A quality inspection scoring method, device and storage medium based on correlation

Abstract

The present invention discloses a quality inspection and scoring method, device, and storage medium based on relevance, comprising the following steps: determining the relevance value range and its corresponding behavioral characteristics according to actual business scenarios, and annotating the relevance of service work order corpus to obtain model training samples; extracting features from the training samples to obtain a model feature set, determining a classifier based on the model feature set, training the selected classifier, and constructing a relevance recognition model; selecting relevance weight influencing factor and time distribution influencing factor parameters according to business needs to construct a relevance scoring model; and constructing a service quality inspection scoring system based on the relevance recognition model and the relevance scoring model. The relevance weight influencing factor and the relevance time distribution factor are used to quantify the influence of the relevance value and time distribution, and a service quality inspection scoring system is constructed to score the work order service corpus text, thereby ensuring the comprehensiveness, objectivity, and accuracy of the scoring results.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a quality inspection scoring method, device and storage medium based on correlation. Background Art

[0002] In the home service scenario, due to the complexity of the service environment involving customer privacy, the supervision and evaluation of service quality is often difficult to achieve. How to identify service items, quantify service quality, and evaluate whether service orders are compliant, thereby forming an effective supervision and assessment system has become a pain point for every service provider. Since it takes huge resources to test all service orders, the current quality inspection method is mainly through customer return visits or random sampling of service recordings. However, there are still problems such as narrow sampling coverage, high labor costs, delayed and objective supervision, a single sample cannot represent the situation of other samples, and low accuracy. These are all obstacles to the standardization, informatization, and intelligentization of home service supervision. Summary of the Invention

[0003] The technical problem to be solved by the present invention is that the current supervision of service work orders through random inspections has the problems of narrow coverage, delayed supervision, lack of objectivity and low accuracy. The purpose is to provide a quality inspection scoring method, equipment and storage medium based on correlation, propose that correlation improves the accuracy of quality inspection scoring, quantify the influence of correlation value and time distribution through correlation weight influencing factor and correlation time distribution factor, construct a service quality inspection scoring system to score the work order service corpus text, and ensure the comprehensiveness and objectivity of the scoring results.

[0004] The present invention is achieved through the following technical solutions:

[0005] A first aspect of the present invention provides a quality inspection scoring method based on correlation, comprising the following steps:

[0006] S1. Determine the relevance value range and its corresponding behavioral characteristics based on the actual business scenario, and annotate the relevance of the service ticket corpus to obtain model training samples;

[0007] S2. Extract features from the training samples to obtain a model feature set, determine a classifier based on the model feature set, perform model training on the selected classifier, and construct a relevance recognition model;

[0008] S3. Select the relevance weight influencing factor and time distribution influencing factor parameters according to business needs and build a relevance scoring model;

[0009] S4. Based on the relevance identification model and relevance scoring model, a service quality inspection scoring system is constructed.

[0010] The present invention quantifies the influence of correlation value and time distribution through the correlation weight influence factor and the correlation time distribution factor, constructs a service quality inspection scoring system to score the work order service corpus text, ensures the comprehensiveness, objectivity and accuracy of the scoring results, quantifies the degree of correlation between the features of the work order service corpus text and the work order service corpus text training samples through the correlation index, and improves the interpretability and practicality of the service quality inspection scoring system.

[0011] Furthermore, the specific steps of determining the correlation value range and its corresponding behavioral characteristics include:

[0012] Formulate a relevance value strategy based on the complexity of the actual business and quality inspection requirements;

[0013] Each relevance value should correspond to a certain type of behavioral feature associated with the service item in a specific business scenario.

[0014] Furthermore, the feature extraction of the training samples to obtain the model feature set includes:

[0015] Extract features of word sequence, part-of-speech characteristics, and syntactic relations from the work order service corpus text;

[0016] The extracted features are processed by N-gram to obtain word segmentation sequence, part-of-speech tagging sequence and dependency syntax sequence;

[0017] The word segmentation sequence, part-of-speech tagging sequence and dependency syntax sequence are combined to obtain the model feature set.

