An RPA automated case filing processing method and system

Abstract

The present invention discloses an RPA automated case filing processing method and system: obtaining case filing material data, performing text recognition on scanned files and image files, generating a second electronic text file, obtaining the case cause of the case to be filed from the electronic text file, using the case information template corresponding to the template case cause with the greatest similarity to the case cause as the case information template of the case to be filed, performing semantic analysis on the electronic text file, filling in the case information template to generate a case information table, performing integrity detection on the case information table, and if the detection passes, executing the case filing application according to the case information table, assigning data identifiers to the case filing material data, and dividing the case filing material data into N data blocks, then dividing the N data blocks into first data and second data, storing the first data in a first server, and storing the second data in a second server. The technical solution of the present invention improves case filing efficiency and data security.

Description

Technical Field

[0001] The present invention belongs to the technical field of big data analysis, and specifically relates to an RPA automated case filing processing method and system. Background Art

[0002] When filing a case, staff usually enter various information and materials manually, which is a time-consuming and labor-intensive process. When the case volume is large and a large amount of materials need to be submitted, staff are unable to complete the acceptance work in a timely manner and cannot guarantee the accuracy of the information entered, which will lead to an increased error rate and relatively low efficiency and accuracy.

[0003] With the development of computers, many traditional offices have been gradually replaced by convenient online offices, and automatic case filing has become more and more of a trend. In the prior art, for example, Chinese patent application CN108460712A discloses a method for controlling an auxiliary case filing terminal robot, which can self-service receive, scan and submit case filing materials; self-service packing and archiving of paper case filing materials; centralized review and management of the electronic document database of case filing materials, and the backstage operator can review the materials after logging into the system on the backstage computer, and check the image files submitted for each document in order; and save and manage the original case filing materials. This method automatically scans and submits case filing materials and directly stores them, and then the staff reviews the materials in the background, which is easy to cause case information leakage. Another example is Chinese patent application CN113793121A, which discloses an automatic filing method, device, electronic device and storage medium for litigation cases. The method obtains litigation documents sent by litigation users, including image materials and text materials; calculates the clarity value of the image materials, and determines whether the text materials meet the preset filing requirements; if the clarity value is greater than or equal to the clarity threshold, and the text materials meet the preset filing requirements, the litigation documents are uploaded to the preset filing system, and the filing number of the litigation documents is automatically generated. After judging that the litigation documents meet the requirements, this method directly uploads the litigation documents to the preset filing system, which is easy to cause case information leakage.

[0004] Therefore, it is an urgent problem to provide an RPA automated case filing method and system to improve case filing efficiency while improving data security. Summary of the invention

[0005] In response to the above-mentioned technical problems, the present invention provides an RPA automated case filing processing method and system.

[0006] In a first aspect, the present invention provides an RPA automated case filing processing method, which is applied to an RPA terminal. The method comprises the following steps:

[0007] Step 1: The RPA terminal obtains the case filing material data of the case to be filed, wherein the case filing material data includes a first electronic text file, a scanned file, and a picture file;

[0008] Step 2: Perform text recognition on the scanned files and the image files, extract text information from each scanned file and each image file, and define the extracted text information as a second electronic text file;

[0009] Step 3: Obtain the case information of the case to be filed from the electronic text file, calculate the first similarity between the case information and any template case information in the case information template library, and use the case information template corresponding to the template case information with the largest first similarity as the case information template of the case to be filed, wherein the electronic text file includes the first electronic text file and the second electronic text file;

[0010] Step 4: Perform semantic analysis on the electronic text file based on the case information template, fill in the corresponding fields of the case information template to generate a case information table;

[0011] Step 5: Perform integrity check on the case information table. If the check fails, a prompt message indicating that the check failed is sent to the user who submitted the case to be filed. If the check passes, proceed to step 6.

[0012] Step 6: Follow the case filing application process according to the case information form;

[0013] Step 7: Assign a data identifier to the case filing material data, divide the case filing material data into N data blocks according to preset rules, then store the data identifier and the preset rules in the first storage module of the RPA terminal accordingly, and divide the N data blocks into first data and second data. After the division is completed, send and store the data identifier and the first data in the first server, and send and store the data identifier and the second data in the second server, wherein the first data includes N1 data blocks, the second data includes N2 data blocks, and N=N1+N2.

[0014] Specifically, the scanned file and the image file are defined as two types of files. In step 2, the text recognition of the scanned file and the image file includes:

[0015] Step 21, obtaining any page file in the second category of files, identifying a character region image of any page file, and preprocessing the character region image to obtain first image data;

[0016] Step 22: obtaining any character image in the first image data, and performing text recognition on any character image;

[0017] Step 23, defining the characters in the recognition result whose similarity with any character image is greater than a first preset value as pending characters;

[0018] Step 24, obtaining the text image of the mth undetermined text, drawing the bounding rectangles of any character image and the text image respectively, and adjusting the size of any character image based on the bounding rectangle of the text image so that the size of the bounding rectangle of any character image is equal to that of the bounding rectangle of the text image, wherein m is a positive integer from 1 to M, and M is the total number of undetermined texts;

[0019] Step 25, obtaining the i-th pixel in any character image, when the pixel value of the i-th pixel is (0, 0, 0) and the pixel value of the i-th pixel in the text image is (0, 0, 0), setting the pixel value of the i-th pixel to (0, 0, 0), otherwise, setting the pixel value of the i-th pixel to (255, 255, 255), after traversing all the pixels in any character image, generating a second text image corresponding to the text image, wherein i is a positive integer from 1 to I, and I is the total number of pixels in any character image;

[0020] Step 26: After traversing all the pending characters, calculate the second similarity between any character image and the second character image corresponding to any pending character, and take the pending character corresponding to the second character image with the largest second similarity as the character recognition result of any character image.

