CN117556449A

CN117556449A - Chemical detection report data encryption method based on big data

Info

Publication number: CN117556449A
Application number: CN202311677196.3A
Authority: CN
Inventors: 刘智明; 黄彦俊; 彭翠冰; 刘娟; 马铭骏
Original assignee: Guangdong Dewei Inspection And Certification Co ltd
Current assignee: Guangdong Dewei Inspection And Certification Co ltd
Priority date: 2023-12-08
Filing date: 2023-12-08
Publication date: 2024-02-13
Anticipated expiration: 2043-12-08
Also published as: CN117556449B

Abstract

The application relates to the technical field of data processing, in particular to a chemical detection report data encryption method based on big data, wherein a detection report encryption decision network can pertinently mine a sensitive report element semantic matrix of an original chemical detection report to be encrypted under any one of target text semantic dimensions, the characteristic characterization performance of the sensitive report element semantic matrix is improved, so that the determination precision of semantic element commonality weights of the sensitive report element semantic matrix and report element semantic matrix in a preset report element semantic matrix pool is improved, the data encryption decision view of the original chemical detection report can be accurately determined according to the semantic element commonality weights, the targeted chemical detection report data encryption processing can be realized through the data encryption decision view, the confidentiality and the integrity of data are greatly enhanced, and the data is ensured to be prevented from being tampered and leaked in the transmission and storage processes.

Description

Chemical detection report data encryption method based on big data

Technical Field

The application relates to the technical field of data processing, in particular to a chemical detection report data encryption method based on big data.

Background

With the advent of the information age, a vast amount of chemical detection report data needs to be transmitted and stored. However, these reports often contain sensitive information that, if tampered with or revealed during transmission and storage, can have serious adverse effects on the relevant parties. Traditional encryption methods may not meet the processing requirements for large amounts of data and may not effectively extract and protect sensitive information in the detection report.

Disclosure of Invention

In order to improve the technical problems in the related art, the application provides a chemical detection report data encryption method based on big data.

In a first aspect, an embodiment of the present application provides a method for encrypting chemical detection report data based on big data, applied to an artificial intelligence encryption system, the method including:

acquiring an original chemical detection report to be subjected to encryption processing, wherein the original chemical detection report belongs to at least one of a plurality of target text semantic dimensions;

transmitting the original chemical detection report into a detection report encryption decision network to obtain a sensitive report element semantic matrix generated by the detection report encryption decision network, wherein the detection report encryption decision network is obtained by adjusting a target report element semantic matrix, the target report element semantic matrix is obtained by mining a detection report semantic relationship spectrum through a basic encryption decision network, and the detection report semantic relationship spectrum is obtained according to chemical detection reports respectively corresponding to the plurality of target text semantic dimensions;

Determining semantic element commonality weights of the sensitive report element semantic matrix and report element semantic matrices in a pre-allocation report element semantic matrix pool, wherein the pre-allocation report element semantic matrix pool comprises at least one report element semantic matrix corresponding to a past chemical detection report, and the at least one past chemical detection report belongs to any one of the plurality of target text semantic dimensions;

and determining a data encryption decision perspective of the original chemical detection report based on the semantic element commonality weights.

Optionally, the step of calibrating the detection report encryption decision network includes:

acquiring chemical detection reports corresponding to semantic dimensions of a plurality of target texts respectively;

processing the chemical detection report under each target text semantic dimension through a semantic knowledge mining sub-network corresponding to each target text semantic dimension in the basic encryption decision network to obtain detection report semantics Guan Jipu of the chemical detection report corresponding to each target text semantic dimension, wherein the basic encryption decision network comprises semantic knowledge mining sub-networks respectively corresponding to different target text semantic dimensions;

Transmitting the detection report semantics Guan Jipu into a semantic matrix mapping subnet in the basic encryption decision network to obtain a target report element semantic matrix generated by the semantic matrix mapping subnet;

and calibrating the basic encryption decision network based on the target report element semantic matrix to obtain the detection report encryption decision network.

Optionally, the step of transmitting the detection report semantics Guan Jipu to a semantic matrix mapping subnet in the basic encryption decision network to obtain a target report element semantic matrix generated by the semantic matrix mapping subnet includes:

carrying out semantic feature integration on the detection report semantic relation spectrums corresponding to the chemical detection reports under the semantic dimensions of the plurality of target texts to obtain detection report global semantics Guan Jipu;

and transmitting the detection report global semantics Guan Jipu into the semantic matrix mapping sub-network to obtain a target report element semantic matrix generated by the semantic matrix mapping sub-network.

Optionally, the calibrating the basic encryption decision network based on the target report element semantic matrix to obtain the detection report encryption decision network includes:

Determining training error variables corresponding to the target report element semantic matrix through an encryption decision discrimination subnet corresponding to the target report element semantic matrix in the basic encryption decision network, wherein the basic encryption decision network comprises an encryption decision discrimination subnet corresponding to the report element semantic matrix of a target encryption suggestion keyword, and the target encryption suggestion keyword is determined based on a target text semantic dimension corresponding to the detection report global semantic relationship spectrum;

and based on the training error variable, adjusting an encryption decision discrimination subnet corresponding to the target report element semantic matrix, semantic knowledge mining subnets corresponding to the plurality of target text semantic dimensions respectively and the semantic matrix mapping subnets to obtain the detection report encryption decision network.

Optionally, the processing, by a semantic knowledge mining subnet corresponding to each target text semantic dimension in the basic encryption decision network, the chemical detection report under each target text semantic dimension to obtain a detection report semantic relationship spectrum of the chemical detection report corresponding to each target text semantic dimension includes:

Performing key text block marking processing and key text block matching processing on the chemical detection report to obtain an intermediate chemical detection report with fine granularity of a target text;

and processing the intermediate chemical detection report under each target text semantic dimension through a semantic knowledge mining subnet corresponding to each target text semantic dimension in the basic encryption decision network to obtain a detection report semantic relation spectrum of the chemical detection report corresponding to each target text semantic dimension.

Optionally, the processing, by a semantic knowledge mining subnet corresponding to each target text semantic dimension in the basic encryption decision network, the intermediate chemical detection report under each target text semantic dimension to obtain a detection report semantic relationship spectrum of the chemical detection report corresponding to each target text semantic dimension includes:

extracting detection reports with detection report sensitivity meeting the requirement of pre-configuration sensitivity from the intermediate chemical detection reports in each target text semantic dimension as target chemical detection reports in each target text semantic dimension;

and processing the target chemical detection report under each target text semantic dimension through a semantic knowledge mining subnet corresponding to each target text semantic dimension in the basic encryption decision network to obtain a detection report semantic relation spectrum of the chemical detection report corresponding to each target text semantic dimension.

Optionally, the plurality of target text semantic dimensions include numerical text semantic dimensions, and the extracting, from the intermediate chemical detection reports in each of the target text semantic dimensions, a detection report with detection report sensitivity meeting a preset sensitivity requirement as a target chemical detection report in each of the target text semantic dimensions includes:

determining a maximum word frequent item variable and a minimum word frequent item variable of an intermediate chemical detection report under the numerical text semantic dimension;

based on a comparison result of the word frequent item variable of each word unit in the intermediate chemical detection report under the numerical text semantic dimension and the minimum word frequent item variable, updating the word frequent item variable of each word unit in the intermediate chemical detection report under the numerical text semantic dimension to obtain a first chemical detection report;

projecting the word frequent item variable of each word unit in the first chemical detection report into the target word frequent item variable quadrant to obtain a second chemical detection report;

performing detection element thermal mapping on the second chemical detection report to obtain a third chemical detection report;

And extracting a detection report with the detection report sensitivity meeting the first pre-configuration sensitivity requirement from the third chemical detection report as a target chemical detection report under the numerical text semantic dimension.

