A data model table generation method and system based on Excel analysis

Abstract

The application relates to the technical field of data processing, and discloses a data model table generation method and system based on Excel analysis, which comprises the following steps: extracting text feature data of a to-be-entered inspection sheet, importing the text feature data into an Excel template, determining an initial form according to an import result, verifying table data based on a twin neural network model, determining whether to replace the table data to determine a target form, comparing the target form with a standard Excel template format, adjusting the target form based on the standard Excel template format when it is determined that the target form needs to be adjusted, determining an HTML form, performing distributed storage of the encrypted HTML form in a storage database, obtaining a storage hash value, and determining whether form entry is completed according to the relationship between the form hash value and the storage hash value. Through the verification of the model and the hash value, the stability of the entry is improved, and the reliability of enterprise business is ensured.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and more specifically, to a method and system for generating data model tables based on Excel parsing. Background Technology

[0002] In traditional manufacturing industries, the inspection process suffers from varying inspection standards and corresponding inspection forms for different products. For example, producing materials with different parameters generates numerous inspection forms. Many companies use spreadJS to determine the appropriate forms. However, when dealing with large batches of inspection forms, data entry often relies on the experience of maintenance personnel, making it difficult to adjust based on actual conditions. This susceptibility to human error leads to low data entry stability and increases maintenance costs. Furthermore, inconsistent data entry often results in incorrect forms being entered simultaneously, compromising the reliability of business operations.

[0003] Therefore, it is necessary to design a data model table generation method and system based on Excel parsing to solve the problems existing in the current technology. Summary of the Invention

[0004] In view of this, the present invention proposes a data model table generation method and system based on Excel parsing, which aims to solve the problems that the input of inspection forms often relies on the experience of maintenance personnel and cannot be adjusted according to the actual situation. This makes it susceptible to human influence, resulting in low input stability, disorder in the input of inspection forms, and the simultaneous input of incorrect inspection forms, which leads to insufficient reliability of enterprise business.

[0005] In one aspect, this invention proposes a method for generating data model tables based on Excel parsing, comprising:

[0006] Establish a template database and a storage database, download an Excel template from the template database, extract the text feature data of the inspection form to be entered, import the text feature data into the Excel template, and determine the initial form based on the import result;

[0007] Extract the table data of each row and each column in the initial form, verify the table data based on the Siamese neural network model, and determine whether to replace the table data to determine the target form;

[0008] The target form is compared with the standard Excel template format. Based on the comparison result, it is determined whether the target form should be adjusted. When it is determined that the target form should be adjusted, the target form is adjusted based on the standard Excel template format to determine the HTML form.

[0009] The HTML form is encrypted and its hash value is obtained. The encrypted HTML form is then distributed and stored in the storage database to obtain its storage hash value. The relationship between the form hash value and the storage hash value is used to determine whether the form entry is complete.

[0010] Furthermore, when downloading Excel templates from the template database, the process includes:

[0011] The Excel template includes a template number and template information;

[0012] The template information includes a template header and a template body, and the template header and template body are located on the same row;

[0013] The template header includes the data name, technical requirements, and header remarks;

[0014] The template table body includes field names, default values, and table body remarks.

[0015] Furthermore, when extracting the text feature data of the inspection form to be entered, importing the text feature data into the Excel template, and determining the initial form based on the import result, the process includes:

[0016] The text feature data includes keywords and noun phrases;

[0017] The system analyzes the test forms to be entered based on NLP, and performs word segmentation, stop word removal and stemming on the test forms. Based on the dependency parsing model, it analyzes the relationship between phrases and extracts noun phrases. The noun phrases that conform to the syntactic structure are identified as the keywords.

[0018] The relationships between phrases are broken down based on the keywords, and the noun phrases are determined by arranging and combining the broken-down phrases based on the NLP.

[0019] Fill the Excel template with the keywords and noun phrases to determine the initial form.

