Digital assessment method and system for supervision quality

Abstract

The embodiment of the invention provides a digital assessment method and system for supervision quality, and belongs to the technical field of electric power engineering supervision quality verification. The digital evaluation method comprises the following steps: acquiring a current supervision quality report to be evaluated; segmenting the supervision quality report to obtain a text material of each project unit; identifying each text material by adopting a text identification method to obtain a digital evaluation index of each item; performing threshold judgment on the digital evaluation index to obtain a digital matrix corresponding to each project unit; combining the digitized matrixes to obtain a digitized total matrix corresponding to the supervision quality report; constructing a weight matrix of the digital total matrix; and calculating an evaluation value of the supervision quality report according to the digital total matrix and the weight matrix. According to the digital evaluation method and system, the evaluation accuracy of supervision quality can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of power engineering supervision quality verification, and in particular to a digital evaluation method and system for supervision quality. Background Art

[0002] In the field of power engineering construction, the assessment of supervision quality is a key link to ensure that the project is implemented in accordance with established standards and specifications. Traditional supervision quality assessment methods usually rely on the review of multiple experts, who review different stages and aspects of power engineering based on their own professional knowledge, experience, and familiarity with relevant regulations and standards. However, this assessment method that relies on subjective judgment has certain limitations.

[0003] First, due to the different knowledge backgrounds, experiences and personal preferences of each expert, their understanding and evaluation criteria of power project quality may be different, which leads to subjectivity and inconsistency in the evaluation results. Secondly, the information exchange and coordination in the expert review process is often complicated and requires a lot of time and resources. In addition, the subjective judgment of experts may be affected by external factors, such as the relationship with the contractor, time pressure, etc., which may affect the fairness and accuracy of the evaluation.

[0004] In order to improve the objectivity and standardization of supervision quality assessment, the industry has been exploring more scientific and systematic assessment methods. Among them, a common practice is to introduce quantitative assessment indicators and scoring systems to reduce the impact of subjective judgment by setting specific quantitative standards. For example, checklists, scorecards or performance-based assessment tools can be used to convert various aspects of power engineering quality into quantifiable indicators, making the assessment process more transparent and comparable.

[0005] In addition, with the development of information technology, some advanced data analysis methods and tools have also been applied to supervision quality assessment. For example, using big data analysis technology to analyze the data of historical power engineering projects can identify potential quality problems and risk factors and provide reference for current power engineering. At the same time, artificial intelligence and machine learning algorithms are also used to assist experts in making decisions. By analyzing a large amount of power engineering data, possible problem areas can be automatically identified and improvement suggestions can be provided.

[0006] Despite these attempts and methods, supervision quality assessment still faces challenges. How to balance expert experience and objective data, how to ensure the applicability and effectiveness of assessment tools and methods, and how to deal with special situations that arise in complex power engineering projects are all issues that need further research and resolution. Therefore, establishing a supervision quality assessment system that can fully utilize expert knowledge and reduce the impact of subjectivity remains an important topic in the field of power engineering construction. Summary of the invention

[0007] The purpose of the embodiments of the present invention is to provide a digital evaluation method and system for supervision quality, which can improve the evaluation accuracy of supervision quality.

[0008] In order to achieve the above object, an embodiment of the present invention provides a digital evaluation method for supervision quality, comprising:

[0009] Obtain the supervision quality report currently awaiting evaluation;

[0010] Divide the supervision quality report into text materials for each project unit;

[0011] Using a text recognition method to recognize each of the text materials to obtain a digital evaluation index for each project;

[0012] Performing threshold judgment on the digital evaluation index to obtain a digital matrix corresponding to each of the project units;

[0013] Combining each of the digital matrices to obtain a digital total matrix corresponding to the supervision quality report;

[0014] Constructing a weight matrix of the digitalized total matrix;

[0015] The evaluation value of the supervision quality report is calculated based on the digital total matrix and the weight matrix.

[0016] Optionally, the supervision quality report is segmented, and the text materials of each project unit include:

[0017] Determining the title of each of the project units based on a keyword search;

[0018] Determine the actual position of the project unit according to the position of the title and the position of the next title;

[0019] The supervision quality report is divided according to the actual location to obtain text materials for each of the project units.