[0018] Furthermore, the S3 specifically includes:

[0019] According to the relevance recognition model, the number of relevant corpora at each relevance is obtained;

[0020] The relevance, the number of relevance-related corpora under the relevance, and the relevance weight influencing factor are weighted to obtain a relevance weight scoring model.

[0021] The relevance weight scoring model is:

[0022]

[0023] Among them, α is the correlation, M α is the number of corpora with relevance α, β α is the weighted influence factor with correlation α, and the value of N is a natural number.

[0024] Furthermore, obtaining the relevance scoring model specifically includes:

[0025] Obtain a relevance time set, and obtain a relevance time distribution factor based on the time difference between adjacent sentences in the relevance time set;

[0026] Obtain the score of the relevance α statement set according to the relevance time distribution factor and the relevance scoring model;

[0027] The scores of the relevance α statement set are normalized according to the hyperbolic tangent function to obtain the relevance scoring model.

[0028] Furthermore, the correlation time distribution factor is:

[0029]

[0030] Among them, x nm =t nm -t n(m-1) is the time difference of the mth related statement in the set with relevance n, t nm Indicates the time of the mth related statement in the set with n relevance, t n(m-1) Indicates the time of the m-1th related statement in a set with relevance n.

[0031] The score of the relevance α statement set is:

[0032]

[0033] The relevance scoring model is:

[0034]

[0035] Among them, α is the correlation, M α is the number of corpora with relevance α, β α is the weighted influence factor with correlation α, σ(x αm ) is the time distribution factor with correlation α.

[0036] Furthermore, the service quality inspection scoring system specifically includes:

[0037] Acquire voice data from the work order service process, translate the voice data into text data, input the text data into the relevance recognition model, and obtain a set of relevant sentences for the project;

[0038] Input the project-related statement set into the relevance scoring model to obtain the quality inspection score results for each project;

[0039] Get the service order quality inspection results based on the quality inspection score of each service item;

[0040] Output the quality inspection score of each service item and the compliance quality inspection results of the entire service work order.

[0041] A second aspect of the present invention provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor is used to implement a quality inspection and scoring method based on correlation when executing the program.

[0042] A third aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is used to implement a quality inspection and scoring method based on correlation.

[0043] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0044] 1. By quantifying the impact of relevance values and time distribution using the relevance weight influencing factor and the relevance time distribution factor, a service quality inspection scoring system is constructed to score the work order service corpus text, ensuring the comprehensiveness, objectivity, and accuracy of the scoring results.

[0045] 2. The correlation index is used to quantify the degree of correlation between the features of the work order service corpus text and the training samples of the work order service corpus text, thereby improving the interpretability and practicality of the service quality inspection scoring system.

[0046] 3. The service quality inspection and scoring system outputs scoring results in a unified format, laying the foundation for the standardization and informatization of work order processing and service personnel assessment systems.

[0047] 4. By analyzing and NLP processing the historical audio text data of the home service process, a relevance recognition and scoring model is constructed for intelligent service process scoring, which solves the problems of standardization, informatization and intelligence of home service supervision. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the technical solutions of the exemplary embodiments of the present invention, the following briefly introduces the drawings required for use in the examples. It should be understood that the following drawings only illustrate certain embodiments of the present invention and should not be considered as limiting the scope. A person of ordinary skill in the art can also derive other relevant drawings based on these drawings without inventive effort. In the drawings:

[0049] Figure 1 is a flow chart in an embodiment of the present invention;

[0050] Figure 2 A flowchart of modeling and training of a relevance recognition model in an embodiment of the present invention;

[0051] Figure 3 is a function curve diagram of σ(x) in an embodiment of the present invention;

[0052] Figure 4 This is a complete flowchart of the service process quality inspection in an embodiment of the present invention. DETAILED DESCRIPTION

[0053] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with examples and drawings. The exemplary embodiments of the present invention and their descriptions are only used to explain the present invention and are not intended to limit the present invention.