Optionally, extracting a detection report with the detection report sensitivity meeting the first preset sensitivity requirement from the third chemical detection report as a target chemical detection report under the numerical text semantic dimension, including:

determining a first report text set including chemical property descriptions in the third chemical detection report;

determining word frequent item average variables of each word unit in the first report text set;

and extracting a detection report with the average variable of the word frequent items higher than a first threshold value from the third chemical detection report, and taking the detection report as a target chemical detection report under the numerical text semantic dimension.

Optionally, the plurality of target text semantic dimensions include chemical formula text semantic dimensions, the extracting, from the intermediate chemical detection reports under each of the target text semantic dimensions, a detection report with detection report sensitivity meeting a preset sensitivity requirement as a target chemical detection report under each of the target text semantic dimensions includes:

Determining a target chemical formula description semantic value corresponding to the intermediate chemical detection report in the chemical formula text semantic dimension, wherein the target chemical formula description semantic value is a target statistical value of chemical formula description semantic values of all word units in the intermediate chemical detection report in the chemical formula text semantic dimension;

determining a target chemical formula description semantic value quadrant based on the target chemical formula description semantic value;

changing a chemical formula description semantic value, which is in the intermediate chemical formula detection report under the chemical formula text semantic dimension and is not in the chemical formula description semantic value quadrant, into a specified value to obtain a fourth chemical formula detection report, and obtaining a minimum chemical formula description semantic value, which is in the fourth chemical formula detection report and is not 0, as a target chemical formula description semantic value;

updating the chemical formula description semantic value of each word unit with the chemical formula description semantic value of not 0 in the fourth chemical detection report based on the comparison result of the chemical formula description semantic value of each word unit with the chemical formula description semantic value of not 0 in the fourth chemical detection report and obtaining a fifth chemical detection report;

And extracting a detection report with the detection report sensitivity meeting the second pre-configuration sensitivity requirement from the fifth chemical detection report as a target chemical detection report under the semantic dimension of the chemical formula text.

Optionally, extracting a detection report with the detection report sensitivity meeting the second preset sensitivity requirement from the fifth chemical detection report as a target chemical detection report under the semantic dimension of the chemical formula text, including:

determining a second report text set including chemical property descriptions in the fifth chemical detection report;

determining the proportion of word units with the semantic value of the chemical formula description not being 0 in the second report text set as the proportion of effective word units;

determining a chemical formula description semantic value quadrant based on the maximum chemical formula description semantic value and the minimum chemical formula description semantic value of the second report text set;

disassembling the chemical formula description semantic value quadrant into a plurality of local numerical value quadrants;

decomposing the chemical formula description semantic values of the chemical formula description semantic values in the same local numerical quadrant in the second report text set into a linear characteristic variable set to obtain a plurality of linear characteristic variable sets;

Taking the linear characteristic variable set with the number of the chemical formula description semantic values reaching a second threshold value in the plurality of linear characteristic variable sets as a target linear characteristic variable set;

and extracting a detection report that the proportion of the effective word units reaches a third threshold value and the number of the target linear characteristic variable sets reaches a fourth threshold value from the fifth chemical detection report, and taking the detection report as a target chemical detection report under the semantic dimension of the chemical formula text.

In a second aspect, the present application also provides an artificial intelligence encryption system comprising a processor and a memory; the processor is in communication with the memory, and the processor is configured to read and execute a computer program from the memory to implement the method described above.

In a third aspect, the present application also provides a computer readable storage medium having stored thereon a program which, when executed by a processor, implements the method described above.

According to the chemical detection report data encryption method based on big data, the detection report encryption decision network is obtained by adjusting the chemical detection report corresponding to each of a plurality of target text semantic dimensions, the detection report encryption decision network can pertinently mine sensitive report element semantic matrixes of the original chemical detection report to be encrypted under any one of the target text semantic dimensions, the characteristic characterization performance of the sensitive report element semantic matrixes is improved, the determination precision of semantic element commonality weights of the sensitive report element semantic matrixes and report element semantic matrixes in a pre-allocation report element semantic matrix pool is improved, so that the data encryption decision view of the original chemical detection report can be accurately determined according to the semantic element commonality weights, the encryption processing of the original chemical detection report data can be realized through the data encryption decision view, the confidentiality and the integrity of data are greatly enhanced, and the tamper and leakage are prevented in the transmission and storage processes.

Drawings

The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the application and together with the description, serve to explain the principles of the application.

Fig. 1 is a schematic flow chart of a chemical detection report data encryption method based on big data according to an embodiment of the present application.

Detailed Description

Reference will now be made in detail to exemplary embodiments, examples of which are illustrated in the accompanying drawings. When the following description refers to the accompanying drawings, the same numbers in different drawings refer to the same or similar elements, unless otherwise indicated. The implementations described in the following exemplary examples are not representative of all implementations consistent with the present application. Rather, they are merely examples of apparatus and methods consistent with aspects of the present application.

It should be noted that the terms "first," "second," and the like in the description of the present application and the above-described figures are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order.

The method embodiments provided by the embodiments of the present application may be performed in an artificial intelligence encryption system, a computer device, or similar computing device. Taking the example of operating on an artificial intelligence encryption system, the artificial intelligence encryption system may include one or more processors (which may include, but is not limited to, a microprocessor MCU, a programmable logic device FPGA, etc. processing means) and memory for storing data, and optionally, transmission means for communication functions. It will be appreciated by those of ordinary skill in the art that the above-described architecture is merely illustrative and is not intended to limit the architecture of the artificial intelligence encryption system described above. For example, the artificial intelligence encryption system may also include more or fewer components than shown above, or have a different configuration than shown above.

The memory may be used to store a computer program, for example, a software program of application software and a module, such as a computer program corresponding to a method for encrypting chemical detection report data based on big data in the embodiments of the present application, and the processor executes the computer program stored in the memory, thereby performing various functional applications and data processing, that is, implementing the method described above. The memory may include high speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid state memory. In some examples, the memory may further include memory remotely located with respect to the processor, the remote memory being connectable to the artificial intelligence encryption system through a network. Examples of such networks include, but are not limited to, the internet, intranets, local area networks, mobile communication networks, and combinations thereof.

The transmission means is used for receiving or transmitting data via a network. Specific examples of the network described above may include a wireless network provided by a communications provider of an artificial intelligence encryption system. In one example, the transmission means comprises a network adapter (Network Interface Controller, simply referred to as NIC) that can be connected to other network devices via a base station to communicate with the internet. In one example, the transmission device may be a Radio Frequency (RF) module, which is used to communicate with the internet wirelessly.

Referring to fig. 1, fig. 1 is a flow chart of a chemical detection report data encryption method based on big data according to an embodiment of the present application, where the method is applied to an artificial intelligence encryption system, and further may include steps 110-140.