[0020] Furthermore, the table data for each row and column of the initial form is extracted, and the table data is validated based on a Siamese neural network model to determine whether to replace the table data. When determining the target form, the process includes:

[0021] Obtain the inspection forms to be entered and construct a dataset. Divide the dataset into a training set and a test set according to the sampling ratio.

[0022] The twin neural network model includes several convolutional layers, pooling layers, and fully connected layers;

[0023] The Siamese neural network model is trained using the training data set, and the trained Siamese neural network model is tested using the test data set to determine a comparison model.

[0024] Substitute the table data into the comparison model and output the model results;

[0025] When there is a discrepancy between the model results and the table data, the table data is replaced according to the model results, and the target form is determined based on the replacement result.

[0026] When the model results and the table data are consistent, the table data is determined as the target form.

[0027] Furthermore, when training the Siamese neural network model based on the data training set, and testing the trained Siamese neural network model based on the data test set to determine the comparison model, the process includes:

[0028] If the test value of the currently trained Siamese neural network model is greater than or equal to the test value of the Siamese neural network model after the previous iteration of training, then the iterative training is stopped, and the currently trained Siamese neural network model is determined as the comparison model.

[0029] If the test value of the currently trained Siamese neural network model is less than the test value of the Siamese neural network model after the previous iteration of training, the learning rate of the currently trained Siamese neural network model is adjusted, and training continues until the test value of the currently trained Siamese neural network model is greater than or equal to the test value of the Siamese neural network model after the previous iteration of training.

[0030] Furthermore, when comparing the target form with the standard Excel template format and determining whether to adjust the target form based on the comparison results, the process includes:

[0031] When the target form and the standard Excel template have the same format, it is determined that the target form will not be adjusted and the target form will be identified as an HTML form;

[0032] If the target form and the standard Excel template format are inconsistent, the target form will be adjusted.

[0033] Furthermore, when it is determined that the target form needs adjustment, the target form is adjusted based on the standard Excel template format. When determining the HTML form, this includes:

[0034] Obtain the target form that does not conform to the standard Excel template format, and determine the table format;

[0035] Obtain the standard table corresponding to the inconsistent table in the standard Excel template format, and determine the standard table format;

[0036] The table format is replaced according to the standard table format, and the HTML form is determined based on the replacement result.

[0037] Furthermore, when distributing the encrypted HTML form in the storage database, the following steps are included:

[0038] The storage database is divided into a first sub-storage database, a second sub-storage database, and a third sub-storage database;

[0039] Get the size of the encrypted HTML form data and record it as the form data volume;

[0040] When the amount of form data exceeds the first preset amount of form data, the encrypted HTML form will be stored in the first sub-storage database;

[0041] When the amount of form data is less than or equal to the first preset form data amount and greater than or equal to the second preset form data amount, the encrypted HTML form is stored in the second sub-storage database.

[0042] When the amount of form data is less than the second preset amount of form data, the encrypted HTML form is stored in the third sub-storage database.

[0043] Furthermore, when determining whether form entry is complete based on the relationship between the form hash value and the stored hash value, the process includes:

[0044] When the form hash value matches the stored hash value, the form entry is considered complete.

[0045] When the form hash value is inconsistent with the storage hash value, the form entry is determined to be abnormal, and the encrypted HTML form is re-stored in the distributed manner. When the number of times the distributed storage is re-stored equals the number of times it is re-stored is equal to the number of times the Excel template is downloaded again.

[0046] Compared with existing technologies, the beneficial effects of this invention are as follows: By using a Siamese neural network model to verify the tabular data of the initial form, it can accurately identify and correct erroneous or unreasonable data, avoiding errors caused by insufficient human experience or negligence, improving the accuracy of verification, and thus ensuring the reliability of enterprise operations. Using an Excel template to import text feature data and generate the initial form changes the previous method of relying on the experience of maintenance personnel for data entry, reducing interference from human factors and ensuring the stability of data entry. On the one hand, stable and accurate data entry reduces repetitive work and correction costs caused by erroneous data entry, lowering maintenance expenses. On the other hand, the distributed storage and encrypted HTML form not only improves the security and accuracy of stored data entry but also effectively utilizes resources, achieving efficient management of enterprise operations. The relationship between the form hash value and the storage hash value determines whether the form entry is complete, ensuring the accuracy and security of data entry.