[0054] Example 1

[0055] like Figure 1 As shown, the first aspect of this embodiment provides a quality inspection scoring method based on correlation, comprising the following steps:

[0056] S1. Determine the relevance value range and its corresponding behavioral characteristics based on the actual business scenario, and annotate the relevance of the service ticket corpus to obtain model training samples;

[0057] S2. Extract features from the training samples to obtain a model feature set, determine a classifier based on the model feature set, perform model training on the selected classifier, and construct a relevance recognition model;

[0058] S3. Select the relevance weight influencing factor and time distribution influencing factor parameters according to business needs and build a relevance scoring model;

[0059] S4. Based on the relevance identification model and relevance scoring model, a service quality inspection scoring system is constructed.

[0060] This embodiment quantifies the influence of the correlation value and time distribution through the correlation weight influence factor and the correlation time distribution factor, constructs a service quality inspection scoring system to score the work order service corpus text, ensures the comprehensiveness, objectivity and accuracy of the scoring results, quantifies the degree of correlation between the features of the work order service corpus text and the work order service corpus text training samples through the correlation index, and improves the interpretability and practicality of the service quality inspection scoring system.

[0061] In some possible embodiments, the work order service corpus text is: the service personnel in the home service scenario collect service process voice data through the audio equipment worn by the service personnel, convert the service process voice data into work order service corpus text through automatic speech recognition technology, perform natural language processing on the work order service corpus text, and obtain the correlation between the work order service corpus text and the service item, weight the correlation of all corpuses of the work order service corpus text in combination with its distribution, and finally combine the time function of the service item to obtain the quality inspection result of the service item.

[0062] In some possible embodiments, the step of determining the relevance range includes formulating a relevance value strategy based on a comprehensive consideration of the complexity of the actual business and quality inspection requirements, wherein:

[0063] Each relevance value should correspond to a certain type of behavioral feature related to the service item in a specific business scenario;

[0064] On the premise of meeting business needs, the smaller the relevance range, the better. An N value between 3 and 9 can meet most business needs.

[0065] In some possible embodiments, during the corpus annotation process, a model that meets production requirements requires a certain number of training samples. Based on model training experience, a model with N relevance levels requires at least N × 10,000 samples. The larger the relevance range, the more samples are required, and the cost of project implementation also increases. For annotation of relevant samples, it is best to select text data generated by real services, and the number of samples for each relevance level should be kept as balanced as possible. A larger range of irrelevant samples can be selected. In addition to corpora generated by real services, corpora from other similar service scenarios can also be selected.

[0066] like Figure 2 As shown, feature extraction is performed on the training samples to obtain the model feature set including:

[0067] Extract features of word sequence, part-of-speech characteristics, and syntactic relations from the work order service corpus text;

[0068] The extracted features are processed by N-gram to obtain word segmentation sequence, part-of-speech tagging sequence and dependency syntax sequence;

[0069] The word segmentation sequence, part-of-speech tagging sequence and dependency syntax sequence are combined to obtain the model feature set.

[0070] By extracting features of word sequence, part-of-speech characteristics and syntactic relations from the work order service corpus text, the feature set contains feature information of the three dimensions of word sequence, part-of-speech characteristics and syntactic relations of the training corpus, which can improve the recognition accuracy of the relevance model.

[0071] In some possible embodiments, S3 specifically includes:

[0072] According to the relevance recognition model, the number of relevant corpora at each relevance is obtained. The number of relevant corpora for a text at each relevance is: M1, M2…M N ;

[0073] The relevance, the number of relevance-related corpora under the relevance, and the relevance weight influencing factor are weighted to obtain a relevance weight scoring model.

[0074] The relevance weight scoring model is:

[0075]

[0076] Among them, α is the correlation, M α is the number of corpora with relevance α, β α is the weighted influence factor with correlation α.

[0077] β α The value range is (0, 1). The role of β is mainly reflected in two aspects. On the one hand, it adjusts the score value. The range of relevance of different types of service items and the number of each relevance corpus also vary greatly. Direct accumulation will cause the score value to fluctuate greatly, which is not conducive to the design of a general compliance judgment rule. Adjusting the value of weight β can make the score value evenly distributed in the range of (0, 1). Based on this, a standardized compliance judgment module is designed. Another role of β is to adjust the impact of each relevance corpus on the score. If there is no β α or β α The value of is the same, then the influence of each related statement on score depends on its relevance value, let β α Taking different values can more flexibly adjust the impact of related statements on the score, which can meet more complex quality inspection scoring requirements.