Step 110, acquiring an original chemical detection report to be subjected to encryption processing, wherein the original chemical detection report belongs to at least one of a plurality of target text semantic dimensions.

Step 120, the original chemical detection report is transmitted into a detection report encryption decision network to obtain a sensitive report element semantic matrix generated by the detection report encryption decision network, the detection report encryption decision network is obtained through adjustment of a target report element semantic matrix, the target report element semantic matrix is obtained through excavation of a detection report semantic relationship spectrum by a basic encryption decision network, and the detection report semantic relationship spectrum is obtained according to chemical detection reports respectively corresponding to the plurality of target text semantic dimensions.

130, determining semantic element commonality weights of the sensitive report element semantic matrix and report element semantic matrices in a pre-allocation report element semantic matrix pool, wherein the pre-allocation report element semantic matrix pool comprises at least one report element semantic matrix corresponding to a past chemical detection report, and the at least one past chemical detection report belongs to any one of the plurality of target text semantic dimensions.

And 140, determining the data encryption decision point of the original chemical detection report based on the semantic element commonality weight.

In step 110, the original chemical detection report is an original, raw text file that records the results of detecting a chemical. For example, an original chemical detection report may contain information about the name of the chemical, the date of the detection, the detection method used, the detection result, etc. The target text semantic dimension refers to a predefined set of classification criteria for dividing and understanding the content of the text. For example, for chemical detection reports, possible semantic dimensions include "chemical name", "detection method", "detection result", and so on.

For example, the primary goal of step 110 is to obtain raw chemical detection reports that require encryption processing. This report will belong to at least one of a predefined number of target text semantic dimensions. The target text semantic dimension herein may be understood as a classification criterion of the text, such as the type of chemical, date of detection, result of detection, etc.

For another example, there is a detection report on sulfuric acid, which belongs to the semantic dimension "chemical name"; if the report shows that the sulfuric acid content exceeds the standard, it may also belong to the semantic dimension "detection results".

After the original chemical detection report is obtained, the next step is to send it into the detection report encryption decision network. The function of this network is to decide how to encrypt the report based on a pre-set target report element semantic matrix. The target report element semantic matrix is obtained by mining a detection report semantic relation spectrum, and the detection report semantic relation spectrum is obtained according to chemical detection reports respectively corresponding to a plurality of target text semantic dimensions. Through the steps, the report content needing encryption processing can be determined, and a corresponding data encryption strategy can be further formulated.

In step 120, the report encryption decision network is a model for determining which chemical report elements should be encrypted. It receives as input the original chemical detection report and generates a sensitive report element semantic matrix that indicates which report elements are deemed sensitive and need to be encrypted. The sensitive report element semantic matrix is a matrix generated by the detection report encryption decision network that indicates which elements in a particular chemical detection report are sensitive and need to be encrypted. The target report element semantic matrix is a matrix that describes the semantic relationships of individual elements in a target report (i.e., a report that needs encryption). Tuning is a process in which a target report element semantic matrix is used to optimize and adjust the detection report encryption decision network to more accurately identify sensitive report elements that need encryption. The test report semantic relationship spectrum is also referred to as a test report semantic feature map, which is a graphical representation showing the semantic relationships between the individual elements in a chemical test report. The basic encryption decision network is a model for mining information from the test report semantic relationship spectrum to generate a target report element semantic matrix. Some of the target text semantic dimensions are various possible features or attributes of the chemical detection report, such as the type of report, the chemical species involved, the detection result, etc.

For example, the raw chemical detection report is first input into a detection report encryption decision network. The network has been calibrated through the target reporting element semantic matrix to more accurately identify sensitive reporting elements that need encryption. The target report element semantic matrix is mined from the detection report semantic relationship spectrum by the basic encryption decision network. And the detection report semantic relation spectrum is created according to chemical detection reports respectively corresponding to a plurality of target text semantic dimensions. After all these steps are completed, the detection report encryption decision network will output a sensitive report element semantic matrix indicating which report elements are deemed sensitive and need to be encrypted.

For another example, there is a chemical detection report that contains the chemical name, chemical property, detection result, and other elements. First, this report will be input into the detection report encryption decision network. Then, based on the target report element semantic matrix (which may represent "chemical name" and "detection result" are sensitive elements), the network may decide that these two elements need to be encrypted. Thus, the sensitive report element semantic matrix will contain both elements.

In step 130, the pre-assigned report element semantic matrix pool is a collection or database of stored report element semantic matrices. These semantic matrices all correspond to past chemical detection reports. The report element semantic matrix is a data structure used to represent and store the various elements in the report and the relationships between them. Each element has its corresponding semantic tag, such as "chemical name", "detection method", etc. The semantic element commonality weight is a metric value representing the similarity between the sensitive report element semantic matrix and the pre-assigned report element semantic matrix. If the two matrices have a high commonality over some elements, then their commonality weights are greater. Past chemical detection reports are old chemical detection reports that have been completed and stored. They are used to generate and store pre-assembled report element semantic matrices into a pool of pre-assembled report element semantic matrices.

Step 130 is mainly to determine semantic element commonality weights of the sensitive report element semantic matrix and the report element semantic matrix in the pre-assigned report element semantic matrix pool. This step is accomplished by comparing the similarity between the sensitive report element semantic matrix and the pre-assigned report element semantic matrix.

For example, there is a detection report on sulfuric acid, which produces a sensitive report element semantic matrix. There is also a pool of pre-configured report element semantic matrices containing a plurality of report element semantic matrices of past chemical detection reports (e.g., ammonia, nitric acid, etc.).

In this step, the similarity of the sensitive report element semantic matrix of the sulfuric acid detection report and each report element semantic matrix in the pre-allocated report element semantic matrix pool is compared, and a semantic element commonality weight is allocated to each comparison result. If the sensitive reporting element semantic matrix of the sulfuric acid detection report has high commonality with the reporting element semantic matrix of a previous report on certain elements, the commonality weight between the sensitive reporting element semantic matrix and the reporting element semantic matrix of the sulfuric acid detection report is larger. In this way, it can be determined which reporting elements are important and which are negligible, thus providing basis for the next data encryption decision.

In step 140, the data encryption decision perspective refers to an encryption decision scheme determined based on the semantic element commonality weights by comparing the similarity of the sensitive report element semantic matrix and the pre-assigned report element semantic matrix.

The semantic element commonality weights are first calculated by comparing the similarity between the sensitive report element semantic matrix (which is the report element that is desired to be encrypted) and the pre-assigned report element semantic matrix pool (which contains the report element semantic matrix corresponding to the past chemical detection report). This weight reflects how semantically the report currently to be encrypted is similar to the existing past report.

Based on this semantic element commonality weight, then, a data encryption decision perspective of the original chemical detection report can be determined. That is, it may be determined from this weight which report elements most need to be encrypted to protect sensitive information in the report.

For example, assuming that the sensitive report element semantic matrix indicates that both elements "chemical name" and "detection result" are sensitive, encryption is required. At the same time, a past report in the pre-configured report element semantic matrix pool also identifies the two elements. After comparing the similarity of the two matrices, a higher semantic element commonality weight may be obtained, which indicates that the current report to be encrypted has a great semantic similarity with the past report.