[0047] On the other hand, this application also provides a data model table generation system based on Excel parsing, used to apply the above-mentioned data model table generation method based on Excel parsing, including:

[0048] The storage unit is configured to establish a template database and a storage database, download an Excel template from the template database, extract text feature data of the inspection form to be entered, import the text feature data into the Excel template, and determine the initial form based on the import result;

[0049] The analysis unit is configured to extract table data from each row and each column of the initial form, verify the table data based on a Siamese neural network model, and determine whether to replace the table data to determine the target form.

[0050] The processing unit is configured to compare the target form with a standard Excel template format, determine whether to adjust the target form based on the comparison result, and when it is determined that the target form should be adjusted, adjust the target form based on the standard Excel template format to determine an HTML form;

[0051] The data entry unit is configured to encrypt the HTML form and obtain the form hash value, distribute the encrypted HTML form in the storage database to obtain the storage hash value, and determine whether the form entry is complete based on the relationship between the form hash value and the storage hash value.

[0052] It is understandable that the above-mentioned method and system for generating data model tables based on Excel parsing have the same beneficial effects, and will not be elaborated further here. Attached Figure Description

[0053] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings:

[0054] Figure 1 A flowchart illustrating a method for generating a data model table based on Excel parsing, provided in an embodiment of the present invention;

[0055] Figure 2 This is a functional block diagram of a data model table generation system based on Excel parsing, provided for an embodiment of the present invention. Detailed Implementation

[0056] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the disclosure to those skilled in the art. It should be noted that, unless otherwise specified, embodiments and features in the embodiments of the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0057] In some embodiments of this application, see Figure 1 As shown, a method for generating data model tables based on Excel parsing includes:

[0058] S100: Establish a template database and a storage database, download an Excel template from the template database, extract the text feature data of the inspection forms to be entered, import the text feature data into the Excel template, and determine the initial form based on the import results.

[0059] S200: Extract table data from each row and column of the initial form, validate the table data based on the Siamese neural network model, determine whether to replace the table data, and identify the target form.

[0060] S300: Compare the target form with the standard Excel template format, and determine whether to adjust the target form based on the comparison results. If it is determined that the target form should be adjusted, the target form is adjusted based on the standard Excel template format to determine the HTML form.

[0061] S400: Encrypt the HTML form and obtain the form hash value. Distribute the encrypted HTML form in the storage database to obtain the storage hash value. Determine whether the form entry is complete based on the relationship between the form hash value and the storage hash value.

[0062] Specifically, a template database and a storage database are established, operating independently to avoid data conflicts and interference. The template database stores standard Excel templates to address inspection standards for different products. After downloading an Excel template from the template database, textual feature data of the inspection forms to be entered, such as product specifications and parameters, is extracted and imported into the Excel template to generate the initial form. This combines the information from the actual inspection forms with the Excel template, transforming traditional manual data entry into data feature extraction and import, avoiding interference from human experience, and improving data entry efficiency and accuracy. A preliminary form framework meeting inspection requirements is constructed. Table data in each row and column of the initial form is extracted and validated using a Siamese neural network model. By learning from a large number of correct table data samples, the Siamese neural network model can identify data patterns and regularities. During the validation process, the Siamese neural network model can determine whether the current table data meets expectations and whether corresponding data needs to be replaced, thus ensuring data accuracy and ultimately obtaining the target form. This ensures the data quality of the target form and improves business reliability. During the import process, differences in text and paragraph formatting between the verification form and the standard input format may occur, leading to discrepancies between the target form and the standard Excel template. By comparing the target form and the standard Excel template format and adjusting them to generate an HTML form, the formatting of the HTML form is ensured to be standardized. This facilitates subsequent data entry and use, avoiding data reading errors or business process obstacles caused by formatting inconsistencies. This allows the HTML form to better adapt to the enterprise's business systems, thereby enhancing its compatibility. Before data entry, the HTML form is encrypted, and a hash value is obtained. This hash value is a unique identifier for the HTML form. The encrypted HTML form is then distributed and stored in a database. After storage, a storage hash value is obtained. By comparing the form hash value and the storage hash value, the completion of HTML form entry is determined, ensuring the integrity of the HTML form during the data entry process. Furthermore, distributed storage and entry improve the stability and reliability of the data entry process.