[0020] Optionally, a text recognition method is used to recognize each of the text materials to obtain a digital evaluation index for each project, including:

[0021] Dividing the text material into equal lengths to obtain a plurality of text regions;

[0022] Traversing each of the text regions to determine whether text information can be directly captured in the text region;

[0023] In the case where it is determined that the text information can be directly captured in the text area, directly capturing the text information;

[0024] In the case where it is determined that the text area cannot directly capture text information, an OCR text recognition method is used to capture the text information;

[0025] The digital evaluation index is determined according to the text information using a keyword matching method.

[0026] Optionally, constructing a weight matrix of the digitized total matrix includes:

[0027] Obtaining a sample set of supervision quality reports, wherein the sample set includes multiple supervision quality reports and corresponding evaluation values;

[0028] Initializing the weight matrix, wherein the weight matrix includes a plurality of sub-matrices, each of the sub-matrices corresponds one-to-one to the digitized matrix, and each element in each of the sub-matrices corresponds one-to-one to each element in the digitized matrix;

[0029] Taking the digital total matrix corresponding to the supervision quality report as input and the corresponding evaluation value as output, primary training is performed on the weight matrix to obtain a primary optimal solution of the weight matrix;

[0030] Determining whether the primary optimal solution meets a preset accuracy requirement;

[0031] When it is determined that the primary optimal solution does not meet the accuracy requirement, secondary training is performed on the weight matrix to obtain a secondary optimal solution, and the secondary optimal solution is used as the weight matrix that is finally output.

[0032] Optionally, the digital evaluation method comprises:

[0033] The weight matrix is ​​determined using the following formula (1) and formula (2):

[0034] S={α1A1,α2A2,…,α i A i ,…,α m A m}, (1)

[0035] A i =(β1,β2,β3,…β k ,…,β K ), (2)

[0036] Where S is the weight matrix, A iis the set of secondary weights of the i-th digitized matrix, i=1,2,…,m, m is the number of digitized matrices of the digitized total matrix, α i is the first-level weight of the i-th submatrix, β k is the secondary weight of the kth digitization evaluation index, k=1, 2, …, K, and K is the number of digitization evaluation indicators in a digitization matrix.

[0037] Optionally, performing primary training on the weight matrix includes:

[0038] Get the distance between the expected evaluation value of the current output and the actual evaluation value;

[0039] Update the primary weight of the weight matrix according to the distance;

[0040] Determine whether the current number of iterations is greater than or equal to a preset number threshold;

[0041] When it is determined that the number of iterations is greater than or equal to the number threshold, selecting the smallest distance from the weight matrix as the optimal solution output, and determining that the primary training is completed;

[0042] When it is determined that the number of iterations is less than the number threshold, the process returns to the step of obtaining the distance between the expected evaluation value of the current output and the actual evaluation value.

[0043] Optionally, updating the primary weight of the weight matrix according to the distance includes:

[0044] Update each of the first-level weights according to formula (3) and formula (4),

[0045] α ′ =α+(-1) σ ·Δα, (3)

[0046]

[0047] Among them, α ′ is the first-level weight after update, α is the first-level weight before update, σ is the negative feedback operator, Δα is the update step length, d i is the distance at the current iteration number, d i-1 is the distance under the previous iteration.

[0048] Optionally, performing secondary training on the weight matrix to obtain a secondary optimal solution includes:

[0049] Get the distance between the expected evaluation value of the current output and the actual evaluation value;

[0050] Copying a plurality of the weight matrices to obtain a population of weight matrices;

[0051] Traversing each of the weight matrices, randomly selecting a set of unselected secondary weights;

[0052] updating the secondary weights in the set of secondary weights according to the distance to obtain a new population;

[0053] Determine the difference between the minimum distance in the new population and the minimum distance in the population before the update;

[0054] Determining whether the difference is less than or equal to a preset difference threshold;

[0055] When it is determined that the difference is less than or equal to the difference threshold, the weight matrix corresponding to the minimum value is selected as the optimal solution output;

[0056] When it is determined that the difference is greater than the difference threshold, the step of returning to traverse each of the weight matrices and randomly selecting a set of unselected secondary weights is performed.