[0078] In some possible embodiments, obtaining the relevance scoring model specifically includes:

[0079] Obtain a relevance time set, and obtain a relevance time distribution factor based on the time difference between adjacent sentences in the relevance time set;

[0080] Obtain the score of the relevance α statement set according to the relevance time distribution factor and the relevance scoring model;

[0081] The scores of the relevance α statement set are normalized according to the hyperbolic tangent function to obtain the relevance scoring model.

[0082] The time sets of each correlation degree are:

[0083] {t 10 ,t 11 …t 1m1}, {t 10 ,t 21 …t 2m1}…{t N0 ,t N1 …t NmN}

[0084] Among them, t n0 is the service start time t0, t nmIndicates the time of the mth related statement in the set with relevance n,

[0085] The correlation time distribution factor is:

[0086]

[0087] Among them, x nm =t nm -t n(m-1) is the time difference of the mth related statement in the set with relevance n, t nm Indicates the time of the mth related statement in the set with n relevance, t n(m-1) Indicates the time of the m-1th related statement in a set with relevance n.

[0088] like Figure 3 As shown in the figure, when the time difference is 0, the value of the factor σ(x) is 0, and the influence of the current relevance statement is completely suppressed. It can be understood that it corresponds to the same service characteristic behavior as the previous relevance statement, so it is only calculated once. As the time difference increases, the influence of the relevance statement gradually increases, but it is not a simple linear growth. Instead, it grows rapidly in the initial stage, then slows down, and finally converges to 1 and stabilizes. It can be understood that after the time difference reaches a certain threshold, the relevance statement has corresponded to different service characteristic behaviors, and its influence is almost not suppressed by the previous relevance statement.

[0089] The score of the set of sentences with relevance α is:

[0090]

[0091] The relevance scoring model is:

[0092]

[0093] Among them, α is the correlation, M α is the number of corpora with relevance α, β α is the weighted influence factor with correlation α, σ(x αm ) is the time distribution factor with correlation α.

[0094] In some possible embodiments, the service quality inspection scoring system specifically includes:

[0095] Acquire voice data from the work order service process, translate the voice data into text data, input the text data into the relevance recognition model, and obtain a set of relevant sentences for the project;

[0096] Input the project-related statement set into the relevance scoring model to obtain the quality inspection score results for each project;

[0097] Output the quality inspection score of each service item and the compliance quality inspection results of the entire service work order.

[0098] A second aspect of this embodiment provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor is used to implement a quality inspection scoring method based on correlation when executing the program.

[0099] A third aspect of this embodiment provides a computer-readable storage medium having a computer program stored thereon. When the program is executed by a processor, it is used to implement a quality inspection scoring method based on correlation.

[0100] Example 2

[0101] like Figure 4 As shown, this embodiment provides the complete steps of service process quality inspection:

[0102] A1. Service personnel collect voice data during the service process through wearable audio devices and upload it to the quality inspection server. The audio is translated into a time-stamped work order service corpus text through ASR.

[0103] A2. Use NLP tools to parse and extract features from service process text;

[0104] A3. Input text features into the relevance recognition model of L service items to obtain a set of relevant sentences with timestamps for each item.

[0105] If there are L items to be inspected, the relevance recognition model for each item is independent and needs to be trained in advance;

[0106] A4. Input the relevant statements of L projects into the service quality inspection scoring system to obtain the quality inspection score of each project;

[0107] A5. Make compliance assessments for the entire service order based on the quality inspection scores of each service item.

[0108] A6. The quality inspection results are pushed to the business assessment system, and the quality inspection of this service work order is completed.

[0109] Example 3

[0110] Based on Example 1 and Example 2, this embodiment takes the home care service project M as an example of the complete steps of the service process quality inspection:

[0111] Step 1: Determine the relevance range. Quality inspection based on the work order service corpus during the service process primarily relies on the text of the work order service corpus. In actual service processes, the corpus can be divided into irrelevant and relevant corpus, and the degree of relevance between relevant corpus and the service varies. Therefore, a more accurate evaluation of service projects requires quantifiable relevance metrics and corresponding scoring algorithms. For specific quality inspection business scenarios, the relevance range [1, N] is not fixed. The most appropriate relevance strategy should be developed collaboratively by the business and algorithm teams, taking into account the complexity of the actual business and quality inspection requirements.