[0063] Understandably, using a Siamese neural network model to verify and determine the target form reduces the subjectivity of human judgment, allowing for accurate adjustments based on the actual situation of the verification form. This improves data entry stability and reduces repetitive work and cost waste caused by data entry errors. Comparing the target form with the standard Excel template format effectively avoids the entry of erroneous verification forms, enhancing the reliability of business operations based on HTML forms. Furthermore, distributed storage and hash value verification reduce the difficulty of storing, entering, and managing HTML forms, ensuring the accuracy and standardization of data entry.

[0064] In some embodiments of this application, when downloading an Excel template from a template database, the process includes: the Excel template includes a template number and template information; the template information includes a template header and a template body; the template header and template body are located on the same row; the template header includes data name, technical requirements, and header remarks; and the template body includes field names, default values, and body remarks.

[0065] Specifically, the template number facilitates quick location and management of the corresponding downloaded Excel template, improving the efficiency of subsequent calls. The template information is divided into a template header and a template body, both on the same row. The data names, technical requirements, and header remarks in the template header clearly define the core inspection indicators and standards. The field names, default values, and body remarks in the template body further refine the inspection form, ensuring the accuracy and standardization of each subsequent initial form generation.

[0066] In some embodiments of this application, when extracting text feature data of the test form to be entered and importing the text feature data into an Excel template, and determining the initial form based on the import results, the process includes: the text feature data includes keywords and noun phrases; the test form to be entered is analyzed based on NLP, and the test form to be entered is segmented, stop words are removed, and stems are extracted; the relationship between phrases is parsed based on a dependency parsing model and noun phrases are extracted; noun phrases that conform to the syntactic structure are identified as keywords; the relationship between phrases is split based on keywords; the split phrases are arranged and combined based on NLP to determine noun phrases; and the keywords and noun phrases are filled into the Excel template to determine the initial form.

[0067] Specifically, when extracting textual feature data from the test forms to be entered, natural language processing (NLP) techniques are used to analyze the forms, including three steps: word segmentation, stop word removal, and stemming. Word segmentation breaks down the continuous text in the test forms into independent words or phrases for easier subsequent processing. Stop word removal removes common words that do not affect the actual semantics (such as "is" or "in") to improve the accuracy of the analysis. Stemming restores words to their root forms (e.g., unifying "technology" and "technical" into "technology") to reduce semantic redundancy. A dependency parsing model is used to analyze the grammatical and semantic relationships between words, extracting noun phrases from the test forms. Dependency parsing accurately identifies noun phrases (such as "technical indicators" or "industry demands"), which reflect the theme of the test forms and are crucial for keyword extraction. The relationships between phrases are broken down based on keywords, and NLP is used to arrange and combine the broken phrases to determine noun phrases, thus achieving accurate extraction of text feature data. Keyword extraction is based on grammatical structure, ensuring the semantic value of the extracted content. When filling keywords and noun phrases into the Excel template, except for the template header and body being in the same row, the other rows and columns of the Excel template are filled in sequentially according to the check sheet to be entered, ensuring the accuracy and standardization of the initial form.