[0021] Specifically, in step 21, the preprocessing includes:

[0022] Step 211, performing image smoothing processing on the character region image to obtain a second character region image;

[0023] Step 212, obtaining the grayscale value of the j-th pixel in the second character region image, when the grayscale value of the j-th pixel is less than the second preset value, setting the grayscale value of the j-th pixel to 0, when the grayscale value of the j-th pixel is greater than or equal to the second preset value, setting the grayscale value of the j-th pixel to 255, and obtaining the third character region image after the processing is completed, wherein j is a positive integer from 1 to J, and J is the total number of pixels of the second character region image;

[0024] Step 213, respectively obtain the number of pixels of each pixel block in the third character area image; if the number of pixels of any pixel block is less than a third preset value, set the grayscale value of all pixels in any pixel block to 255, wherein the pixel block is a collection of pixels with continuous coordinates.

[0025] Specifically, in step 22, before obtaining any character image in the first image data, the process includes:

[0026] Step 221, sequentially intercepting a character sequence diagram in the first image data, wherein the character sequence diagram is a block diagram of a string of characters written continuously;

[0027] Step 222: extract any character sequence diagram, and draw the circumscribed rectangle of each character in any character sequence diagram;

[0028] Step 223: Obtain the first height value of any character sequence diagram, and obtain the second height value and center point coordinates of the circumscribed rectangle of each character, calculate the character height standard deviation based on all the second height values, calculate the character center point standard deviation based on the Y-axis coordinates of all the center point coordinates, and define the ratio of the average value of all the second height values ​​to the first height value as the character height ratio;

[0029] Step 224: when the character height standard deviation is less than or equal to the fourth preset value, the character center point standard deviation is less than or equal to the fifth preset value, and the character height ratio is less than or equal to the sixth preset value, it is determined that the characters intercepted in any character sequence diagram are in one row; otherwise, it is determined that the characters intercepted in any character sequence diagram are in multiple rows;

[0030] Step 225: if the characters contained in any character sequence diagram are multiple lines, divide any character sequence diagram into multiple character sequence block images along the direction of the character sequence, sequentially intercept the character sequence diagram in each character sequence block image, then return to step 222, and after traversing all the character sequence block images, enter step 226; if the characters contained in any character sequence diagram are one line, enter step 226;

[0031] Step 226, determine whether all character sequence diagrams in the first image data have been traversed. If not, return to step 222. If they have been traversed, align each line of character sequence in the first image data to generate second image data, and then obtain any character image in the second image data for text recognition.

[0032] Specifically, in step 225, dividing any character sequence diagram into a plurality of character sequence block images along the direction of the character sequence includes:

[0033] Step 2251, obtaining the number of character columns in any character sequence diagram, and marking each column of characters in order from left to right;

[0034] Step 2252: Set the position between the g-1th column character and the gth column character as the position to be divided, where g is a positive integer from 2 to G, and G is the total number of character columns;

[0035] Step 2253, respectively calculate the sum of pixel values ​​of each row of pixels from the 1st column of characters to the g-1th column of characters, obtain the minimum value of the sum of pixel values, and determine whether the minimum value of the sum of pixel values ​​is less than or equal to the seventh preset value. If so, proceed to step 2254; if so, proceed to step 2255.

[0036] Step 2254, define the position to be divided as the division position, then determine whether g is equal to G, if so, proceed to step 2256, if not, set the mark of the character in the g-th column to 1, adjust the marks of each column of characters after the g-th column in turn, and then return to step 2252;

[0037] Step 2255, g=g-1, determine whether g is equal to 2, if so, define the position between the first column character image and the second column character image as the dividing position, and set the mark of the second column character to 1, adjust the mark of each column character after the second column character in turn, and then return to step 2252, if not, return to step 2253;

[0038] Step 2256: Divide any character sequence image into a plurality of character sequence block images based on the division position.

[0039] Specifically, in step 3, calculating the first similarity between the case and any template case in the case template library includes:

[0040] Step 31, obtaining a first sentence vector of the case and a second sentence vector of any template case, and calculating a third similarity between the case and any template case based on the first sentence vector and the second sentence vector;

[0041] Step 32: Generate a case image based on the case, generate a template case image based on any template case, and calculate a fourth similarity between the case image and the template case image;

[0042] Step 33: Calculate the first similarity based on the third similarity and the fourth similarity, and the calculation formula is:

[0043]

[0044] Among them, SI1 is the first similarity, SI3 is the third similarity, SI4 is the fourth similarity, and α3 and α4 are parameter weight coefficients.

[0045] Specifically, after step 7, the following steps are also included:

[0046] Step 71: When the staff member wants to obtain the case filing material data, the staff member sends a data acquisition request to the server through the terminal device, wherein the server includes a first server and a second server, and the data acquisition request includes a data identifier;

[0047] Step 72: After receiving the data acquisition request, the server generates a test frame, and then sends the test frame to the terminal device. Based on the response of the terminal device, the response time of receiving the response is calculated, and the number of route hops for sending the test frame to the terminal device is obtained.

[0048] Step 73: When the response time is less than the eighth preset value and the number of routing hops is less than the ninth preset value, extract the data corresponding to the data identifier and send it to the terminal device.

[0049] In a second aspect, the present invention further provides an RPA automated case filing processing system, the system comprising: an RPA terminal, a first server and a second server, the RPA terminal comprising a data acquisition module, a text recognition module, a template selection module, an information table generation module, a case filing application module and a data storage module;

[0050] A data acquisition module, used to acquire case filing material data of a case to be filed, wherein the case filing material data includes a first electronic text file, a scanned file, and a picture file;

[0051] A text recognition module, used to perform text recognition on scanned files and image files, extract text information from each scanned file and each image file, and define the extracted text information as a second electronic text file;

[0052] A template selection module, used to obtain the case information of the case to be filed from the electronic text file, calculate the similarity between the case information and any template case information in the case information template library, and use the case information template corresponding to the template case information with the greatest similarity as the case information template of the case to be filed, wherein the electronic text file includes a first electronic text file and a second electronic text file;

[0053] An information table generation module is used to perform semantic analysis on the electronic text file based on the case information template, and fill in the corresponding fields of the case information template to generate a case information table;

[0054] The case filing application module is used to perform integrity check on the case information table. If the check fails, a prompt message indicating that the check failed is sent to the user who submitted the case to be filed. If the check passes, the case filing application process is executed according to the case information table.