[0112] This scheme sets three relevance levels, 1 to 3, as shown in Table 1 (the relevance of irrelevant corpus is defined as 0):

[0113] Table 1 Correlation of corresponding behavioral characteristics

[0114] Relevance Corresponding behavioral characteristics 0 irrelevant behavior 1 Home Service Etiquette 2 Inquiry about nursing care 3 Implementation of nursing actions

[0115] Step 2: Corpus Annotation. After determining the relevance range, training samples need to be annotated. Models with three relevance levels require at least 30,000 samples. For annotating relevant samples, it is best to select text data from the work order service corpus generated by real services, and try to keep the number of samples for each relevance level balanced. The range of irrelevant samples can be larger, and in addition to the corpus generated by real services, corpus data from other similar service scenarios can also be selected.

[0116] Step 3: Feature extraction. The feature extraction of the relevance corpus is designed based on syntactic analysis, taking each sentence in the text as a processing unit. The feature engineering of the model is constructed through the three dimensions of text segmentation, part-of-speech tagging, and syntactic structure analysis, so that the feature set contains feature information of the three dimensions of word sequence, part-of-speech characteristics, and syntactic relations of the training corpus, which can improve the recognition accuracy of the relevance model. Use the NLP tool HanLP to perform syntactic analysis on all training corpora, and then perform N-gram processing on the word segmentation sequence, part-of-speech tagging sequence, and dependency syntactic sequence of the text to obtain their respective feature sets. Finally, the three feature sets are merged to obtain the feature set required by the model.

[0117] Step 4: Classification model selection. The principle of relevance identification is to obtain the relevance value through the model algorithm based on the input corpus information, so a classification model is selected to implement it. You can choose a traditional classification model based on machine learning or a classification model based on neural networks and deep learning. The accuracy and running speed of the classifier are two performance indicators that need to be considered. The classifier based on deep learning has the highest accuracy, but the running speed is relatively slow and the project implementation cost is high. Therefore, on the premise of meeting the needs of the project, machine learning and neural network classifiers are given priority. For example, SVM (support vector machine), MLP (neural network model), Fasttext, etc. are all classifiers with good comprehensive performance. In this embodiment, SVM (support vector machine) is selected as the classifier.

[0118] Step 5: Model training. After completing the relevance corpus annotation, feature engineering, and classification model selection, the relevance recognition model is obtained through model training. There are many mature open source frameworks that can be used for modeling and training. In this example, the sk-learn framework is used as a tool to train the SVM model.

[0119] Step 6: Determine the parameters of the service quality inspection scoring system. The design of the service quality inspection scoring system introduces the relevance weight influencing factor β and the time distribution influencing factor σ. This abstracts and quantifies the characteristics of the two dimensions of relevance value and time distribution, thereby providing a relatively comprehensive and accurate service process score.

[0120] The parameter values are: weight influence factor β1 = β2 = β3 = 0.033, and the final form of the relevance scoring model is:

[0121]

[0122] At this point, the model training and preparation for service project M are completed.

[0123] Step 7: The service personnel collect voice data during the service process through the audio device worn by the wearable device and upload it to the quality inspection server. The service audio is translated into time-stamped work order service corpus text data through ASR.

[0124] Step 8: Use NLP tools to parse and extract features from the process text of service project M. The text features are input into the relevance recognition model of service project M to obtain a set of relevant sentences with timestamps for each project. The set values can be set as shown in Table 2:

[0125] Table 2 Correlation of time distribution sets

[0126] Relevance Time distribution set 1 {0,1.2,4.5,9.4,20.1,28.9} 2 {0,2.5,8.3,16.5,26.4} 3 {0,3.6,9.2,18.6,24.8,27.8}

[0127] Step 9: Input the relevant statement set in the previous step into the service quality inspection scoring system, and the quality inspection score of service project M can be obtained as: 0.84 or 84 points.