[0068] In some embodiments of this application, the table data of each row and each column in the initial form is extracted, and the table data is verified based on the Siamese neural network model to determine whether to replace the table data. When determining the target form, the process includes: obtaining the verification form to be entered and constructing a dataset; dividing the dataset into a data training set and a data test set according to the sampling ratio; the Siamese neural network model includes several convolutional layers, pooling layers, and fully connected layers; training the Siamese neural network model based on the data training set; testing the trained Siamese neural network model based on the data test set; determining a comparison model; substituting the table data into the comparison model and outputting the model result; when there is an inconsistency between the model result and the table data, the table data is replaced according to the model result; the target form is determined based on the replacement result; and when the model result and the table data are consistent, the table data is determined as the target form.

[0069] Specifically, the process involves acquiring the verification forms to be entered and constructing a dataset containing all the data from these forms, which serves as the training basis to ensure the reliability and stability of the verification. The dataset is divided into a training set and a test set according to a sampling ratio, typically 7:3, to improve the model's generalization ability. The specific sampling ratio can be adjusted based on the actual dataset size. A Siamese Neural Network (SiN) is a special neural network architecture that compares the similarity of inputs by sharing weights, thereby determining the degree of similarity. The training set is used to allow the Siamese Neural Network model to learn the features and patterns of the verification forms, while the test set is used to evaluate the performance of the trained model. Feedback from the test set allows for optimization and adjustment of the Siamese Neural Network model, ensuring accuracy and generalization ability when facing data. Compared to the subjectivity of human judgment, this improves the stability of data verification. Finally, the trained Siamese neural network model was determined as the comparison model. The comparison model can output the similarity between data in a good way. The table data is substituted into the comparison model and the model results are output. When there is a discrepancy between the model results and the table data, the table data is replaced according to the model results. When the model results and the table data are consistent, the table data is determined as the target form, which ensures the accuracy and stability of the target form.

[0070] Understandably, the Siamese neural network model contains several convolutional layers, pooling layers, and fully connected layers. The trained Siamese neural network model is used as the comparison model, and its specific structure remains the same. Convolutional layers are used to extract features from the data, pooling layers are used to reduce feature dimensionality, and fully connected layers are used for the final classification. The collaborative operation of the model's multi-layered structure allows the comparison model to deeply analyze the tabular data. If the model's results are inconsistent with the tabular data, the model's results are used to replace the data, ensuring the objectivity and consistency of data verification and reducing interference from human factors. When the results are consistent, the target form is directly determined, thereby reducing labor costs and the accumulation of errors from human verification, and improving the stability of data entry.

[0071] In some embodiments of this application, when training a Siamese neural network model based on a training data set, testing the trained Siamese neural network model based on a test data set, and determining a comparison model, the process includes: if the test value of the currently trained Siamese neural network model is greater than or equal to the test value of the Siamese neural network model after the previous iteration of training, then stop the iterative training and determine the currently trained Siamese neural network model as the comparison model; if the test value of the currently trained Siamese neural network model is less than the test value of the Siamese neural network model after the previous iteration of training, then adjust the learning rate of the currently trained Siamese neural network model and continue training until the test value of the currently trained Siamese neural network model is greater than or equal to the test value of the Siamese neural network model after the previous iteration of training.

[0072] Specifically, the Siamese neural network model is iteratively trained using data from the training set. During each iteration, the model attempts to learn patterns and relationships in the data to improve its prediction or classification abilities. After each iteration, the model is tested using data from the test set. Test metrics include accuracy, loss function value, and recall, used to measure model performance. If the test value after the current iteration is lower than the previous test value, it indicates a performance decline. In this case, the learning rate needs to be adjusted, and training continues until the test value of the currently trained Siamese neural network model is greater than or equal to the test value after the previous iteration. When the test value is greater than or equal to the test value after the previous iteration, it indicates that the model's performance has improved or stabilized. At this point, iterative training can be stopped, and the model is considered to have reached a satisfactory performance level. This Siamese neural network model is then designated as the comparison model, ensuring the stability and reliability of the comparison model.

[0073] In some embodiments of this application, when comparing the target form with the standard Excel template format and determining whether to adjust the target form based on the comparison result, the following steps are taken: when the target form and the standard Excel template format are consistent, it is determined that the target form will not be adjusted and the target form is identified as an HTML form; when the target form and the standard Excel template format are inconsistent, it is determined that the target form will be adjusted.