Thus, based on this high weight, the data encryption decision perspective may be: the two elements of "chemical name" and "detection result" are encrypted to protect sensitive information in the chemical detection report.

In some more detailed examples, the sensitive report element semantic matrix is generated by a detection report encryption decision network to describe various elements and their interrelationships in a chemical detection report. This matrix helps to understand which information is sensitive and requires encryption processing. For example, there is an original report on sulfuric acid detection, which contains the following elements: chemical name (sulfuric acid), date of detection (2022, 3, 1 day), detection result (ph=1), and some other possible details. When the report is entered into the test report encryption decision network, the network generates a sensitive report element semantic matrix indicating whether the information is sensitive or not according to predefined rules and criteria. After determining which elements are sensitive, it is next determined how to encrypt the sensitive information. This is typically done by calculating a semantic element commonality weight and determining from this weight the data encryption decision perspective of the original chemical detection report. Specific encryption methods may include various encryption techniques such as symmetric encryption, asymmetric encryption, and the like.

In other more detailed examples, the sensitive report element semantic matrix (or what may be considered as a sensitive report element semantic feature vector) is a data structure that is used to represent and store elements of a chemical detection report that are deemed sensitive and that require encryption. These elements typically include information such as chemical name, chemical nature, test results, etc. Each element is given a semantic tag and converted to a numerical feature for processing. For example, "chemical name" may be labeled 1, "chemical property" may be labeled 2, and "detection result" may be labeled 3. Such digital labels can help more conveniently process and analyze data. At the same time, each element is given a weight indicating the importance of the element. For example, if "detection result" is considered to be the most important element, it may be given the highest weight, such as 0.7; accordingly, the weight of "chemical name" and "chemical property" may be lower, such as 0.2 and 0.1, respectively. This sensitive report element semantic matrix is then input into the pre-assigned report element semantic matrix pool for comparison and matching. From the comparison, semantic element commonality weights can be calculated and based on this, a data encryption decision perspective, i.e. which elements should be encrypted and how to encrypt, is determined.

In still other more detailed examples, when feature vectors are used to represent the sensitive report element semantic matrix, the sensitivity of each element may be taken as one dimension of the vector. For the example of sulfuric acid detection report, the semantic matrix of the sensitive report element can be expressed as a three-dimensional vector. For example, "chemical name", "date of detection" and "detection result" are set to correspond to the first, second and third dimensions of the vector, respectively. Thus, the corresponding sensitive report element semantic vector may be [1,0,1]. In this vector, 1 indicates that the corresponding element is sensitive and needs to be encrypted; 0 indicates that the corresponding element is insensitive and may not be encrypted. Thus, the sensitive report element semantic matrix is represented in a compact manner by a feature vector that explicitly indicates which elements are sensitive and require encryption.

In some examples, exemplary content of the raw chemical detection report may be as follows:

company name: XYZ laboratory;

report number: 78945612;

reporting date: 2022, 3, 15;

project name: evaluating comprehensive performance;

sample information:

chemical name: a carbamate;

CAS number: 60-00-4;

sample source: ABC chemical plant;

sample state: a liquid state;

detecting parameters and results:

appearance and properties: colorless transparent liquid;

specific gravity (20 ℃): 1.01 g/cm;

pH value: 5.5;

flash point: 94 ℃;

boiling point: 157 ℃;

solubility (20 ℃ C. In water): fully dissolving;

content of harmful substances:

lead (Pb): <0.001%;

arsenic (As): <0.00001%;

mercury (Hg): <0.00001%;

microbial index:

total colony count: <100CFU/mL;

coli group: undetected;

mould and yeast: not detected.

Conclusion: the performance indexes of the samples all meet the related regulation requirements, and harmful substances and microorganisms exceeding the safety standard are not detected. This report provides more detailed information about the sample, including various parameters of appearance and character, specific gravity, PH, flash point, boiling point, solubility, content of hazardous substances, and microbiological indicators. These are sensitive information that may require encryption processing.

Based on the above example, through steps 110-140, the following pieces of sensitive information in the report may be chosen to be encrypted:

chemical name: a carbamate;

CAS number: 60-00-4;

Content of harmful substances:

lead (Pb): <0.001%;

arsenic (As): <0.00001%;

mercury (Hg): <0.00001%;

microbial index:

total colony count: <100CFU/mL;

coli group: undetected;

mould and yeast: not detected.

Other information such as company name, report number, report date, item name, sample source, sample status, and general performance parameters of the sample (such as appearance and properties, specific gravity, PH, flash point, boiling point, and solubility) are not generally considered sensitive and may not be encrypted. Based on this, the data encryption decision point may be to encrypt the chemical name, CAS number, content of harmful substances, and microbial index.

In summary, the detection report encryption decision network is obtained by adjusting chemical detection reports corresponding to a plurality of target text semantic dimensions respectively, the detection report encryption decision network can pertinently mine a sensitive report element semantic matrix of an original chemical detection report to be encrypted under any one of the target text semantic dimensions, the characteristic characterization performance of the sensitive report element semantic matrix is improved, the determination precision of semantic element commonality weights of the sensitive report element semantic matrix and report element semantic matrix in a pre-allocation report element semantic matrix pool is improved, and thus, the data encryption decision viewpoint of the original chemical detection report can be accurately determined according to the semantic element commonality weights, so that the targeted chemical detection report data encryption processing is realized through the data encryption decision viewpoint, the confidentiality and the integrity of data are greatly enhanced, and the data is prevented from being tampered and leaked in the transmission and storage processes.

In detail, by encrypting the sensitive information, the data may be protected from unauthorized access, tampering, and risk of leakage during transmission and storage. Even if the data is acquired by a third party, the data cannot be read from the data because the data is in an encrypted state, so that confidentiality of the data is protected. The detection report encryption decision network can accurately determine which elements need to be encrypted according to the semantic element commonality weight, so that unnecessary total encryption or encryption omission is avoided, and the pertinence and the effectiveness of encryption are improved. The relation and importance among the elements can be better reflected through the sensitive report element semantic matrix mined by the network, so that the matching degree of the report element semantic matrix in the pre-allocation report element semantic matrix pool is improved, and the overall reliability is enhanced. Not all data need to be encrypted, in this way, more resources can be used for protection of sensitive information, instead of unnecessary data encryption, so that resource allocation is optimized and efficiency is improved.

In some alternative embodiments, the tuning step of the detection report encryption decision network includes steps 210-240.

Step 210, acquiring chemical detection reports corresponding to the semantic dimensions of the target texts.

And 220, processing the chemical detection report under each target text semantic dimension through a semantic knowledge mining sub-network corresponding to each target text semantic dimension in the basic encryption decision network to obtain a detection report semantic Guan Jipu of the chemical detection report corresponding to each target text semantic dimension, wherein the basic encryption decision network comprises semantic knowledge mining sub-networks respectively corresponding to different target text semantic dimensions.

And 230, transmitting the detection report semantics Guan Jipu into a semantic matrix mapping subnet in the basic encryption decision network to obtain a target report element semantic matrix generated by the semantic matrix mapping subnet.

And 240, calibrating the basic encryption decision network based on the target report element semantic matrix to obtain the detection report encryption decision network.