[0057] Optionally, updating the secondary weights in the set of secondary weights according to the distance comprises:

[0058] β ′ =β+(-1) σ ·Δβ, (3)

[0059]

[0060] Among them, β ′ is the secondary weight after update, β is the secondary weight before update, σ is the negative feedback operator, Δβ is the update step length, d i is the distance at the current iteration number, d i-1 is the distance under the previous iteration.

[0061] On the other hand, the present invention further provides a digital evaluation system for supervision quality, the digital evaluation system comprising a processor, and the processor is configured to execute any of the digital evaluation methods described above.

[0062] Through the above technical solution, the embodiment of the present invention provides a digital evaluation method and system for supervision quality, which divides each part of the supervision quality report, and determines the digital evaluation index in combination with each divided part, and finally calculates the evaluation value of the supervision quality report by matrix weighting. Compared with the prior art, since the weight matrix is ​​trained based on historical evaluation values, the finally calculated evaluation value can accurately and fairly reflect the overall quality of the supervision quality report while overcoming the subjective influence of human beings.

[0063] Other features and advantages of the embodiments of the present invention will be described in detail in the subsequent detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] The accompanying drawings are used to provide a further understanding of the embodiments of the present invention and constitute a part of the specification. Together with the following specific implementations, they are used to explain the embodiments of the present invention, but do not constitute a limitation on the embodiments of the present invention. In the accompanying drawings:

[0065] Figure 1 is a flow chart of a digital evaluation method for supervision quality according to one embodiment of the present invention;

[0066] Figure 2 is a flow chart of a method for segmenting the project unit according to one embodiment of the present invention;

[0067] Figure 3 is a flow chart of a method for obtaining a digital evaluation index according to an embodiment of the present invention;

[0068] Figure 4 is a flow chart of a method for constructing a weight matrix of a digitized total matrix according to one embodiment of the present invention;

[0069] Figure 5 is a flowchart of a specific method of primary training according to one embodiment of the present invention; and

[0070] Figure 6 is a flowchart of a method for secondary training according to one embodiment of the present invention. DETAILED DESCRIPTION

[0071] The specific implementation of the embodiment of the present invention is described in detail below in conjunction with the accompanying drawings. It should be understood that the specific implementation described here is only used to illustrate and explain the embodiment of the present invention, and is not used to limit the embodiment of the present invention.

[0072] like Figure 1 FIG. 1 is a flow chart of a digital evaluation method for supervision quality according to an embodiment of the present invention. Figure 1 In the digital evaluation method, the following steps may be included:

[0073] In step S10, the supervision quality report to be evaluated is obtained;

[0074] In step S11, the supervision quality report is segmented into text materials for each project unit;

[0075] In step S12, each text material is recognized using a text recognition method to obtain a digital evaluation index for each project;

[0076] In step S13, a threshold value is determined for the digital evaluation index to obtain a digital matrix corresponding to each project unit;

[0077] In step S14, each digital matrix is ​​combined to obtain a digital total matrix corresponding to the supervision quality report;

[0078] In step S15, a weight matrix of the digitized total matrix is ​​constructed;

[0079] In step S16, the evaluation value of the supervision quality report is calculated based on the digitized total matrix and the weight matrix.

[0080] In this Figure 1 In the digital evaluation method shown, step S10 can be used to obtain the supervision quality report to be evaluated. The supervision quality report can be in a form known to those skilled in the art, including but not limited to basic project information, supervision work overview, construction quality records, engineering problems and suggestions, quality accident reports, regulations and contract documents, engineering progress records, etc.

[0081] Step S11 can be used to segment the supervision quality report into text materials of each project unit. The specific method for segmenting the project unit can be various forms known to those skilled in the art. In one example of the present invention, the method for segmenting the project unit can further include: Figure 2 In the steps shown in Figure 2 In the method of dividing the project unit, the method may include the following steps:

[0082] In step S20, the title of each project unit is determined based on keyword search, wherein the keyword search may be preset with multiple standard keywords, and then the title of each project unit is matched based on the preset multiple keywords, so as to determine the location of each title.

[0083] In step S21, the actual position of the item unit is determined according to the position of the title and the position of the next title;

[0084] In step S22, the supervision quality report is segmented according to the actual location to obtain text materials for each project unit.