[0055] The data storage module is used to assign data identifiers to case filing material data, divide the case filing material data into N data blocks according to preset rules, and then store the data identifiers and the preset rules in the first storage module of the RPA terminal accordingly, and divide the N data blocks into first data and second data. After the division is completed, the data identifier and the first data are sent and stored in the first server, and the data identifier and the second data are sent and stored in the second server, wherein the first data includes N1 data blocks, the second data includes N2 data blocks, and N=N1+N2.

[0056] The present invention discloses an RPA automated case filing processing method and system, which obtains case filing material data of a case to be filed, the case filing material data including a first electronic text file, a scan file and a picture file, performs text recognition on the scan file and the picture file, defines the extracted text information as a second electronic text file, performs semantic analysis on the first electronic text file and the second electronic text file based on a case filing information template, fills in corresponding fields of the case information template to generate a case information table, thereby improving the accuracy and comprehensiveness of filling in the case information table, then performs integrity detection on the completed case information table to determine whether all blank fields have been filled in, and if they are filled in completely, executes the case filing application process, thereby improving the case filing efficiency, and at the same time, divides the case filing material data into N data blocks according to preset rules, divides the N data blocks into first data and second data, and stores the first data and the second data in different servers, thereby improving data security and avoiding leakage of case filing material data. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0058] Figure 1 This is a flow chart of an RPA automated case filing method of the present invention;

[0059] Figure 2a This is a first schematic diagram of text recognition according to an embodiment of the present invention;

[0060] Figure 2b A second schematic diagram of text recognition according to an embodiment of the present invention;

[0061] Figure 3 A schematic diagram of a character sequence diagram according to an embodiment of the present invention;

[0062] Figure 4 It is a structural schematic diagram of an RPA automated case filing processing system of the present invention. DETAILED DESCRIPTION

[0063] In order to make the purpose, technical scheme and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. Obviously, the specific embodiments described herein are only used to explain the present invention and are part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work belong to the scope of protection of the present invention.

[0064] It should be noted that if there are descriptions involving "first", "second", etc. in the embodiments of the present invention, the descriptions of "first", "second", etc. are only used for descriptive purposes and cannot be understood as indicating or implying their relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In addition, the technical solutions between the various embodiments can be combined with each other, but they must be based on the ability of ordinary technicians in the field to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be deemed that such a combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0065] RPA stands for Robotic Process Automation, which uses specific "robot software" to simulate human operations on computers. Its main function is to use robots to execute work information and business interactions according to pre-designed processes. RPA can efficiently solve complex processes and save labor costs.

[0066] Figure 1 FIG. 1 is a flowchart of an embodiment of an RPA automated case filing method provided by the present invention, and the flowchart specifically includes the following steps:

[0067] Step 1: The RPA terminal obtains the case filing material data of the case to be filed, where the case filing material data includes a first electronic text file, a scanned file, and a picture file.

[0068] In the embodiment of the present invention, the case filing material data refers to the materials required by the user to apply for case filing, including the complaint, identity document materials, power of attorney, evidence, etc.

[0069] The source of the case filing materials data can be the case filing materials submitted by the party applying for case filing through the case filing system, or the paper version of the case filing materials submitted by the party applying for case filing, and then the staff obtains the electronic version of the case filing materials data through scanning equipment.

[0070] Step 2: Perform text recognition on the scanned files and the image files, extract text information from each scanned file and each image file respectively, and define the extracted text information as a second electronic text file.

[0071] Specifically, the scanned file and the image file are defined as two types of files. In step 2, the text recognition of the scanned file and the image file includes:

[0072] Step 21 : obtaining any page file in the second category of files, identifying the character area image of any page file, and preprocessing the character area image to obtain first image data.

[0073] Step 22: Obtain any character image in the first image data, and perform character recognition on any character image.

[0074] Step 23: Define the characters in the recognition result whose similarity to any character image is greater than the first preset value as the pending characters.

[0075] Step 24: Obtain the character image of the m-th pending character, respectively draw the circumscribed rectangles of any character image and the character image, and adjust the size of any character image based on the circumscribed rectangle of the character image so that the circumscribed rectangle of any character image is equal to the circumscribed rectangle of the character image, where m is a positive integer from 1 to M, and M is the total number of pending characters.

[0076] Step 25: Obtain the i-th pixel point in any character image. When the pixel value of the i-th pixel point is (0, 0, 0) and the pixel value of the i-th pixel point in the character image is (0, 0, 0), set the pixel value of the i-th pixel point to (0, 0, 0); otherwise, set the pixel value of the i-th pixel point to (255, 255, 255). After traversing all the pixel points in any character image, generate the second character image corresponding to the character image, where i is a positive integer from 1 to I, and I is the total number of pixel points of any character image.

[0077] Step 26: After traversing all the pending characters, calculate the second similarity between any character image and the second character image corresponding to any pending character, and use the pending character corresponding to the second character image with the largest second similarity as the character recognition result of any character image.

[0078] The first preset value can be set according to the experience of those skilled in the art or according to the actual application scenario, and the embodiments of the present application do not limit this.

[0079] Exemplarily, when recognizing the image of the character "天", due to stains or recognition errors in the character image, multiple characters may be recognized. Calculate the similarity between the character images of multiple recognition results and the image of the character "天", and define the characters whose similarity is greater than the first preset value as the pending characters, such as "天", "太", "夫". Obtain the character images of the three characters "天", "太", "夫", and adjust the circumscribed rectangle of the character image of the character "天" to be the same as the circumscribed rectangles of the three character images of "天", "太", "夫", that is, the ratios are equal. For example Figure 2aas shown (noise exists on the "Tian" character image, not shown). Since the sizes of the four images are equal and the coordinates of each pixel point are equal, obtain the pixel value of the i-th pixel point in the "Tian" character image. When the pixel value of the i-th pixel point is (0, 0, 0) and the pixel value of the i-th pixel point in the character image "Fu" is (0, 0, 0), set the pixel value of the i-th pixel point in the character image "Fu" to (0, 0, 0). In other cases, set the pixel value of the i-th pixel point to (255, 255, 255). That is, only when the i-th pixel point in the "Tian" character image and the i-th pixel point in the character image "Fu" are both black, set the i-th pixel point in the character image "Fu" to black, and set it to white in other cases, generating the second character image corresponding to the character image "Fu". Similarly, generate the second character images corresponding to the character images "Tian" and "Tai" respectively. As Figure 2b shown, the second character images corresponding to the character images "Tian", "Tai", and "Fu" are "Tian", "Ren", and "Da" respectively. Calculate the similarity between each second character image and the "Tian" character image, and use the pending character corresponding to the second character image with the largest similarity as the text recognition result of the "Tian" character image.