[0074] In some embodiments of this application, when it is determined that the target form should be adjusted, the target form should be adjusted based on the standard Excel template format. When determining the HTML form, the following steps are included: obtaining tables in the target form that are inconsistent with the standard Excel template format and determining the table format; obtaining the standard table corresponding to the inconsistent table in the standard Excel template format and determining the standard table format; replacing the table format according to the standard table format; and determining the HTML form based on the replacement result.

[0075] Specifically, by comparing the target form and the inspection form to be entered, consistency between the target form and the inspection form can be ensured. However, during the filling and replacement process, there may be differences in format. By obtaining a table whose format is inconsistent with the standard Excel template, the table format can be determined and compared with the standard table format. The table format is then replaced according to the standard table format to determine the HTML form. This improves the reliability and compatibility of the HTML form, avoids the risk of data entry failure due to non-standard format, and the intelligent adjustment mechanism reduces manual intervention, lowers the risk of errors caused by manual format adjustment, and improves the resource utilization and business operation stability of enterprises when entering inspection forms.

[0076] In some embodiments of this application, when distributing the encrypted HTML form in a storage database, the process includes: dividing the storage database into a first sub-storage database, a second sub-storage database, and a third sub-storage database; obtaining the form data size of the encrypted HTML form and recording it as the form data volume; when the form data volume is greater than a first preset form data volume, storing the encrypted HTML form in the first sub-storage database; when the form data volume is less than or equal to the first preset form data volume and greater than or equal to a second preset form data volume, storing the encrypted HTML form in the second sub-storage database; and when the form data volume is less than the second preset form data volume, storing the encrypted HTML form in the third sub-storage database.

[0077] In some embodiments of this application, when determining whether form entry is complete based on the relationship between the form hash value and the storage hash value, the following steps are taken: when the form hash value and the storage hash value are consistent, it is determined that the form entry is complete; when the form hash value and the storage hash value are inconsistent, it is determined that the form entry is abnormal, and the encrypted HTML form is re-distributed and stored; when the number of times the distributed storage is re-distributed is equal to the number of times threshold is reached, the Excel template is re-downloaded.

[0078] Specifically, the storage database is divided into different sub-storage databases, and storage allocation is based on the amount of form data. This achieves refined management of data entry resources. Larger amounts of form data are stored in the first sub-storage database, while smaller amounts are stored in the third. Distributed storage can quickly respond to data entry needs, improving efficiency and ensuring rational resource utilization. This also helps reduce the risk of data loss due to excessive data in a single sub-storage database. Categorizing and storing data according to its volume allows for rapid location of encrypted HTML forms during subsequent data retrieval and management, reducing retrieval time and improving data processing efficiency.

[0079] Understandably, the HTML form is encrypted before being stored, and a form hash value is generated. The form hash value is a unique digital representation of the data content, ensuring that the data is not tampered with during transmission and entry. After the HTML form is uploaded to the distributed storage database, the storage hash value of the stored HTML form is determined again. The storage hash value is a digital representation of the actual data content after storage in the sub-storage database. The form hash value is compared with the storage hash value. If these two values ​​match, it means that the encrypted HTML form has not changed during transmission, storage, and entry, indicating that the entry has been completed correctly. If the form hash value does not match the storage hash value, it indicates that the encrypted HTML form may have been lost, damaged, or tampered with during storage and entry. In this case, the form entry is considered abnormal, and the encrypted HTML form is re-distributed for storage. The preferred number of re-distributed storage attempts is 5. When the number of re-distributed storage attempts equals the threshold (the form hash value and storage hash value are inconsistent multiple times), the Excel template needs to be downloaded again and the corresponding data needs to be re-filled to confirm the encrypted HTML form. By checking the hash value consistency, the encrypted HTML form is prevented from being tampered with during transmission, improving the overall efficiency of data management.