[0085] Step S12 may be used to recognize each text material using a text recognition method to obtain a digital evaluation index for each project. The specific steps of the method for obtaining the digital evaluation index may be in various forms known to those skilled in the art. In one example of the present invention, the method for obtaining the digital evaluation index may include: Figure 3 In the steps shown in Figure 3The method for obtaining the digital evaluation index may include the following steps:

[0086] In step S30, the text material is segmented into equal lengths to obtain a plurality of text regions;

[0087] In step S31, each text area is traversed to determine whether text information can be directly captured in the text area;

[0088] In step S32, if it is determined that the text information can be directly captured in the text area, the text information is directly captured;

[0089] In step S33, when it is determined that the text area cannot directly capture text information, an OCR text recognition method is used to capture text information;

[0090] In step S34, a keyword matching method is used to determine a digital evaluation index based on the text information.

[0091] In this Figure 3 In the method shown, step S30 can be used to divide the text material into equal lengths to obtain multiple text areas. The specific method of the equal length division can be, for example, to determine the total row and column values ​​or the total height of the text material according to the row and column values ​​or the height value of the document, and then divide the text material into multiple text areas according to the principle of average division. Steps S31 to S33 can be used to scan each text area separately and identify the corresponding text information separately and specifically. Since the supervision report materials generally include various types of text materials, including but not limited to text, images, tables, etc., these information are arranged continuously together. Therefore, in this Figure 3 In the step S30, the text material is first divided into a plurality of text regions, thereby preliminarily dividing different types of information. Then, different types of information are targetedly identified through steps S31 to S33, thereby obtaining text information. Among them, the OCR text recognition method used in step S33 is a commonly used text recognition method in the art, and the specific process is not repeated here.

[0092] Step S15 can be used to construct a weight matrix of the digitized total matrix. Specifically, in this embodiment, step S15 can further include: Figure 4 In the method shown in Figure 4 In the step S15, the step S15 may include the following steps:

[0093] In step S40, a sample set of supervision quality reports is obtained, wherein the sample set includes a plurality of supervision quality reports and corresponding evaluation values;

[0094] In step S41, a weight matrix is ​​initialized, wherein the weight matrix may include a plurality of sub-matrices, each sub-matrix corresponds to a digitized matrix one-to-one, and each element in each sub-matrix corresponds to each element in the digitized matrix one-to-one. In an example of the present invention, the weight matrix may also be expressed by the following formula (1) and formula (2):

[0095] S={α1A1,α2A2,…,α i A i ,…,α m A m}, (1)

[0096] A i =(β1,β2,β3,…β k ,…,β K ), (2)

[0097] Among them, S is the weight matrix, A i is the set of secondary weights of the i-th digitized matrix, i = 1, 2, ..., m, m is the number of digitized matrices in a digitized total matrix, α i is the first-level weight of the I-th submatrix, β k is the secondary weight of the kth digitization evaluation index, k=1, 2, …, K, and K is the number of digitization evaluation indicators in a digitization matrix.

[0098] In step S42, the digital total matrix corresponding to the supervision quality report is used as input, and the corresponding evaluation value is used as output, and the weight matrix is ​​preliminarily trained to obtain a preliminarily optimal solution of the weight matrix. The specific process of the preliminarily training can be in various forms known to those skilled in the art. In one example of the present invention, the specific method of the preliminarily training can include: Figure 5 In the steps shown in Figure 5 The specific method of the primary training may include the following steps:

[0099] In step S50, the distance between the expected evaluation value currently output and the actual evaluation value is obtained;

[0100] In step S51, the primary weight of the weight matrix is ​​updated according to the distance. The method for updating the primary weight can be various forms known to those skilled in the art. In this example, the primary weight can be updated using the following formula (3) and formula (4):

[0101] α ′ =α+(-1) σ ·Δα, (3)

[0102]

[0103] Among them, α ′ is the first-level weight after update, α is the first-level weight before update, σ is the negative feedback operator, Δα is the update step size, d i is the distance at the current iteration number, d i-1 is the distance under the previous iteration.

[0104] In step S52, it is determined whether the current number of iterations is greater than or equal to a preset number threshold;

[0105] In step S53, when it is determined that the number of iterations is greater than or equal to the number threshold, the smallest distance is selected from the weight matrix as the optimal solution output, and it is determined that the primary training is completed;

[0106] When it is determined that the number of iterations is less than the number threshold, the process returns to the step of obtaining the distance between the expected evaluation value of the current output and the actual evaluation value.