[0080] Preferably, when generating the second character image based on the pixel values of each pixel point of the character image and the pixel values of each pixel point of the character image, the pixel values of each pixel point of the character image can be changed, the pixel values of each pixel point of the character image can also be changed, or a new blank image can be generated, and the pixel values of each pixel point of the new image are set based on the pixel values of each pixel point of the character image and the pixel values of each pixel point of the character image.

[0081] According to the technical solution of the present invention, based on the character image to be recognized and the character image of the pending character, a second character image is generated, and then the character image to be recognized is text-recognized based on the second character image, which can effectively remove the noise in the character image and improve the text recognition accuracy.

[0082] Specifically, in step 21, the preprocessing includes:

[0083] Step 211: Perform image smoothing processing on the character region image to obtain a second character region image.

[0084] Step 212: Obtain the gray value of the j-th pixel point in the second character region image. When the gray value of the j-th pixel point is less than the second preset value, set the gray value of the j-th pixel point to 0. When the gray value of the j-th pixel point is greater than or equal to the second preset value, set the gray value of the j-th pixel point to 255. After processing, a third character region image is obtained, where j is a positive integer from 1 to J, and J is the total number of pixel points in the second character region image.

[0085] Step 213, respectively obtain the number of pixels of each pixel block in the third character area image; if the number of pixels of any pixel block is less than a third preset value, set the grayscale value of all pixels in any pixel block to 255, wherein the pixel block is a collection of pixels with continuous coordinates.

[0086] The second preset value and the third preset value may be set according to the experience of those skilled in the art or according to actual application scenarios, and the embodiments of the present application do not limit this.

[0087] The character area image is preprocessed to remove some noise from each text image.

[0088] Specifically, in step 22, before obtaining any character image in the first image data, the process includes:

[0089] Step 221 : sequentially intercepting a character sequence diagram in the first image data, wherein the character sequence diagram is a block diagram of a string of characters written continuously.

[0090] Step 222: extract any character sequence diagram, and draw the circumscribed rectangle of each character in any character sequence diagram.

[0091] Step 223, obtain the first height value of any character sequence diagram, and obtain the second height value and center point coordinates of the circumscribed rectangle of each character, calculate the character height standard deviation based on all the second height values, calculate the character center point standard deviation based on the Y-axis coordinates of all the center point coordinates, and define the ratio of the average value of all the second height values ​​to the first height value as the character height ratio.

[0092] Step 224: when the character height standard deviation is less than or equal to the fourth preset value, the character center point standard deviation is less than or equal to the fifth preset value, and the character height ratio is less than or equal to the sixth preset value, it is determined that the characters intercepted in any character sequence diagram are in one row; otherwise, it is determined that the characters intercepted in any character sequence diagram are in multiple rows.

[0093] Step 225: If any character sequence diagram contains multiple lines of characters, divide any character sequence diagram into multiple character sequence block images along the direction of the character sequence, sequentially intercept the character sequence diagram in each character sequence block image, and then return to step 222. After traversing all character sequence block images, enter step 226; if any character sequence diagram contains a line of characters, enter step 226.

[0094] Step 226, determine whether all character sequence diagrams in the first image data have been traversed. If not, return to step 222. If they have been traversed, align each line of character sequence in the first image data to generate second image data, and then obtain any character image in the second image data for text recognition.

[0095] The fourth preset value, the fifth preset value, and the sixth preset value can be set according to the experience of those skilled in the art or according to the actual application scenario, and the embodiments of the present application do not limit this.

[0096] In the case where the document is handwritten or the printed paper is not placed properly, it will be inclined, resulting in the recognition of the upper and lower lines as the same string of characters. For example Figure 3 As shown, in the case of handwritten, the line spacing becomes smaller from left to right. When the line spacing becomes smaller, the upper and lower lines will be recognized as the same string of characters and intercepted into a character sequence diagram. Since two lines of characters are recognized as one line, when performing character recognition, "结, 为" will be recognized as one character, and "合, 了"... "明, 案" will be recognized as one character in turn. When drawing the circumscribed rectangle of each character in the character sequence diagram, "结, 为", "合, 了", etc. are drawn as one character for the circumscribed rectangle respectively. Since the line spacing becomes smaller from left to right, the height of the character circumscribed rectangles from left to right decreases in turn. As Figure 3 shown by the character circumscribed rectangles corresponding to the identifications 03 and 04 in, the line spacing between "行" and "例" is smaller than the line spacing between "结" and "为", resulting in the height value of the circumscribed rectangle of "行, 例" being smaller than that of the circumscribed rectangle of "结, 为". Based on the drawn circumscribed rectangles, calculate the standard deviation of character height, the standard deviation of character center points, and the character height ratio. If any one of the standard deviation of character height, the standard deviation of character center points, and the character height ratio does not meet its corresponding preset value, it is determined that the character sequence diagram contains multiple lines of characters.

[0097] When the character sequence diagram contains multiple lines of characters, a dividing line is generated along the direction perpendicular to the character arrangement, and the character sequence diagram is divided into multiple character sequence block images, and then each character sequence block image is re-divided and recognized for the character sequence diagram.