[0080] In summary, the beneficial effects of this invention are as follows: By using a Siamese neural network model to verify the tabular data of the initial form, it can accurately identify and correct erroneous or unreasonable data, avoiding errors caused by insufficient human experience or negligence, improving the accuracy of verification, and thus ensuring the reliability of enterprise operations. Using an Excel template to import text feature data and generate the initial form changes the previous method of relying on the experience of maintenance personnel for data entry, reducing interference from human factors and ensuring the stability of data entry. On the one hand, stable and accurate data entry reduces repetitive work and correction costs caused by erroneous data entry, lowering maintenance expenses. On the other hand, the distributed storage and encrypted HTML form not only improves the security and accuracy of stored data entry but also effectively utilizes resources, achieving efficient management of enterprise operations. The relationship between the form hash value and the storage hash value determines whether the form entry is complete, ensuring the accuracy and security of data entry.

[0081] In another preferred embodiment based on the above embodiments, see [reference] Figure 2 As shown, this embodiment provides a data model table generation system based on Excel parsing, used to apply the above-described data model table generation method based on Excel parsing, including:

[0082] The storage unit is configured to create a template database and a storage database, download an Excel template from the template database, extract text feature data of the inspection forms to be entered, import the text feature data into the Excel template, and determine the initial form based on the import results.

[0083] The analysis unit is configured to extract tabular data from each row and column of the initial form, validate the tabular data based on a Siamese neural network model, and determine whether to replace the tabular data to identify the target form.

[0084] The processing unit is configured to compare the target form with the standard Excel template format, determine whether to adjust the target form based on the comparison result, and when it is determined that the target form should be adjusted, the target form is adjusted based on the standard Excel template format to determine the HTML form.

[0085] The data entry unit is configured to encrypt the HTML form and obtain the form hash value, then distribute the encrypted HTML form in the storage database to obtain the storage hash value, and determine whether the form entry is complete based on the relationship between the form hash value and the storage hash value.

[0086] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program goods. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program goods embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0087] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program goods according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0088] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0089] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0090] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for generating data model tables based on Excel parsing, characterized in that, include: Establish a template database and a storage database, download an Excel template from the template database, extract the text feature data of the inspection form to be entered, import the text feature data into the Excel template, and determine the initial form based on the import result; Extract the table data of each row and each column in the initial form, verify the table data based on the Siamese neural network model, and determine whether to replace the table data to determine the target form; The target form is compared with the standard Excel template format. Based on the comparison result, it is determined whether the target form should be adjusted. When it is determined that the target form should be adjusted, the target form is adjusted based on the standard Excel template format to determine the HTML form. The HTML form is encrypted and its hash value is obtained. The encrypted HTML form is then distributed and stored in the storage database to obtain a storage hash value. The relationship between the form hash value and the storage hash value is used to determine whether the form entry is complete. When downloading an Excel template from the template database, the following is included: The Excel template includes a template number and template information; The template information includes a template header and a template body, and the template header and template body are located on the same row; The template header includes the data name, technical requirements, and header remarks; The template table body includes field names, default values, and table body remarks; Extract table data from each row and column of the initial form, validate the table data based on a Siamese neural network model, and determine whether to replace the table data. When determining the target form, the process includes: Obtain the inspection forms to be entered and construct a dataset. Divide the dataset into a training set and a test set according to the sampling ratio. The twin neural network model includes several convolutional layers, pooling layers, and fully connected layers; The Siamese neural network model is trained using the training data set, and the trained Siamese neural network model is tested using the test data set to determine a comparison model. Substitute the table data into the comparison model and output the model results; When there is a discrepancy between the model results and the table data, the table data is replaced according to the model results, and the target form is determined based on the replacement result. When the model results and the table data are consistent, the table data is determined as the target form; When it is determined that the target form needs adjustment, the target form is adjusted based on the standard Excel template format. When determining the HTML form, this includes: Obtain the target form that does not conform to the standard Excel template format, and determine the table format; Obtain the standard table corresponding to the inconsistent table in the standard Excel template format, and determine the standard table format; The table format is replaced according to the standard table format, and the HTML form is determined based on the replacement result; When determining whether form entry is complete based on the relationship between the form hash value and the stored hash value, the process includes: When the form hash value matches the stored hash value, the form entry is considered complete. When the form hash value is inconsistent with the storage hash value, the form entry is determined to be abnormal, and the encrypted HTML form is re-stored in the distributed manner. When the number of times the distributed storage is re-stored equals the number of times it is re-stored is equal to the number of times the Excel template is downloaded again.