[0107] In step S43, it is determined whether the primary optimal solution meets the preset accuracy requirement;

[0108] In step S44, if it is determined that the primary optimal solution does not meet the accuracy requirement, the weight matrix is ​​subjected to secondary training to obtain a secondary optimal solution, and the secondary optimal solution is used as the weight matrix for the final output. The specific method of the secondary training may be a form known to those skilled in the art. In one example of the present invention, the secondary training method may include: Figure 6 In the steps shown in Figure 6 In the method of secondary training, the secondary training method may include the following steps:

[0109] In step S60, the distance between the expected evaluation value currently output and the actual evaluation value is obtained;

[0110] In step S61, a plurality of weight matrices are copied to obtain a population of weight matrices;

[0111] In step S62, each weight matrix is ​​traversed and a set of unselected secondary weights is randomly selected;

[0112] In step S63, the secondary weights in the set of secondary weights are updated according to the distance to obtain a new population. The specific method for updating the secondary weights may be in various forms known to those skilled in the art. In one example of the present invention, the method for updating the secondary weights may be to use the following formulas (5) and (6) to update:

[0113] β ′ =β+(-1) σ ·Δβ, (5)

[0114]

[0115] Among them, β ′ is the updated secondary weight, β is the secondary weight before update, σ is the negative feedback operator, Δβ is the update step size, d i is the distance at the current iteration number, d i-1 is the distance under the previous iteration.

[0116] In step S64, the difference between the minimum value of the distance in the new population and the minimum value in the population before the update is determined;

[0117] In step S65, it is determined whether the difference is less than or equal to a preset difference threshold;

[0118] In step S66, when it is determined that the difference is less than or equal to the difference threshold, the weight matrix corresponding to the minimum value is selected as the optimal solution output;

[0119] When it is determined that the difference is greater than the difference threshold, the process returns to the step of traversing each weight matrix and randomly selecting a set of unselected secondary weights, that is, returns to step S62.

[0120] In step S45, when it is determined that the primary optimal solution meets the accuracy requirement, the primary optimal solution is used as the final output weight matrix.

[0121] On the other hand, the present invention also provides a digital evaluation system for supervision quality, the digital evaluation system comprising a processor, the processor being configured to execute any of the digital evaluation methods described above. Specifically, the digital evaluation method may include the following steps:

[0122] In step S10, the supervision quality report to be evaluated is obtained;

[0123] In step S11, the supervision quality report is segmented into text materials for each project unit;

[0124] In step S12, each text material is recognized using a text recognition method to obtain a digital evaluation index for each project;

[0125] In step S13, a threshold value is determined for the digital evaluation index to obtain a digital matrix corresponding to each project unit;

[0126] In step S14, each digital matrix is ​​combined to obtain a digital total matrix corresponding to the supervision quality report;

[0127] In step S15, a weight matrix of the digitized total matrix is ​​constructed;

[0128] In step S16, the evaluation value of the supervision quality report is calculated based on the digitized total matrix and the weight matrix.

[0129] In this Figure 1 In the digital evaluation method shown, step S10 can be used to obtain the supervision quality report to be evaluated. The supervision quality report can be in a form known to those skilled in the art, including but not limited to basic project information, supervision work overview, construction quality records, engineering problems and suggestions, quality accident reports, regulations and contract documents, engineering progress records, etc.

[0130] Step S11 can be used to segment the supervision quality report into text materials of each project unit. The specific method for segmenting the project unit can be various forms known to those skilled in the art. In one example of the present invention, the method for segmenting the project unit can further include: Figure 2 In the steps shown in Figure 2 In the method of dividing the project unit, the method may include the following steps:

[0131] In step S20, the title of each project unit is determined based on keyword search, wherein the keyword search may be preset with multiple standard keywords, and then the title of each project unit is matched based on the preset multiple keywords, so as to determine the location of each title.

[0132] In step S21, the actual position of the item unit is determined according to the position of the title and the position of the next title;

[0133] In step S22, the supervision quality report is segmented according to the actual location to obtain text materials for each project unit.