[0098] If there is a situation where the character sequence diagram contains multiple lines of characters, then determine the inclination direction of the character arrangement in the multi-line character sequence diagram, and perform alignment processing on each line of characters based on this inclination direction. Exemplarily, if it is inclined upward, it is adjusted downward for alignment, and if it is inclined downward, it is adjusted upward for alignment.

[0099] Specifically, in step 225, dividing any character sequence diagram into multiple character sequence block images along the direction of the character sequence includes:

[0100] Step 2251: Obtain the number of character columns in any character sequence diagram, and sequentially mark each column of characters in the order from left to right.

[0101] Step 2252: Set the position between the (g - 1)-th column of characters and the g-th column of characters as the position to be divided, where g is a positive integer from 2 to G, and G is the total number of character columns.

[0102] Step 2253: Calculate the sum of pixel values of each row of pixels from the characters in the first column to the characters in the (g - 1)-th column respectively, obtain the minimum value of the sum of pixel values, and determine whether the minimum value of the sum of pixel values is less than or equal to the seventh preset value. If it is less than or equal to, proceed to Step 2254; if it is greater, proceed to Step 2255.

[0103] Step 2254: Define the position to be divided as the division position. Subsequently, determine whether g is equal to G. If it is equal, proceed to Step 2256; if it is not equal, set the mark of the characters in the g-th column to 1, sequentially adjust the marks of each column of characters after the g-th column, and then return to Step 2252.

[0104] Step 2255: g = g - 1. Determine whether g is equal to 2. If it is, define the position between the character image in the first column and the character image in the second column as the division position, and set the mark of the characters in the second column to 1. Sequentially adjust the marks of each column of characters after the second column, and then return to Step 2252. If it is not, return to Step 2253.

[0105] Step 2256: Divide any character sequence diagram into multiple character sequence block images based on the division position.

[0106] The seventh preset value can be set according to the experience of those skilled in the art or according to the actual application scenario, and the embodiments of the present application do not limit this.

[0107] Preferably, the seventh preset value is 0.

[0108] Take Figure 3 as an example to illustrate the technical solution of the present invention. Figure 3 For the case where there are multiple rows of characters in a character sequence diagram, the number of character columns is 19 columns. Among them, "jie, wei" is one column, "he, le" is one column,... "ming, an" is one column. Starting from right to left for division, first take the position between the 18th column "shuo, fang" and the 19th column "ming, an" (shown as position 1 in the figure) as the position to be divided. Subsequently, calculate the sum of pixel values of each row of pixels from the characters in the first column to the characters in the 18th column respectively. The calculation formula is: Among them, TP h is the sum of pixel values of the h-th row, and p h,qis the pixel value of the q-th pixel in the h-th row, and Q is the total number of pixels in the h-th row. At this time, the sum of pixel values is greater than 0 (the seventh preset value), indicating that there are character strokes in this row. Then, move the position to be divided to the left, to between the 17th column character and the 18th column character, and repeat the operation until the minimum value of the sum of pixel values is equal to 0. Exemplarily, when the position to be divided is moved to between the 11th column characters "发、明" and the 12th column characters "明、实" (shown as position 2 in the figure), the minimum value of the sum of pixel values of each row of pixels from the 1st column character to the 11th column character is 0, indicating that there are no character strokes in this row and this row is white. Then, define this position to be divided as the division position, and based on this division position, the character sequence diagram can be divided into two character sequence block images on the left and right. Subsequently, the upper and lower two character sequence diagrams in the first image data can be intercepted from the left part of the character sequence block image. Then, define the 12th column as the 1st column, and sequentially define the 13th column as the 2nd column,..., and define the 19th column as the 8th column, and then re-divide the 1st column to the 8th column.

[0109] Based on the technical solution of the present invention, it is possible to prevent extracting multiple lines during text recognition due to writing problems, etc., so that characters in different lines are recognized as the same character, and the recognition accuracy is improved.

[0110] Step 3: Obtain the cause of action of the case to be filed from the electronic text file, calculate the first similarity between the cause of action and any template cause of action in the cause of action template library, and use the case information template corresponding to the template cause of action with the largest first similarity as the case information template of the case to be filed, where the electronic text file includes a first electronic text file and a second electronic text file.

[0111] In the embodiment of the present invention, the case information template includes the specific case information of a case. Different case information templates are set according to different case natures or causes of action. Exemplarily, the case information template includes jurisdiction, accepting court, cause of action, case name, party information, complaint, material list, etc. Among them, the cause of action is the case name formed after summarizing the nature of the legal relationship involved in the litigation case. For example, portrait right dispute, alimony dispute, testamentary succession dispute, patent infringement dispute, etc.

[0112] Preferably, the case information template can also be selected according to other case information (such as case name, case details, etc.).

[0113] Specifically, in step 3, calculating the first similarity between the cause of action and any template cause of action in the cause of action template library includes:

[0114] Step 31: Obtain the first sentence vector of the cause of action and the second sentence vector of any template cause of action, and calculate the third similarity between the cause of action and any template cause of action based on the first sentence vector and the second sentence vector.

[0115] Step 32: Generate a case image based on the case, generate a template case image based on any template case, and calculate a fourth similarity between the case image and the template case image.

[0116] Step 33: Calculate the first similarity based on the third similarity and the fourth similarity, and the calculation formula is:

[0117]

[0118] Among them, SI1 is the first similarity, SI3 is the third similarity, SI4 is the fourth similarity, and α3 and α4 are parameter weight coefficients.

[0119] Sentence vectors are generated by adding up the word vectors corresponding to all the words that make up the sentence. Sentences generated by different words with similar word vectors have similar sentence vectors and high sentence similarity, which will have a certain error.

[0120] According to the technical solution of the present invention, when calculating the case similarity, the third similarity is first calculated based on the sentence vector of the case, then the case is used to generate a case image, the fourth similarity is calculated based on the case image, and the case similarity is further calculated based on the third similarity and the fourth similarity, thereby improving the accuracy of similarity calculation.

[0121] Step 4: Perform semantic analysis on the electronic text file based on the case information template, fill in the corresponding fields of the case information template to generate a case information table.