[0128] Step 10: Based on the quality inspection score of each service item, the entire service ticket is judged as compliant. The compliance judgment logic is: projects with a score of 75 or higher are compliant, and tickets with at least one compliant item are compliant. Therefore, this ticket is considered compliant.

[0129] Step 11: The quality inspection results are pushed to the business assessment system: {Work order status: Compliant, included projects: {Project A: 84 points}}, and the quality inspection of this service work order is completed.

[0130] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A quality inspection scoring method based on correlation, characterized in that: The following steps are involved: S1. Determine the relevance value range and its corresponding behavioral characteristics based on the actual business scenario, and annotate the relevance of the service ticket corpus to obtain model training samples; S2. Extract features from the training samples to obtain a model feature set, determine a classifier based on the model feature set, perform model training on the selected classifier, and construct a relevance recognition model; S3. Selecting relevance weight influencing factors and time distribution influencing factor parameters based on business needs to construct a relevance scoring model; the relevance scoring model specifically includes: obtaining a relevance time set, and obtaining a relevance time distribution factor based on the time difference between adjacent sentences in the relevance time set; The scores of the relevance α statement set are obtained according to the relevance time distribution factor and the relevance scoring model; the scores of the relevance α statement set are normalized according to the hyperbolic tangent function to obtain the relevance scoring model; S4. Based on the relevance identification model and relevance scoring model, a service quality inspection scoring system is constructed.

2. The quality inspection scoring method based on correlation according to claim 1, characterized in that: The specific steps of determining the correlation value range and its corresponding behavior characteristics include: Formulate a relevance value strategy based on the complexity of the actual business and quality inspection requirements; Each relevance value should correspond to a certain type of behavioral feature associated with the service item in a specific business scenario.

3. The quality inspection scoring method based on correlation according to claim 1, characterized in that: Extracting features from the training samples to obtain a model feature set includes: Extract features of word sequence, part-of-speech characteristics, and syntactic relations from the work order service corpus text; The extracted features are processed by N-gram to obtain word segmentation sequence, part-of-speech tagging sequence and dependency syntax sequence; The word segmentation sequence, part-of-speech tagging sequence and dependency syntax sequence are combined to obtain the model feature set.

4. The quality inspection scoring method based on correlation according to claim 1, characterized in that: The S3 specifically includes: According to the relevance recognition model, the number of relevant corpora at each relevance is obtained; The relevance, the number of relevance-related corpora under the relevance, and the relevance weight influencing factor are weighted to obtain a relevance weight scoring model.

5. The quality inspection scoring method based on correlation according to claim 4, characterized in that: The relevance weight scoring model is: Among them, α is the correlation, M α is the number of corpora with correlation α, β α is the weighted influence factor with correlation α, and the value of N is a natural number.

6. The quality inspection scoring method based on correlation according to claim 5, characterized in that: The correlation time distribution factor is: Among them, x nm =t nm -t n(m-1) is the time difference of the mth related statement in the set with relevance n, t nm Indicates the time of the mth related statement in the set with n relevance, t n(m-1) Indicates the time of the m-1th related statement in a set with relevance n.

7. The quality inspection scoring method based on correlation according to claim 6, characterized in that: The score of the relevance α statement set is: The relevance scoring model is: Among them, α is the correlation, M α is the number of corpora with correlation α, β α is the weighted influence factor with correlation α, σ(x αm ) is the time distribution factor with correlation α.

8. The quality inspection scoring method based on correlation according to claim 7, characterized in that: The service quality inspection scoring system specifically includes: Acquire voice data from the work order service process, translate the voice data into text data, input the text data into the relevance recognition model, and obtain a set of relevant sentences for the project; Input the project-related statement set into the relevance scoring model to obtain the quality inspection score results for each project; Output the quality inspection score of each service item and the compliance quality inspection results of the entire service work order.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the quality inspection and scoring method based on correlation as described in any one of claims 1 to 8 is implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the quality inspection and scoring method based on correlation as described in any one of claims 1 to 8 is implemented.

Images