[0134] Step S12 may be used to recognize each text material using a text recognition method to obtain a digital evaluation index for each project. The specific steps of the method for obtaining the digital evaluation index may be in various forms known to those skilled in the art. In one example of the present invention, the method for obtaining the digital evaluation index may include: Figure 3 In the steps shown in Figure 3 The method for obtaining the digital evaluation index may include the following steps:

[0135] In step S30, the text material is segmented into equal lengths to obtain a plurality of text regions;

[0136] In step S31, each text area is traversed to determine whether text information can be directly captured in the text area;

[0137] In step S32, if it is determined that the text information can be directly captured in the text area, the text information is directly captured;

[0138] In step S33, when it is determined that the text area cannot directly capture text information, an OCR text recognition method is used to capture text information;

[0139] In step S34, a keyword matching method is used to determine a digital evaluation index based on the text information.

[0140] In this Figure 3 In the method shown, step S30 can be used to divide the text material into equal lengths to obtain multiple text areas. The specific method of the equal length division can be, for example, to determine the total row and column values ​​or the total height of the text material according to the row and column values ​​or the height value of the document, and then divide the text material into multiple text areas according to the principle of average division. Steps S31 to S33 can be used to scan each text area separately and identify the corresponding text information separately and specifically. Since the supervision report materials generally include various types of text materials, including but not limited to text, images, tables, etc., these information are arranged continuously together. Therefore, in this Figure 3 In the step S30, the text material is first divided into a plurality of text regions, thereby preliminarily dividing different types of information. Then, different types of information are targetedly identified through steps S31 to S33, thereby obtaining text information. Among them, the OCR text recognition method used in step S33 is a commonly used text recognition method in the art, and the specific process is not repeated here.

[0141] Step S15 can be used to construct a weight matrix of the digitized total matrix. Specifically, in this embodiment, step S15 can further include: Figure 4 In the method shown in Figure 4 In the step S15, the step S15 may include the following steps:

[0142] In step S40, a sample set of supervision quality reports is obtained, wherein the sample set includes a plurality of supervision quality reports and corresponding evaluation values;

[0143] In step S41, a weight matrix is ​​initialized, wherein the weight matrix may include a plurality of sub-matrices, each sub-matrix corresponds to a digitized matrix one-to-one, and each element in each sub-matrix corresponds to each element in the digitized matrix one-to-one. In an example of the present invention, the weight matrix may also be expressed by the following formula (1) and formula (2):

[0144] S={α1A1,α2A2,…,α i A i ,…,αm A m}, (1)

[0145] A i =(β1,β2,β3,…β k ,…,β K ), (2)

[0146] Among them, S is the weight matrix, A i is the set of secondary weights of the i-th digitized matrix, i = 1, 2, ..., m, m is the number of digitized matrices in a digitized total matrix, α i is the first-level weight of the i-th submatrix, β k is the secondary weight of the kth digitization evaluation index, k=1, 2, …, K, and K is the number of digitization evaluation indicators in a digitization matrix.

[0147] In step S42, the digital total matrix corresponding to the supervision quality report is used as input, and the corresponding evaluation value is used as output, and the weight matrix is ​​preliminarily trained to obtain a preliminarily optimal solution of the weight matrix. The specific process of the preliminarily training can be in various forms known to those skilled in the art. In one example of the present invention, the specific method of the preliminarily training can include: Figure 5 In the steps shown in Figure 5 The specific method of the primary training may include the following steps:

[0148] In step S50, the distance between the expected evaluation value currently output and the actual evaluation value is obtained;

[0149] In step S51, the primary weight of the weight matrix is ​​updated according to the distance. The method for updating the primary weight can be various forms known to those skilled in the art. In this example, the primary weight can be updated using the following formula (3) and formula (4):

[0150] α ′ =α+(-1) σ ·Δα, (3)

[0151]

[0152] Among them, α ′ is the first-level weight after update, α is the first-level weight before update, σ is the negative feedback operator, Δα is the update step size, d i is the distance at the current iteration number, d i-1 is the distance under the previous iteration.