[0122] Step 5: Perform an integrity check on the case information table. If the check fails, a prompt message indicating that the check failed is sent to the user who submitted the case to be filed. If the check passes, proceed to step 6.

[0123] The integrity check is to determine whether the above case information form is completed and there are no blank fields.

[0124] If there are blank fields, the test is determined to have failed, and a prompt message indicating that the test has failed is sent to the user. Exemplarily, the prompt message includes that the test has failed and what the blank fields that have not been filled in are.

[0125] Step 6: Follow the case filing application process according to the case information form.

[0126] Step 7: Assign a data identifier to the case filing material data, divide the case filing material data into N data blocks according to preset rules, then store the data identifier and the preset rules in the first storage module of the RPA terminal accordingly, and divide the N data blocks into first data and second data. After the division is completed, send and store the data identifier and the first data in the first server, and send and store the data identifier and the second data in the second server, wherein the first data includes N1 data blocks, the second data includes N2 data blocks, and N=N1+N2.

[0127] Preferably, when the first data and the second data are generated according to the second preset rule, after the division is completed, the data identifier and the second preset rule are stored correspondingly.

[0128] Preferably, the first server and the second server are servers used to store data within the court.

[0129] The case filing material data is divided into N data blocks according to preset rules, and the N data blocks are further divided into first data and second data. When a user wants to obtain the case filing material data, he needs to obtain data from two servers respectively to generate the above case filing material data, which improves the security of data storage and avoids leakage. At the same time, the case filing material data is divided into different data and stored on different servers, which can reduce the storage burden of the server.

[0130] Preferably, the case filing material data may also be stored on other storage devices other than the server.

[0131] Preferably, the N data blocks may also be divided into N3 data blocks, which are stored on N3 servers or storage devices respectively.

[0132] Specifically, after step 7, the following steps are also included:

[0133] Step 71: When the staff member wants to obtain the case filing material data, he / she sends a data acquisition request to the server through the terminal device, wherein the server includes a first server and a second server, and the data acquisition request includes a data identifier.

[0134] Step 72: After receiving the data acquisition request, the server generates a test frame and then sends the test frame to the terminal device. Based on the response of the terminal device, the response time of receiving the response is calculated, and the number of route hops for sending the test frame to the terminal device is obtained.

[0135] Step 73: When the response time is less than the eighth preset value and the number of routing hops is less than the ninth preset value, extract the data corresponding to the data identifier and send it to the terminal device.

[0136] The eighth preset value and the ninth preset value may be set according to the experience of those skilled in the art or according to actual application scenarios, and the embodiments of the present application are not limited thereto.

[0137] Based on the response time and the number of routing hops, it is determined whether the terminal device sending the data acquisition request is located inside the court. If so, the data corresponding to the data identifier is sent to the terminal device, further improving data security.

[0138] Figure 4 FIG. 1 is a schematic diagram of a structure of an embodiment of an RPA automated case filing processing system provided by the present invention. Figure 4 As shown, the system includes: an RPA terminal 100, a first server 200 and a second server 300, and the RPA terminal 100 includes a data acquisition module 101, a text recognition module 102, a template selection module 103, an information table generation module 104, a case filing application module 105 and a data storage module 106.

[0139] The data acquisition module 101 is used to acquire the case filing material data of the case to be filed, wherein the case filing material data includes a first electronic text file, a scan file and a picture file.

[0140] The text recognition module 102 is used to perform text recognition on the scanned files and the image files, extract text information from each scanned file and each image file, and define the extracted text information as a second electronic text file.

[0141] The template selection module 103 is used to obtain the case cause of the case to be filed from the electronic text file, calculate the similarity between the case cause and any template case cause in the case cause template library, and use the case information template corresponding to the template case cause with the greatest similarity as the case information template of the case to be filed, wherein the electronic text file includes a first electronic text file and a second electronic text file.

[0142] The information table generating module 104 is used to perform semantic analysis on the electronic text file based on the case information template, fill in the corresponding fields of the case information template to generate a case information table.

[0143] The case filing application module 105 is used to perform integrity check on the case information table. If the check fails, a prompt message indicating that the check failed is sent to the user who submitted the case to be filed. If the check passes, the case filing application process is executed according to the case information table.

[0144] The data storage module 106 is used to assign data identifiers to the case filing material data, divide the case filing material data into N data blocks according to preset rules, and then store the data identifiers and the preset rules correspondingly in the first storage module of the RPA terminal 100, and divide the N data blocks into first data and second data. After the division is completed, the data identifier and the first data are sent and stored in the first server 200, and the data identifier and the second data are sent and stored in the second server 300, wherein the first data includes N1 data blocks, and the second data includes N2 data blocks, N=N1+N2.

[0145] It should be understood that, although each step in the flow chart of each embodiment of the present invention is shown in sequence according to the indication of the arrow, these steps are not necessarily performed in sequence according to the order indicated by the arrow. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be performed in other orders. Moreover, at least a portion of the steps in each embodiment may include a plurality of sub-steps or a plurality of stages, and these sub-steps or stages are not necessarily performed at the same time, but can be performed at different times, and the execution order of these sub-steps or stages is not necessarily performed in sequence, but can be performed in turn or alternately with at least a portion of other steps or sub-steps or stages of other steps.

