A procuratorate civil supervision method and device based on text positioning technology

Abstract

The application discloses a procuratorial office civil supervision method and device based on a text positioning technology, relates to the technical field of text processing, and realizes efficient auxiliary perusal of files and documents, and the main technical scheme of the application is as follows: the type and the trial level of a judgment document are determined; a combination information obtained from the type and the trial level is searched for a preset mapping relationship between the combination information and a dimension tree, a target dimension tree corresponding to the judgment document is acquired, the target dimension tree is composed of a root node with a tree structure and leaf nodes under the root node, and the root node and the leaf nodes are all previously bound with corresponding node labels; a preset document analysis model is used to analyze the judgment document, dimensions contained in the judgment document and paragraph information corresponding to the dimensions are obtained; the paragraph information corresponding to the dimensions is bound with node labels contained in the target dimension tree; and when a trigger click on a target node label is received, paragraph information corresponding to the target node label is positioned. The application is mainly applied to auxiliary perusal of content information of the judgment document.

Description

Technical Field

[0001] This invention relates to the field of text processing technology, and in particular to a method and apparatus for civil supervision by procuratorates based on text positioning technology. Background Technology

[0002] Judgments are records of the trial process and outcomes of the People's Court. They serve as the carrier of the results of litigation activities and the sole evidence for the People's Court to determine and allocate the substantive rights and obligations of the parties. A well-structured, complete, and logically rigorous judgment is not only proof of the parties' rights and obligations but also an important basis for higher People's Courts to supervise the civil trial activities of lower People's Courts. Judgments have their own specific rules of creation, and common types include civil judgments, criminal judgments, administrative judgments, and other general litigation documents.

[0003] Currently, access to prosecutorial documents is still limited to paper or electronic versions. However, prosecutors need to read the entire document to obtain the desired content, which consumes a lot of their time and energy. Summary of the Invention

[0004] In view of this, the present invention provides a method and device for civil supervision by the procuratorate based on text positioning technology. The main purpose is to quickly locate the specified content information in the case file according to the need for reviewing the case file, thereby achieving efficient assistance in reviewing the case file and greatly improving the office efficiency of the procurator.

[0005] To achieve the above objectives, the present invention mainly provides the following technical solutions:

[0006] The first aspect of this application provides a method for civil supervision by procuratorates based on text positioning technology, the method comprising:

[0007] Determine the types and levels of review of judgment documents;

[0008] By finding the pre-defined mapping relationship between the combined information obtained from the type and the level of review and the dimension tree, the target dimension tree corresponding to the judgment document is obtained. The target dimension tree is composed of a root node with a tree structure and leaf nodes below it. The root node and the leaf nodes are pre-bound with corresponding node labels.

[0009] The judgment document is parsed using a preset document parsing model to obtain the dimensions contained in the judgment document and the paragraph information corresponding to the dimensions;

[0010] Bind the paragraph information corresponding to the dimension to the node labels contained in the target dimension tree;

[0011] When a trigger is received to click on a target node label, the paragraph information corresponding to the target node label is located.

[0012] In some modified embodiments of the first aspect of this application, the step of parsing the judgment document using a preset document parsing model to obtain the dimensions contained in the judgment document and the paragraph information corresponding to the dimensions includes:

[0013] By utilizing the underlying model of the pre-defined document parsing model, the text corresponding to the judgment document is processed to obtain the word vector representation contained therein;

[0014] Using the sentence representation layer of the model, the document text after word vector representation is processed to obtain the sentence vector representation corresponding to the document text;

[0015] The sentence vectors, represented by sentence vectors, are input into a multi-layer bidirectional recurrent neural network model for processing to obtain the segment vector representations corresponding to the document text.

[0016] The segment vectors, represented by segment vectors, are input into the softmax layer and the CRF layer to predict sequence labels, and the output is the dimension matched by each paragraph in the document text.

[0017] In some modified embodiments of the first aspect of this application, the step of binding the paragraph information corresponding to the dimension with the node labels contained in the target dimension tree includes:

[0018] Obtain the dimensions contained in the judgment document and the paragraph information corresponding to the dimensions;

[0019] Retrieve the node labels contained in the target dimension tree;

[0020] By comparing each dimension contained in the judgment document with the node labels contained in the target dimension tree, it is determined whether the similarity between the dimension and the node label is greater than a preset threshold.

[0021] If so, then perform a matching operation between the dimension and the node label;

[0022] Based on the matching relationship between the dimension and the node label, the paragraph information corresponding to the dimension is bound to the node label.

[0023] In some modified embodiments of the first aspect of this application, the step of locating the paragraph information corresponding to the target node label when a trigger click on the target node label is received includes:

[0024] When a trigger is received to click on a target node label, the paragraph information bound to the target node label is determined;

[0025] The page displays the location of the paragraph information within the judgment document and displays the paragraph information using a feature-based display method.

[0026] In some modified embodiments of the first aspect of this application, displaying the paragraph information in a feature-based manner includes:

[0027] The paragraph information is displayed in a highlighted manner; and / or,

[0028] The paragraph information is displayed using a combination of enlarged text and an additional pop-up window.

[0029] A second aspect of this application provides a civil supervision device for procuratorates based on text positioning technology, the device comprising:

[0030] The unit of determination is used to identify the type and level of the judgment document.

[0031] The acquisition unit is used to obtain the target dimension tree corresponding to the judgment document by finding the preset mapping relationship between the combined information obtained by the type and the level of review and the dimension tree. The target dimension tree is composed of a root node with a tree structure and leaf nodes below it. The root node and the leaf nodes are pre-bound with corresponding node labels.