The target text semantic dimension is a method for classifying information in the detection report. For example, chemical names, test results, test dates, etc. may all be considered different semantic dimensions. The underlying encryption decision network is a neural network that is used to decide which reporting elements need to be encrypted. It includes multiple subnets, each of which is responsible for handling one particular target text semantic dimension. The semantic knowledge mining subnetwork is a part of the basic encryption decision network and is responsible for extracting and understanding semantic information in the detection report. The test report semantic relationship spectrum is a data structure that describes the relationships between the various elements in the test report. The semantic matrix mapping subnetwork is another part of the basic encryption decision network and is responsible for converting the detection report semantic relation spectrum into a target report element semantic matrix. The target report element semantic matrix is a matrix that indicates which report elements need to be encrypted.

For example, there is a chemical detection report containing the following information: chemical name (sulfuric acid), date of detection (2022, 3, 1 day), and detection result (ph=1). This information can be considered as different target text semantic dimensions.

In step 210, the report is first obtained. Then, in step 220, the report is processed using semantic knowledge mining sub-networks in the underlying cryptographic decision network corresponding to each target text semantic dimension, resulting in a detection report semantic relationship spectrum. Next, in step 230, the detection report semantics Guan Jipu are passed into a semantic matrix mapping subnet to obtain a target report element semantic matrix. For example, this matrix may be [1,0,1], indicating that "chemical name" and "test result" need to be encrypted, while "date of test" is not. Finally, in step 240, the basic encryption decision network is calibrated based on the target report element semantic matrix to obtain the detection report encryption decision network. This network can more accurately determine which report elements need to be encrypted.

Thus, by encrypting the sensitive information, the data can be prevented from being illegally acquired and utilized. Not all information needs to be encrypted, and by means of a decision network it can be determined which information is more sensitive and should be encrypted with priority, so that limited computing resources can be used more efficiently. Through machine learning and neural network techniques, it is possible to more accurately determine which information is sensitive based on a large amount of historical data and complex algorithms.

In some preferred embodiments, the step 230 of transferring the detection report semantics Guan Jipu into a semantic matrix mapping subnet in the underlying encryption decision network, to obtain a target report element semantic matrix generated by the semantic matrix mapping subnet, includes steps 231-232.

And 231, carrying out semantic feature integration on the detection report semantic relation spectrums corresponding to the chemical detection reports under the plurality of target text semantic dimensions respectively to obtain a detection report global semantic relation spectrum.

And 232, transmitting the detection report global semantics Guan Jipu into the semantic matrix mapping sub-network to obtain a target report element semantic matrix generated by the semantic matrix mapping sub-network.

The semantic matrix mapping sub-network is a neural network structure, and its task is to receive the input global semantic Guan Jipu of the detection report, and then map (convert) it into the semantic matrix of the target report element. The test report global semantic relationship spectrum is a data structure containing the semantic information of the chemical test report in all the target text semantic dimensions. By integrating these semantic information, a more comprehensive, higher level semantic representation can be obtained.

For example, there are two target text semantic dimensions, the "chemical property" and the "test result", respectively. There are several chemical detection reports in each dimension, for example:

report A, B, C under the "chemical properties" dimension;

the "detection results" dimension is reported D, E, F.

First, detection report semantics Guan Jipu need to be generated separately for all reports in these two dimensions, which can be achieved by natural language processing techniques such as word embedding. Next, in step 231, all semantic relationship spectra in these two dimensions are subjected to semantic feature integration, i.e. they are combined together, resulting in a detection report global semantic relationship spectrum containing all report semantic information. This global semantic Guan Jipu is then passed into a semantic matrix mapping subnet in step 232. The sub-network learns and extracts the important features and generates a target report element semantic matrix containing semantic information of sensitive report elements that need encryption.

Therefore, through semantic feature integration and neural network mapping, the method realizes mining, extracting and representing the sensitive information needing encryption processing from a large number of chemical detection reports, improves the accuracy and effectiveness of data encryption decisions, is beneficial to protecting the safety of sensitive data, and prevents data leakage and tampering, thereby enhancing the confidentiality and integrity of the data.

In some preferred embodiments, the adjusting the basic encryption decision network based on the target report element semantic matrix in step 240 to obtain the detection report encryption decision network includes steps 241-242.

Step 241, determining a training error variable corresponding to the target report element semantic matrix through an encryption decision discrimination subnet corresponding to the target report element semantic matrix in the basic encryption decision network, wherein the basic encryption decision network comprises an encryption decision discrimination subnet corresponding to the report element semantic matrix of a target encryption suggestion keyword, and the target encryption suggestion keyword is determined based on a target text semantic dimension corresponding to the detection report global semantic relationship spectrum.

And step 242, based on the training error variable, calibrating an encryption decision discrimination subnet corresponding to the semantic matrix of the target report element, semantic knowledge mining subnets corresponding to the semantic dimensions of the plurality of target texts and the semantic matrix mapping subnets respectively to obtain the detection report encryption decision network.

Wherein the target report element semantic matrix is a data structure that indicates which elements of the report need to be encrypted. The underlying encryption decision network is a neural network that is used to determine which reporting elements need to be encrypted. It includes multiple subnets, each of which is responsible for handling one particular target text semantic dimension. The encryption decision making subnet is part of the basic encryption decision making network and is responsible for determining which report elements need to be encrypted. Training error variable: in machine learning, the training error variable is the difference between the model predicted value and the true value. The model is optimized by minimizing this error. The target encryption suggestion keyword is key information or a keyword for which the model is expected to recognize and propose an encryption suggestion. The detection report global semantic relationship spectrum is a data structure that describes the global relationships between the various elements in the detection report.

For example, a target report element semantic matrix, such as [1,0,1], has been obtained through the above steps, indicating that "chemical name" and "test result" need to be encrypted, while "date of test" does not.

In step 241, training error variables are determined using an encryption decision making subnet in the underlying encryption decision network corresponding to the target reporting element semantic matrix. For example, if the model predicts that the "detection date" also needs encryption, then the training error variable is 1. Next, in step 242, the encryption decision making sub-network, the semantic knowledge mining sub-network, and the semantic matrix mapping sub-network are adjusted based on the training error variable to obtain a detection report encryption decision network.

In this way, by encrypting sensitive information, the data may be protected from unauthorized access, tampering, and risk of leakage during transmission and storage. Not all information needs to be encrypted, and by means of a decision network it can be determined which information is more sensitive and should be encrypted with priority, so that limited computing resources can be used more efficiently. Through machine learning and neural network techniques, it is possible to more accurately determine which information is sensitive based on a large amount of historical data and complex algorithms. Through continuous training and adjustment, the model can continuously improve the accuracy of judgment, thereby improving the overall system performance.

In some optional examples, processing the chemical detection report in each of the target text semantic dimensions through the semantic knowledge mining sub-network corresponding to each of the target text semantic dimensions in the basic encryption decision network in step 220 to obtain a detection report semantic relationship spectrum of the chemical detection report corresponding to each of the target text semantic dimensions includes steps 221-222.

And 221, performing key text block marking processing and key text block matching processing on the chemical detection report to obtain an intermediate chemical detection report with fine granularity of a target text.