[0146] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program. The above-mentioned program can be stored in a non-volatile computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0147] The above-mentioned embodiments only express the preferred implementation mode of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the patent of the present invention. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

Claims

1. An RPA automated case filing method, applied to an RPA terminal, characterized in that: The steps include: Step 1: The RPA terminal obtains the case filing material data of the case to be filed, wherein the case filing material data includes a first electronic text file, a scanned file, and a picture file; Step 2: performing text recognition on the scanned file and the image file, extracting text information from each scanned file and each image file, and defining the extracted text information as a second electronic text file; Step 3: Obtain the case cause of the case to be filed from the electronic text file, calculate the first similarity between the case cause and any template case cause in the case cause template library, and use the case information template corresponding to the template case cause with the largest first similarity as the case information template of the case to be filed, wherein the electronic text file includes the first electronic text file and the second electronic text file; Step 4: Perform semantic analysis on the electronic text file based on the case information template, fill in the corresponding fields of the case information template to generate a case information table; Step 5: Perform integrity check on the case information table. If the check fails, a prompt message indicating that the check failed is sent to the user who submitted the case to be filed. If the check passes, proceed to step 6. Step 6: Execute the case filing application process according to the case information form; Step 7: assign a data identifier to the case filing material data, divide the case filing material data into N data blocks according to a preset rule, then store the data identifier and the preset rule in the first storage module of the RPA terminal correspondingly, and divide the N data blocks into first data and second data. After the division is completed, send and store the data identifier and the first data in the first server, and send and store the data identifier and the second data in the second server, wherein the first data includes N1 data blocks, and the second data includes N2 data blocks, N=N1+N2; The scanned file and the image file are defined as two types of files. In step 2, the text recognition of the scanned file and the image file includes: Step 21, obtaining any page file of the two types of files, identifying a character region image of the any page file, and preprocessing the character region image to obtain first image data; Step 22: obtaining any character image in the first image data, and performing text recognition on any character image; Step 23, defining the characters in the recognition result whose similarity with any of the character images is greater than a first preset value as pending characters; Step 24, obtaining the character image of the mth undetermined character, drawing the circumscribed rectangle of any character image and the character image respectively, and adjusting the size of any character image based on the circumscribed rectangle of the character image so that the size of the circumscribed rectangle of any character image is equal to that of the circumscribed rectangle of the character image, wherein m is a positive integer from 1 to M, and M is the total number of the undetermined characters; Step 25, obtaining the ith pixel point in any of the character images, and when the pixel value of the ith pixel point is (0, 0, 0) and the pixel value of the ith pixel point in the text image is (0, 0, 0), setting the pixel value of the ith pixel point to (0, 0, 0); otherwise, setting the pixel value of the ith pixel point to (255, 255, 255), and after traversing all the pixels in any of the character images, generating a second text image corresponding to the text image, wherein i is a positive integer from 1 to I, and I is the total number of pixels in any of the character images; Step 26, after traversing all the pending characters, calculating the second similarity between any of the character images and the second character image corresponding to any of the pending characters, and taking the pending character corresponding to the second character image with the largest second similarity as the character recognition result of any of the character images; In the step 22, before obtaining any character image in the first image data, the step includes: Step 221, sequentially intercepting a character sequence diagram in the first image data, wherein the character sequence diagram is a block diagram of a string of characters written continuously; Step 222: extract any character sequence diagram, and draw a circumscribed rectangle of each character in any character sequence diagram; Step 223, obtaining the first height value of any of the character sequence graphs, and obtaining the second height value and center point coordinates of the circumscribed rectangle of each character, calculating the character height standard deviation based on all the second height values, calculating the character center point standard deviation based on the Y-axis coordinates of all the center point coordinates, and defining the ratio of the average value of all the second height values ​​to the first height value as the character height ratio; Step 224: when the character height standard deviation is less than or equal to the fourth preset value, the character center point standard deviation is less than or equal to the fifth preset value, and the character height ratio is less than or equal to the sixth preset value, it is determined that the characters intercepted in any of the character sequence diagrams are in one row; otherwise, it is determined that the characters intercepted in any of the character sequence diagrams are in multiple rows; Step 225: if the characters contained in any of the character sequence diagrams are multiple lines, divide any of the character sequence diagrams into multiple character sequence block images along the direction of the character sequence, sequentially intercept the character sequence diagram in each character sequence block image, then return to the step 222, and after traversing all the character sequence block images, proceed to step 226; if the characters contained in any of the character sequence diagrams are one line, proceed to step 226; Step 226, determine whether all character sequence diagrams in the first image data have been traversed. If not, return to step 222. If they have been traversed, align each line of character sequence in the first image data to generate second image data, and then obtain any character image in the second image data for text recognition.

2. According to claim 1, the RPA automated case filing method is characterized in that: In the step 21, the preprocessing includes: Step 211, performing image smoothing processing on the character region image to obtain a second character region image; Step 212, obtaining the grayscale value of the j-th pixel in the second character area image, when the grayscale value of the j-th pixel is less than the second preset value, setting the grayscale value of the j-th pixel to 0, when the grayscale value of the j-th pixel is greater than or equal to the second preset value, setting the grayscale value of the j-th pixel to 255, and obtaining the third character area image after the processing is completed, wherein j is a positive integer from 1 to J, and J is the total number of pixels of the second character area image; Step 213, respectively obtain the number of pixels of each pixel block in the third character area image; if the number of pixels of any pixel block is less than a third preset value, set the grayscale value of all pixels in any pixel block to 255, wherein the pixel block is a collection of pixels with continuous coordinates.

3. According to claim 1, the RPA automated case filing method is characterized in that: In step 225, dividing any of the character sequence images into a plurality of character sequence block images along the direction of the character sequence includes: Step 2251, obtaining the number of character columns in any of the character sequence diagrams, and marking each column of characters in order from left to right; Step 2252: Set the position between the g-1th column character and the gth column character as the position to be divided, where g is a positive integer from 2 to G, and G is the total number of character columns; Step 2253, respectively calculate the sum of pixel values ​​of each row of pixels from the character in the 1st column to the character in the g-1th column, obtain the minimum value of the sum of pixel values, and determine whether the minimum value of the sum of pixel values ​​is less than or equal to the seventh preset value. If so, proceed to step 2254; if so, proceed to step 2255. Step 2254, define the position to be divided as the division position, then determine whether g is equal to G, if so, proceed to step 2256, if not, set the mark of the g-th column character to 1, adjust the mark of each column character after the g-th column character in turn, and then return to step 2252; Step 2255, g=g-1, determine whether g is equal to 2, if yes, define the position between the first column character image and the second column character image as the division position, set the mark of the second column character to 1, adjust the mark of each column character after the second column character in turn, and then return to step 2252, if no, return to step 2253; Step 2256: Divide any of the character sequence images into a plurality of character sequence block images based on the division positions.