[0153] In step S52, it is determined whether the current number of iterations is greater than or equal to a preset number threshold;

[0154] In step S53, when it is determined that the number of iterations is greater than or equal to the number threshold, the smallest distance is selected from the weight matrix as the optimal solution output, and it is determined that the primary training is completed;

[0155] When it is determined that the number of iterations is less than the number threshold, the process returns to the step of obtaining the distance between the expected evaluation value of the current output and the actual evaluation value.

[0156] In step S43, it is determined whether the primary optimal solution meets the preset accuracy requirement;

[0157] In step S44, if it is determined that the primary optimal solution does not meet the accuracy requirement, the weight matrix is ​​subjected to secondary training to obtain a secondary optimal solution, and the secondary optimal solution is used as the weight matrix for the final output. The specific method of the secondary training may be a form known to those skilled in the art. In one example of the present invention, the secondary training method may include: Figure 6 In the steps shown in Figure 6 In the method of secondary training, the secondary training method may include the following steps:

[0158] In step S60, the distance between the expected evaluation value currently output and the actual evaluation value is obtained;

[0159] In step S61, a plurality of weight matrices are copied to obtain a population of weight matrices;

[0160] In step S62, each weight matrix is ​​traversed and a set of unselected secondary weights is randomly selected;

[0161] In step S63, the secondary weights in the set of secondary weights are updated according to the distance to obtain a new population. The specific method for updating the secondary weights may be in various forms known to those skilled in the art. In one example of the present invention, the method for updating the secondary weights may be to use the following formulas (5) and (6) to update:

[0162] β ′ =β+(-1) σ ·Δβ, (5)

[0163]

[0164] Among them, β ′ is the updated secondary weight, β is the secondary weight before update, σ is the negative feedback operator, Δβ is the update step size, d i is the distance at the current iteration number, d i-1 is the distance under the previous iteration.

[0165] In step S64, the difference between the minimum value of the distance in the new population and the minimum value in the population before the update is determined;

[0166] In step S65, it is determined whether the difference is less than or equal to a preset difference threshold;

[0167] In step S66, when it is determined that the difference is less than or equal to the difference threshold, the weight matrix corresponding to the minimum value is selected as the optimal solution output;

[0168] When it is determined that the difference is greater than the difference threshold, the process returns to the step of traversing each weight matrix and randomly selecting a set of unselected secondary weights, that is, returns to step S62.

[0169] In step S45, when it is determined that the primary optimal solution meets the accuracy requirement, the primary optimal solution is used as the final output weight matrix.

[0170] Through the above technical solution, the embodiment of the present invention provides a digital evaluation method and system for supervision quality, which divides each part of the supervision quality report, and determines the digital evaluation index in combination with each divided part, and finally calculates the evaluation value of the supervision quality report by matrix weighting. Compared with the prior art, since the weight matrix is ​​trained based on historical evaluation values, the finally calculated evaluation value can accurately and fairly reflect the overall quality of the supervision quality report while overcoming the subjective influence of human beings.

[0171] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0172] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1A device that provides the functions specified in a block or multiple blocks.

[0173] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0174] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0175] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0176] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0177] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0178] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0179] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included within the scope of the claims of the present application.

Claims

1. A digital evaluation method for supervision quality, characterized in that: The digital evaluation method comprises: Obtain the supervision quality report currently awaiting evaluation; Divide the supervision quality report into text materials for each project unit; Using a text recognition method to recognize each of the text materials to obtain a digital evaluation index for each project; Performing threshold judgment on the digital evaluation index to obtain a digital matrix corresponding to each of the project units; Combining each of the digital matrices to obtain a digital total matrix corresponding to the supervision quality report; Constructing a weight matrix of the digitalized total matrix; The evaluation value of the supervision quality report is calculated based on the digital total matrix and the weight matrix.

2. The digital evaluation method according to claim 1, characterized in that: The supervision quality report is divided into text materials for each project unit, including: Determining the title of each of the project units based on a keyword search; Determine the actual position of the project unit according to the position of the title and the position of the next title; The supervision quality report is divided according to the actual location to obtain text materials for each of the project units.

3. The digital evaluation method according to claim 1, characterized in that: Each of the text materials is identified using a text recognition method to obtain digital evaluation indicators for each project, including: Dividing the text material into equal lengths to obtain a plurality of text regions; Traversing each of the text regions to determine whether text information can be directly captured in the text region; If it is determined that the text information can be directly captured in the text area, directly capture the text information; In the case where it is determined that the text area cannot directly capture text information, an OCR text recognition method is used to capture the text information; The digital evaluation index is determined according to the text information using a keyword matching method.