And 222, processing the intermediate chemical detection report in each target text semantic dimension through a semantic knowledge mining subnet corresponding to each target text semantic dimension in the basic encryption decision network to obtain a detection report semantic relation spectrum of the chemical detection report corresponding to each target text semantic dimension.

The key text block marking process is a method of processing the original chemical detection report, and identifies and marks important pieces of information in the report, or "key text blocks". And (3) matching the key text blocks: after the key text blocks are marked, this processing step matches the key text blocks with predefined target text semantic dimensions to determine the semantic dimensions to which each key text block belongs. Intermediate chemical detection report: the chemical detection report obtained after the key text block marking and matching process has higher structuring degree and more definite semantic classification than the original report.

For example, there is a chemical detection report containing the following information: chemical name (sulfuric acid), date of detection (2022, 3, 1 day), and detection result (ph=1). This information can be considered as different blocks of key text.

In step 221, the report is first subjected to a text block labeling process, i.e., the above three text blocks are identified. Then, a key text block matching process is performed to match the key text blocks with predefined target text semantic dimensions. For example, it is possible to match "chemical name" to the dimension of "substance attribute", and match both "detection date" and "detection result" to the dimension of "detection information". Thus, a target text fine-grained intermediate chemical detection report is obtained. Next, in step 222, the intermediate chemical detection report in each target text semantic dimension is transmitted to a corresponding semantic knowledge mining subnet for processing, so as to obtain a detection report semantic relationship spectrum in each dimension.

By adopting the embodiment, the original report is firstly structured, and then semantic mining is carried out by utilizing the neural network, so that sensitive information needing encryption processing is extracted from a large number of chemical detection reports, the accuracy and effectiveness of data encryption decision are improved, the safety of sensitive data is protected, the data leakage and tampering are prevented, and the confidentiality and the integrity of the data are enhanced.

In some alternative embodiments, processing the intermediate chemical detection report in each of the target text semantic dimensions through a semantic knowledge mining subnet corresponding to each of the target text semantic dimensions in the base encryption decision network as described in step 222 results in a detection report semantic relationship spectrum corresponding to the chemical detection report in each of the target text semantic dimensions, including steps 2221-2222.

Step 2221, extracting a detection report with detection report sensitivity meeting the requirement of pre-configuration sensitivity from the intermediate chemical detection report in each target text semantic dimension, and taking the detection report as a target chemical detection report in each target text semantic dimension.

Step 2222, processing, by a semantic knowledge mining subnet corresponding to each target text semantic dimension in the basic encryption decision network, the target chemical detection report in each target text semantic dimension, so as to obtain a detection report semantic relationship spectrum of the chemical detection report corresponding to each target text semantic dimension.

In the above embodiment, the intermediate chemical detection report is a preliminary, not fully processed chemical detection report, which contains a large amount of raw data and information. The pre-configured sensitivity requirement is a pre-set requirement for which information in the chemical detection report needs to be considered sensitive and encrypted. The target chemical detection report is the portion of the intermediate chemical detection report that is extracted based on the pre-set sensitivity requirements.

For example, there is an intermediate chemical detection report that contains information of semantic dimensions of multiple target texts, such as chemical names, detection results, detection dates, etc. Meanwhile, the pre-formulation sensitivity requirements are: the "chemical name" and "detection result" need to be regarded as sensitive information.

In step 2221, the portion meeting the pre-configured sensitivity requirements, i.e., the chemical name and the test result, is first extracted from the intermediate chemical test report, which information constitutes the target chemical test report. Next, in step 2222, these target chemical detection reports are processed using the semantic knowledge mining sub-networks in the underlying encryption decision network corresponding to each target text semantic dimension, resulting in a detection report semantic relationship spectrum for each target text semantic dimension.

By the design, only the part meeting the requirement of the pre-allocation sensitivity is extracted, so that only information really needing encryption can be processed and transmitted, and the risk of data leakage is greatly reduced. Not all information needs to be encrypted, and such policies can help to more efficiently use limited computing resources, processing only those that are truly important, sensitive. Because only the information which really needs to be encrypted is processed, the processing time can be greatly reduced, and the overall efficiency of the system is improved.

In some possible examples, the number of target text semantic dimensions includes a numeric text semantic dimension. Based on this, the detection report with detection report sensitivity meeting the pre-configured sensitivity requirement is extracted from the intermediate chemical detection report in each of the target text semantic dimensions in step 2221 as the target chemical detection report in each of the target text semantic dimensions, including steps 22211-22215.

Step 22211, determining the maximum word frequent item variable and the minimum word frequent item variable of the intermediate chemical detection report under the numerical text semantic dimension.

Step 22212, based on the comparison result of the word frequent item variable of each word unit in the intermediate chemical detection report in the numerical text semantic dimension and the minimum word frequent item variable, updates the word frequent item variable of each word unit in the intermediate chemical detection report in the numerical text semantic dimension, and obtains a first chemical detection report.

Step 22213, projecting the word frequent item variable of each word unit in the first chemical detection report into the target word frequent item variable quadrant, so as to obtain a second chemical detection report.

Step 22214, performing a detection element thermal mapping on the second chemical detection report to obtain a third chemical detection report.

And 22215, extracting a detection report with detection report sensitivity meeting the first pre-configuration sensitivity requirement from the third chemical detection report, and taking the detection report as a target chemical detection report under the numerical text semantic dimension.

The numerical text semantic dimension is one of the target text semantic dimensions, and mainly comprises numerical information in a chemical detection report, such as concentration, PH value and the like of chemicals. The word frequent item variable is a measure of the frequency of occurrence of each word unit (e.g., keyword, phrase, etc.) in a chemical detection report, and may also be understood as the importance of that word unit throughout the report. The detection element thermodynamic mapping is a method of converting various elements (e.g., chemical names, detection results, etc.) in a detection report into a visual chart (e.g., thermodynamic diagram) for visually displaying which elements are more important or more sensitive.

For example, there is a report of intermediate chemical detection in a number of text semantic dimensions, which contains the detection result of "sulfuric acid", the PH of which is 1, and the frequency of occurrence is 10.

In step 22211, the maximum and minimum word frequent term variables of the report are first determined, say, the maximum possible "sulfuric acid" and the minimum possible "pH". Then, in step 22212, the word frequent item variable of each word unit is updated based on the comparison result of the word frequent item variable of each word unit and the minimum word frequent item variable. For example, if the occurrence frequency of "sulfuric acid" is found to be far higher than "pH", the word frequent item variable of "sulfuric acid" can be lowered, and the word frequent item variable of "pH" can be raised. Next, in step 22213, the updated word frequent item variable is projected into the target word frequent item variable quadrant, resulting in a second chemical detection report. This step can help to better understand which words are more important in the overall report. Then, in step 22214, the second chemical detection report is subjected to detection element thermal mapping to obtain a third chemical detection report. This step can help to more intuitively see which detection elements are more sensitive or important. Finally, in step 22215, the portion meeting the first pre-configured sensitivity requirement is extracted from the third chemical detection report as the target chemical detection report in the numeric text semantic dimension.

Thus, by continuously screening and optimizing the data in each step, the interference of invalid data can be effectively reduced, thereby improving the efficiency of data processing. The data can be more intuitively and clearly understood through the updating and projection of the word frequent item variable and the thermal mapping of the detection element, so that the readability of the data is improved. By pre-configuring the sensitivity requirement, only the truly sensitive information can be extracted and encrypted, so that the security of the sensitive information is effectively protected.