4. According to claim 1, the RPA automated case filing method is characterized in that: In step 3, calculating the first similarity between the case and any template case in the case template library includes: Step 31, obtaining a first sentence vector of the case and a second sentence vector of any of the template cases, and calculating a third similarity between the case and any of the template cases based on the first sentence vector and the second sentence vector; Step 32: Generate a case image based on the case, generate a template case image based on any of the template cases, and calculate a fourth similarity between the case image and the template case image; Step 33: Calculate the first similarity based on the third similarity and the fourth similarity, using the following calculation formula: Among them, SI1 is the first similarity, SI3 is the third similarity, SI4 is the fourth similarity, and α3 and α4 are parameter weight coefficients.

5. According to claim 1, the RPA automated case filing method is characterized in that: After step 7, the following steps are also included: Step 71: When the staff member wants to obtain the case filing material data, the staff member sends a data acquisition request to the server through the terminal device, wherein the server includes the first server and the second server, and the data acquisition request includes the data identifier; Step 72: After receiving the data acquisition request, the server generates a test frame, and then sends the test frame to the terminal device. Based on the response of the terminal device, the server calculates the response time of receiving the response, and obtains the number of route hops for sending the test frame to the terminal device. Step 73: When the response time is less than an eighth preset value and the number of routing hops is less than a ninth preset value, extract data corresponding to the data identifier and send it to the terminal device.

6. An RPA automated case filing processing system, used to implement the RPA automated case filing processing method according to any one of claims 1 to 5, characterized in that: include: An RPA terminal, a first server and a second server, wherein the RPA terminal includes a data acquisition module, a text recognition module, a template selection module, an information table generation module, a case filing application module and a data storage module; The data acquisition module is used to acquire the case filing material data of the case to be filed, wherein the case filing material data includes a first electronic text file, a scan file and a picture file; The text recognition module is used to perform text recognition on the scanned file and the image file, extract text information from each scanned file and each image file, and define the extracted text information as a second electronic text file; The template selection module is used to obtain the case information of the case to be filed from the electronic text file, calculate the similarity between the case information and any template case information in the case information template library, and use the case information template corresponding to the template case information with the greatest similarity as the case information template of the case to be filed, wherein the electronic text file includes the first electronic text file and the second electronic text file; The information table generating module is used to perform semantic analysis on the electronic text file based on the case information template, fill in the corresponding fields of the case information template to generate a case information table; The case filing application module is used to perform integrity detection on the case information table. If the detection fails, a prompt message indicating that the detection fails is sent to the user who submitted the case to be filed. If the detection passes, the case filing application process is executed according to the case information table. The data storage module is used to assign a data identifier to the case filing material data, divide the case filing material data into N data blocks according to a preset rule, and then store the data identifier and the preset rule in a first storage module of the RPA terminal correspondingly, and divide the N data blocks into first data and second data. After the division is completed, the data identifier and the first data are sent and stored in the first server, and the data identifier and the second data are sent and stored in the second server, wherein the first data includes N1 data blocks, and the second data includes N2 data blocks, N=N1+N2; The scanned file and the image file are defined as two types of files, and the text recognition of the scanned file and the image file includes: Step 21, obtaining any page file of the two types of files, identifying a character region image of the any page file, and preprocessing the character region image to obtain first image data; Step 22: obtaining any character image in the first image data, and performing text recognition on any character image; Step 23, defining the characters in the recognition result whose similarity with any of the character images is greater than a first preset value as pending characters; Step 24, obtaining the character image of the mth undetermined character, drawing the circumscribed rectangle of any character image and the character image respectively, and adjusting the size of any character image based on the circumscribed rectangle of the character image so that the size of the circumscribed rectangle of any character image is equal to that of the circumscribed rectangle of the character image, wherein m is a positive integer from 1 to M, and M is the total number of the undetermined characters; Step 25, obtaining the ith pixel point in any of the character images, and when the pixel value of the ith pixel point is (0, 0, 0) and the pixel value of the ith pixel point in the text image is (0, 0, 0), setting the pixel value of the ith pixel point to (0, 0, 0); otherwise, setting the pixel value of the ith pixel point to (255, 255, 255), and after traversing all the pixels in any of the character images, generating a second text image corresponding to the text image, wherein i is a positive integer from 1 to I, and I is the total number of pixels in any of the character images; Step 26, after traversing all the pending characters, calculating the second similarity between any of the character images and the second character image corresponding to any of the pending characters, and taking the pending character corresponding to the second character image with the largest second similarity as the character recognition result of any of the character images; In the step 22, before obtaining any character image in the first image data, the step includes: Step 221, sequentially intercepting a character sequence diagram in the first image data, wherein the character sequence diagram is a block diagram of a string of characters written continuously; Step 222: extract any character sequence diagram, and draw a circumscribed rectangle of each character in any character sequence diagram; Step 223, obtaining the first height value of any of the character sequence graphs, and obtaining the second height value and center point coordinates of the circumscribed rectangle of each character, calculating the character height standard deviation based on all the second height values, calculating the character center point standard deviation based on the Y-axis coordinates of all the center point coordinates, and defining the ratio of the average value of all the second height values ​​to the first height value as the character height ratio; Step 224: when the character height standard deviation is less than or equal to the fourth preset value, the character center point standard deviation is less than or equal to the fifth preset value, and the character height ratio is less than or equal to the sixth preset value, it is determined that the characters intercepted in any of the character sequence diagrams are in one row; otherwise, it is determined that the characters intercepted in any of the character sequence diagrams are in multiple rows; Step 225: if the characters contained in any of the character sequence diagrams are multiple lines, divide any of the character sequence diagrams into multiple character sequence block images along the direction of the character sequence, sequentially intercept the character sequence diagram in each character sequence block image, then return to the step 222, and after traversing all the character sequence block images, proceed to step 226; if the characters contained in any of the character sequence diagrams are one line, proceed to step 226; Step 226, determine whether all character sequence diagrams in the first image data have been traversed. If not, return to step 222. If they have been traversed, align each line of character sequence in the first image data to generate second image data, and then obtain any character image in the second image data for text recognition.

Images