; The clustering results of each time are combined to obtain the final clustering result, which is divided into the sensitive attribute set SA and the suspected sensitive attribute set PSA.

7. The API sensitive data identification method of multiple DBSCAN and Apriori algorithms as claimed in claim 6, characterized in that: The sensitive data identification includes using association rule algorithms to mine strong association rules between attributes, optimizing the identification effect of sensitive attributes, and realizing automatic identification of API sensitive attributes; Define association rules. If the set of transaction items in the database is L = {l1,l2,…,l m }, the association rule formula is expressed as: in, And x1 and x2 represent two disjoint subsets of transaction items, x1∩x2=φ; The Apriori algorithm is used to mine the association between the sensitive attribute set SA and the suspected sensitive attribute set PSA, and the association rule expression is used To mark each association rule, the classification processing of suspected sensitive attribute sets is realized according to the number of successfully established association rules between two attribute sets. The attributes in the suspected sensitive attribute set that have a strong correlation with the sensitive attributes are accurately divided into the sensitive attribute set, and the data set is divided into the sensitive attribute set SA and the non-sensitive attribute set NSA.

8. A system for API sensitive data identification using the multiple DBSCAN and Apriori algorithms as described in any one of claims 1 to 7, characterized in that: The first building module is used to build an API sensitive data set based on data sequence correlation; The second building module is used to build an API sensitivity quantification model, wherein the API sensitivity quantification indicators include information entropy and maximum discrete entropy; The third construction module is used to construct an API sensitivity clustering model based on multiple DBSCAN, wherein the API sensitivity clustering samples include a sensitive attribute set and a suspected sensitive attribute set; The fourth building block is used to build an automatic identification model for API sensitive attributes based on the Apriori algorithm.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the API sensitive data identification method of multiple DBSCAN and Apriori algorithms described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the API sensitive data identification method of multiple DBSCAN and Apriori algorithms described in any one of claims 1 to 7 are implemented.

Images