Under some preferred design considerations, extracting a detection report in step 22215 having a detection report sensitivity that meets the first pre-configured sensitivity requirement from the third chemical detection report as a target chemical detection report in the numerical text semantic dimension includes: determining a first report text set including chemical property descriptions in the third chemical detection report; determining word frequent item average variables of each word unit in the first report text set; and extracting a detection report with the average variable of the word frequent items higher than a first threshold value from the third chemical detection report, and taking the detection report as a target chemical detection report under the numerical text semantic dimension.

Wherein the chemical property descriptive term is an item or indicator describing the chemical property, such as odor, color, solubility, toxicity, etc. The first report text set is extracted from the entire test report and contains a text block set of chemical property description items. The word frequent item average variable is a numerical value obtained by averaging the word frequent item variables of all word units in a text set. The first threshold is a preset threshold for screening out information requiring special processing.

For example, in the third chemical detection report there are the following paragraphs: the results of the "sulfuric acid is highly corrosive", "2022, 3-month, 1-day test show that the sample ph=1" and "the chemical has extremely high toxicity". Wherein the two paragraphs "sulfuric acid is highly corrosive" and "the chemical is extremely toxic" include chemical performance descriptive terms, so they constitute the first report text set. Next, a word frequent term average variable for each word unit in the first report text set is determined. For example, the word frequent term variables of "sulfuric acid", "strong corrosiveness", "chemicals", and "extremely high toxicity" may be 10, 8, 15, and 12, respectively, and then the word frequent term average variable is (10+8+15+12)/4=11.25. Then, the first threshold value is set to 10, so that in the third chemical detection report, only detection reports with word frequent item average variable higher than 10 are extracted. That is, the paragraph "the chemical has extremely high toxicity" will be used as a report of target chemical detection in the numeric text semantic dimension.

Therefore, by setting the threshold value, the information needing special treatment can be screened out from a large number of chemical detection reports more accurately, and excessive treatment on inconsequential information is avoided. Through the calculation of the word frequent term average variable, the importance degree of each word in a report text set can be intuitively reflected, and the user is helped to better understand and interpret the data. And the information meeting the requirements is extracted from the preliminary report, so that the workload of the subsequent processing stage can be greatly reduced, and the operation efficiency of the whole system is improved.

In some examples, the number of target text semantic dimensions includes a chemical formula text semantic dimension based on which detection reports having detection report sensitivities meeting a pre-configured sensitivity requirement are extracted from the intermediate chemical detection reports under each of the target text semantic dimensions in step 2221 as target chemical detection reports under each of the target text semantic dimensions, including steps 2221 a-2221 e.

Step 2221a, determining a target chemical formula description semantic value corresponding to the intermediate chemical detection report in the chemical formula text semantic dimension, where the target chemical formula description semantic value is a target statistical value of chemical formula description semantic values of each word unit in the intermediate chemical detection report in the chemical formula text semantic dimension.

Step 2221b, determining a target chemical formula description semantic value quadrant based on the target chemical formula description semantic value.

Step 2221c, changing the chemical formula description semantic value of the intermediate chemical formula description semantic value in the chemical formula text semantic dimension, which is not in the chemical formula description semantic value quadrant, to a specified value, obtaining a fourth chemical formula detection report, and obtaining the minimum chemical formula description semantic value of the chemical formula description semantic value in the fourth chemical formula detection report, which is not 0, as the target chemical formula description semantic value.

Step 2221d, based on the comparison result of the chemical formula description semantic value of each word unit with the chemical formula description semantic value not being 0 in the fourth chemical detection report and the target chemical formula description semantic value, updating the chemical formula description semantic value of each word unit with the chemical formula description semantic value not being 0 in the fourth chemical detection report, so as to obtain a fifth chemical detection report.

Step 2221e, extracting a detection report with the detection report sensitivity meeting the second pre-configuration sensitivity requirement from the fifth chemical detection report, and using the detection report as a target chemical detection report under the semantic dimension of the chemical formula text.

Among them, the chemical text semantic dimension is a type of the target text semantic dimension, and mainly focuses on text blocks containing chemical information such as chemical structures or reactions. The chemical formula description semantic value is a measure of the chemical nature or behavior depicted for each word unit (e.g., keyword, phrase, etc.) in a chemical detection report. The target chemical formula description semantic value is a statistical target value, such as an average value or median, for a set of chemical formula description semantic values. The target formula describes the semantic value quadrant: according to the target chemical formula description semantic value, the distribution of the chemical formula description semantic value is divided into different areas or quadrants.

For example, there are the following paragraphs in the intermediate chemical detection report in the text semantic dimension of the formula: "sulfuric acid (H2 SO 4) is strongly corrosive", "aqueous ammonia (NH3.H2O) is an alkaline substance" and "formaldehyde (CH 2O) is strongly malodorous". Wherein, "sulfuric acid (H2 SO 4)", "ammonia water (NH3.H2O)", and "formaldehyde (CH 2O)" are chemical formulas describing semantic values of the respective word units.

In step 2221a, target statistics, such as average or median, of the chemical formula description semantic values are determined. Then, in step 2221b, a target chemical formula description semantic value quadrant is determined based on this target statistic. For example, a higher than average chemical formula description semantic value may be set as the first quadrant and a lower than average chemical formula description semantic value may be set as the second quadrant. Next, in step 2221c, the chemical formula description semantic value that is not in the target chemical formula description semantic value quadrant is changed to a specified value (e.g., 0), resulting in a fourth chemical detection report. Meanwhile, the minimum chemical formula description semantic value with the chemical formula description semantic value not being 0 in the fourth chemical detection report is obtained and used as a new target chemical formula description semantic value. Then, in step 2221d, the chemical formula description semantic value in the fourth chemical detection report is updated based on the comparison result of the chemical formula description semantic value in the fourth chemical detection report and the new target chemical formula description semantic value, to obtain a fifth chemical detection report. Finally, in step 2221e, the portion meeting the second pre-configured sensitivity requirement is extracted from the fifth chemical detection report as the target chemical detection report in the semantic dimension of the chemical formula text.

Therefore, by extracting the data meeting the specific sensitivity requirement, the obtained target chemical detection report can be ensured to have high quality, and the actual situation is accurately reflected. Optimizing resource allocation: when a large amount of data is processed, the use of calculation and storage resources can be optimized by setting reasonable thresholds and indexes, and the system performance is improved. For sensitive chemical detection reports, in this way it is ensured that only reports meeting preset sensitivity requirements are further processed and transmitted, thereby effectively preventing sensitive information from leaking.