4. The digital evaluation method according to claim 3, characterized in that: Constructing a weight matrix of the digital total matrix includes: Obtaining a sample set of supervision quality reports, wherein the sample set includes multiple supervision quality reports and corresponding evaluation values; Initializing the weight matrix, wherein the weight matrix includes a plurality of sub-matrices, each of the sub-matrices corresponds one-to-one to the digitized matrix, and each element in each of the sub-matrices corresponds one-to-one to each element in the digitized matrix; Taking the digital total matrix corresponding to the supervision quality report as input and the corresponding evaluation value as output, primary training is performed on the weight matrix to obtain a primary optimal solution of the weight matrix; Determining whether the primary optimal solution meets a preset accuracy requirement; When it is determined that the primary optimal solution does not meet the accuracy requirement, secondary training is performed on the weight matrix to obtain a secondary optimal solution, and the secondary optimal solution is used as the weight matrix that is finally output.

5. The digital evaluation method according to claim 4, characterized in that: The digital evaluation method comprises: The weight matrix is ​​determined using the following formula (1) and formula (2): S={α1A1,α2A2,…,α i A i ,…,a m A m }, (1) A i ＝(β1,β2,β3,…β k ,…,β K )， (2) Where S is the weight matrix, A i is the set of secondary weights of the i-th digitized matrix, i=1,2,…,m, m is the number of digitized matrices of the digitized total matrix, α i is the first-level weight of the i-th submatrix, β k is the secondary weight of the kth digitization evaluation index, k=1, 2, …, K, and K is the number of digitization evaluation indicators in a digitization matrix.

6. The digital evaluation method according to claim 5, characterized in that: The weight matrix is ​​subjected to primary training, including: Get the distance between the expected evaluation value of the current output and the actual evaluation value; Update the primary weight of the weight matrix according to the distance; Determine whether the current number of iterations is greater than or equal to a preset number threshold; When it is determined that the number of iterations is greater than or equal to the number threshold, selecting the smallest distance from the weight matrix as the optimal solution output, and determining that the primary training is completed; When it is determined that the number of iterations is less than the number threshold, the process returns to the step of obtaining the distance between the expected evaluation value of the current output and the actual evaluation value.

7. The digital evaluation method according to claim 6, characterized in that: Updating the primary weight of the weight matrix according to the distance includes: Update each of the first-level weights according to formula (3) and formula (4), a ′ =α+(-1) σ ·Da, (3) Among them, α ′ is the first-level weight after update, α is the first-level weight before update, σ is the negative feedback operator, Δα is the update step length, d i is the distance at the current iteration number, d i-1 is the distance under the previous iteration.

8. The digital evaluation method according to claim 5, characterized in that: Performing secondary training on the weight matrix to obtain a secondary optimal solution includes: Get the distance between the expected evaluation value of the current output and the actual evaluation value; Copying a plurality of the weight matrices to obtain a population of weight matrices; Traversing each of the weight matrices, randomly selecting a set of unselected secondary weights; updating the secondary weights in the set of secondary weights according to the distance to obtain a new population; Determine the difference between the minimum distance in the new population and the minimum distance in the population before the update; Determining whether the difference is less than or equal to a preset difference threshold; When it is determined that the difference is less than or equal to the difference threshold, the weight matrix corresponding to the minimum value is selected as the optimal solution output; When it is determined that the difference is greater than the difference threshold, the step of returning to traverse each of the weight matrices and randomly selecting a set of unselected secondary weights is performed.

9. The digital evaluation method according to claim 8, characterized in that: Updating the secondary weights in the set of secondary weights according to the distance includes: b ′ =β+(-1) σ ·Db, (3) Among them, β ′ is the secondary weight after update, β is the secondary weight before update, σ is the negative feedback operator, Δβ is the update step length, d i is the distance at the current iteration number, d i-1 is the distance under the previous iteration.

10. A digital evaluation system for supervision quality, characterized in that: The digital assessment system comprises a processor, and the processor is configured to execute the digital assessment method according to any one of claims 1 to 9.

Images