2. The method for generating a data model table based on Excel parsing according to claim 1, characterized in that, When extracting the text feature data of the inspection form to be entered, importing the text feature data into the Excel template, and determining the initial form based on the import results, the process includes: The text feature data includes keywords and noun phrases; The system analyzes the test forms to be entered based on NLP, and performs word segmentation, stop word removal and stemming on the test forms. Based on the dependency parsing model, it analyzes the relationship between phrases and extracts noun phrases. The noun phrases that conform to the syntactic structure are identified as the keywords. The relationships between phrases are broken down based on the keywords, and the noun phrases are determined by arranging and combining the broken-down phrases based on the NLP. Fill the Excel template with the keywords and noun phrases to determine the initial form.

3. The method for generating a data model table based on Excel parsing according to claim 2, characterized in that, When training the Siamese neural network model based on the training data set, testing the trained Siamese neural network model based on the test data set, and determining the comparison model, the process includes: If the test value of the currently trained Siamese neural network model is greater than or equal to the test value of the Siamese neural network model after the previous iteration of training, then the iterative training is stopped, and the currently trained Siamese neural network model is determined as the comparison model. If the test value of the currently trained Siamese neural network model is less than the test value of the Siamese neural network model after the previous iteration of training, the learning rate of the currently trained Siamese neural network model is adjusted, and training continues until the test value of the currently trained Siamese neural network model is greater than or equal to the test value of the Siamese neural network model after the previous iteration of training.

4. The method for generating a data model table based on Excel parsing according to claim 3, characterized in that, When comparing the target form with a standard Excel template format and determining whether to adjust the target form based on the comparison results, the process includes: When the target form and the standard Excel template have the same format, it is determined that the target form will not be adjusted and the target form will be identified as an HTML form; If the target form and the standard Excel template format are inconsistent, the target form will be adjusted.

5. The method for generating a data model table based on Excel parsing according to claim 4, characterized in that, When distributing the encrypted HTML form in the storage database, the following steps are included: The storage database is divided into a first sub-storage database, a second sub-storage database, and a third sub-storage database; Get the size of the encrypted HTML form data and record it as the form data volume; When the amount of form data exceeds the first preset amount of form data, the encrypted HTML form will be stored in the first sub-storage database; When the amount of form data is less than or equal to the first preset form data amount and greater than or equal to the second preset form data amount, the encrypted HTML form is stored in the second sub-storage database. When the amount of form data is less than the second preset amount of form data, the encrypted HTML form is stored in the third sub-storage database.

6. A data model table generation system based on Excel parsing, used to apply the data model table generation method based on Excel parsing as described in any one of claims 1-5, characterized in that, include: The storage unit is configured to establish a template database and a storage database, download an Excel template from the template database, extract text feature data of the inspection form to be entered, import the text feature data into the Excel template, and determine the initial form based on the import result; The analysis unit is configured to extract table data from each row and each column of the initial form, verify the table data based on a Siamese neural network model, and determine whether to replace the table data to determine the target form. The processing unit is configured to compare the target form with a standard Excel template format, determine whether to adjust the target form based on the comparison result, and when it is determined that the target form should be adjusted, adjust the target form based on the standard Excel template format to determine an HTML form; The data entry unit is configured to encrypt the HTML form and obtain the form hash value, distribute the encrypted HTML form in the storage database to obtain the storage hash value, and determine whether the form entry is complete based on the relationship between the form hash value and the storage hash value.

Images