Under other preferred design considerations, extracting a detection report having a detection report sensitivity meeting a second pre-formulation sensitivity requirement from the fifth chemical detection report in step 2221e as a target chemical detection report in the text semantic dimension of the chemical formula includes: determining a second report text set including chemical property descriptions in the fifth chemical detection report; determining the proportion of word units with the semantic value of the chemical formula description not being 0 in the second report text set as the proportion of effective word units; determining a chemical formula description semantic value quadrant based on the maximum chemical formula description semantic value and the minimum chemical formula description semantic value of the second report text set; disassembling the chemical formula description semantic value quadrant into a plurality of local numerical value quadrants; decomposing the chemical formula description semantic values of the chemical formula description semantic values in the same local numerical quadrant in the second report text set into a linear characteristic variable set to obtain a plurality of linear characteristic variable sets; taking the linear characteristic variable set with the number of the chemical formula description semantic values reaching a second threshold value in the plurality of linear characteristic variable sets as a target linear characteristic variable set; and extracting a detection report that the proportion of the effective word units reaches a third threshold value and the number of the target linear characteristic variable sets reaches a fourth threshold value from the fifth chemical detection report, and taking the detection report as a target chemical detection report under the semantic dimension of the chemical formula text.

Wherein the second report text set is extracted from the entire test report and contains another type of text block set of chemical property description items. The effective word unit proportion is the proportion of word units with the semantic value of 0 in all word units in a report text set. The local numerical quadrant is a small range interval obtained by further subdividing the chemical formula description semantic value quadrant. The linear characteristic variable set is a group of characteristic variables obtained by linearly converting the chemical formula description semantic value.

For example, in the fifth chemical detection report there are the following paragraphs: the H2O is colorless and odorless liquid, the CO2 is colorless gas at normal temperature, carbonic acid can be formed by dissolving the liquid in water, and the H2SO4 is colorless and transparent oily liquid, SO that the liquid has strong corrosiveness to human bodies. Wherein, the two paragraphs of 'H2O is colorless and odorless liquid' and 'H2 SO4 is colorless and transparent oily liquid and has strong corrosiveness to human body' contain chemical performance description items, SO that the two paragraphs form a second report text set. Next, the proportion of word units with the semantic value of the chemical formula description not being 0 in the second report text set is determined as the effective word unit proportion. Then, based on the maximum chemical formula description semantic value and the minimum chemical formula description semantic value of the second report text set, determining a chemical formula description semantic value quadrant, and disassembling the chemical formula description semantic value quadrant into a plurality of local numerical value quadrants. And then, decomposing the chemical formula description semantic values in the same local numerical quadrant of the chemical formula description semantic values in the second report text set into a linear characteristic variable set to obtain a plurality of linear characteristic variable sets. And then, taking the linear characteristic variable set with the number of the chemical formula description semantic values reaching a second threshold value in the plurality of linear characteristic variable sets as a target linear characteristic variable set. And finally, extracting a detection report that the proportion of the effective word units reaches a third threshold value and the number of the target linear characteristic variable sets reaches a fourth threshold value from the fifth chemical detection report, and taking the detection report as a target chemical detection report under the semantic dimension of the chemical formula text.

It can be seen that this solution allows more accurate screening of information requiring special treatment from a large number of chemical detection reports by setting a threshold value and using a set of linear characteristic variables. The complex chemical formula description semantic values are disassembled into the linear characteristic variable sets, so that the data processing process is more visual and easier to carry out. Through the preset sensitivity requirement, only the detection report meeting the specific condition can be extracted, and the false leakage of sensitive information is effectively avoided.

Further, there is also provided a computer-readable storage medium having stored thereon a program which, when executed by a processor, implements the above-described method.

In the embodiments provided in the present application, it should be understood that the disclosed apparatus and method may be implemented in other manners. The apparatus and method embodiments described above are merely illustrative, for example, flow diagrams and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of apparatus, methods and computer program products according to various embodiments of the present application. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.

In addition, the functional modules in the embodiments of the present application may be integrated together to form a single part, or each module may exist alone, or two or more modules may be integrated to form a single part.

The functions, if implemented in the form of software functional modules and sold or used as a stand-alone product, may be stored in a computer-readable storage medium. Based on such understanding, the technical solution of the present application may be embodied essentially or in a part contributing to the prior art or in a part of the technical solution, in the form of a software product stored in a storage medium, comprising several instructions for causing a computer device (which may be a personal computer, a network device, or the like) to perform all or part of the steps of the methods described in the embodiments of the present application. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a random access Memory (RAM, random Access Memory), a magnetic disk, or an optical disk, or other various media capable of storing program codes. It should be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.

The foregoing description is only of the preferred embodiments of the present application and is not intended to limit the same, but rather, various modifications and variations may be made by those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present application should be included in the protection scope of the present application.

Claims

1. A chemical detection report data encryption method based on big data, characterized in that it is applied to an artificial intelligence encryption system, the method comprising:

2. The method of claim 1, wherein the step of tuning the detection report encryption decision network comprises:

3. The method of claim 2, wherein said passing the detection report semantics Guan Jipu into a semantic matrix mapping subnet in the underlying encryption decision network results in a target report element semantic matrix generated by the semantic matrix mapping subnet, comprising:

4. The method of claim 3, wherein said calibrating the underlying encryption decision network based on the target reporting element semantic matrix to obtain the detection report encryption decision network comprises:

5. The method of claim 2, wherein said processing the chemical detection report in each of said target text semantic dimensions through a semantic knowledge mining subnet in the underlying encryption decision network corresponding to each of said target text semantic dimensions to obtain a detection report semantic relationship spectrum for the chemical detection report corresponding to each of said target text semantic dimensions, comprises:

6. The method of claim 5, wherein said processing intermediate chemical detection reports in each of said target text semantic dimensions through a semantic knowledge mining subnet in said underlying cryptographic decision network corresponding to each of said target text semantic dimensions to obtain a detection report semantic relationship spectrum for chemical detection reports corresponding to each of said target text semantic dimensions, comprises:

7. The method of claim 6, wherein the plurality of target text semantic dimensions includes a numeric text semantic dimension, wherein the extracting a detection report having a detection report sensitivity that meets a pre-configured sensitivity requirement from the intermediate chemical detection reports in each of the target text semantic dimensions as a target chemical detection report in each of the target text semantic dimensions comprises:

extracting a detection report with detection report sensitivity meeting the first pre-configuration sensitivity requirement from the third chemical detection report as a target chemical detection report under the numerical text semantic dimension;

wherein, the extracting the detection report with the detection report sensitivity meeting the first preset sensitivity requirement from the third chemical detection report as the target chemical detection report under the numerical text semantic dimension comprises: determining a first report text set including chemical property descriptions in the third chemical detection report; determining word frequent item average variables of each word unit in the first report text set;

8. The method of claim 6, wherein the plurality of target text semantic dimensions includes a chemical formula text semantic dimension, wherein the extracting a detection report having a detection report sensitivity that meets a pre-configured sensitivity requirement from the intermediate chemical detection report in each of the target text semantic dimensions as a target chemical detection report in each of the target text semantic dimensions comprises:

extracting a detection report with detection report sensitivity meeting the second pre-configuration sensitivity requirement from the fifth chemical detection report as a target chemical detection report under the chemical formula text semantic dimension;

wherein, the extracting the detection report with the detection report sensitivity meeting the second pre-configuration sensitivity requirement from the fifth chemical detection report as the target chemical detection report under the semantic dimension of the chemical formula text comprises:

9. An artificial intelligence encryption system, comprising a processor and a memory; the processor is communicatively connected to the memory, the processor being configured to read a computer program from the memory and execute the computer program to implement the method of any of claims 1-8.

10. A computer readable storage medium, characterized in that a program is stored thereon, which program, when being executed by a processor, implements the method of any of claims 1-8.