[0032] The parsing unit is used to parse the judgment document using a preset document parsing model to obtain the dimensions contained in the judgment document and the paragraph information corresponding to the dimensions;

[0033] A binding unit is used to bind the paragraph information corresponding to the dimension parsed by the parsing unit to the node labels contained in the target dimension tree obtained by the acquisition unit;

[0034] The positioning unit is used to locate the paragraph information corresponding to the target node label when a trigger click on the target node label is received.

[0035] In some modified embodiments of the second aspect of this application, the parsing unit includes:

[0036] The processing module is used to process the text of the judgment document to obtain the word vector representation contained therein by utilizing the underlying model of the preset document parsing model;

[0037] The processing module is also used to process the document text after word vector representation processing using the model sentence representation layer to obtain the sentence vector representation corresponding to the document text;

[0038] The processing module is also used to input the sentence vectors represented by sentence vectors into a multi-layer bidirectional recurrent neural network model for processing, so as to obtain the segment vector representations corresponding to the document text;

[0039] The output module is used to input the segment vectors, which are represented by segment vectors obtained by the processing module, into the softmax layer and the CRF layer for sequence label prediction, and output the dimension matched by each paragraph in the document text.

[0040] In some modified embodiments of the second aspect of this application, the binding unit includes:

[0041] The acquisition module is used to acquire the dimensions contained in the judgment document and the paragraph information corresponding to the dimensions;

[0042] The acquisition module is also used to acquire the node labels contained in the target dimension tree;

[0043] The judgment module is used to determine whether the similarity between the dimensions contained in the judgment document and the node labels contained in the target dimension tree is greater than a preset threshold by comparing each dimension with the node labels contained in the target dimension tree.

[0044] The matching module is used to perform matching on the dimension and the node label when the judgment module determines that the similarity between the dimension and the node label is greater than a preset threshold.

[0045] The binding module is used to bind the paragraph information corresponding to the dimension to the node label based on the matching relationship between the dimension and the node label obtained by the matching module.

[0046] In some modified embodiments of the second aspect of this application, the positioning unit includes:

[0047] The determination module is used to determine the paragraph information bound to the target node label when a trigger click on the target node label is received;

[0048] The positioning module is used to locate and jump to the position of the paragraph information determined by the determining module in the judgment document within the page display area;

[0049] The display module is used to display the paragraph information in a feature-based display manner.

[0050] In some modified embodiments of the second aspect of this application, the display module is further configured to:

[0051] The paragraph information is displayed in a highlighted manner; and / or,

[0052] The paragraph information is displayed using a combination of enlarged text and an additional pop-up window.

[0053] A third aspect of this application provides a storage medium comprising a stored program, wherein, when the program is executed, it controls the device containing the storage medium to perform the aforementioned text-based location technology-based civil supervision method for procuratorates.

[0054] A fourth aspect of this application provides an electronic device, the device including at least one processor, and at least one memory and bus connected to the processor;

[0055] The processor and the memory communicate with each other via the bus.

[0056] The processor is used to invoke program instructions in the memory to execute the civil supervision method of the procuratorate based on text positioning technology as described above.

[0057] By employing the above-described technical solution, the technical solution provided by the present invention has at least the following advantages:

[0058] This invention provides a method and apparatus for civil supervision by procuratorates based on text positioning technology. The invention obtains a corresponding target dimension tree based on the type and level of the judgment document, parses the judgment document using a preset document parsing model to obtain the dimensions contained in the judgment document and the corresponding paragraph information, and binds the paragraph information corresponding to the dimensions to the root node label or leaf node label of the target dimension tree. Therefore, when a target node label is clicked, the bound paragraph information can be located based on that node label. Compared with existing technologies, this invention solves the technical problem of the high manpower and time costs associated with reviewing case files in procuratorate work. This invention can quickly locate the specified content information location in a case file, thereby efficiently assisting in the review of case files and greatly improving the work efficiency of prosecutors.

[0059] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it in accordance with the contents of the specification, and in order to make the above and other objects, features and advantages of the present invention more apparent and understandable, specific embodiments of the present invention are described below. Attached Figure Description

[0060] 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:

[0061] Figure 1 A flowchart of a civil supervision method for procuratorates based on text positioning technology is provided for an embodiment of the present invention;

[0062] Figure 2 This is an example of a human-computer interaction interface for viewing court documents using a dimension tree, as exemplified in an embodiment of the present invention;

[0063] Figure 3 A flowchart of another civil supervision method for procuratorates based on text positioning technology provided in an embodiment of the present invention;

[0064] Figure 4 This is an example of a human-computer interaction interface for assisting civil procuratorial supervision, as described in an embodiment of the present invention.

[0065] Figure 5 A block diagram illustrating the components of a civil supervision device for procuratorates based on text positioning technology, provided in an embodiment of the present invention;

[0066] Figure 6 A block diagram illustrating the composition of another civil supervision device for procuratorates based on text positioning technology, provided in an embodiment of the present invention;

[0067] Figure 7 An electronic device for civil supervision by the procuratorate based on text positioning technology, provided as an embodiment of the present invention. Detailed Implementation

[0068] Exemplary embodiments of the invention will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the invention are shown in the drawings, it should be understood that the invention may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this invention will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art.

[0069] This invention provides a method for civil supervision by procuratorates based on text location technology, such as... Figure 1 As shown, this method, based on the need to search case files, can quickly locate the specified content information within the case file. The embodiments of this invention provide the following specific steps:

[0070] 101. Determine the types and levels of judicial documents.

[0071] In this embodiment of the invention, judgment documents have their own specific production rules. Common types include civil judgment documents, criminal judgment documents, administrative judgment documents, and other general litigation documents. Documents of the same type generally contain the same dimensions, but there can be differences in details. For example, civil judgment documents are divided into first instance, second instance, retrial, etc., indicating different levels of jurisdiction. Therefore, even though judgment documents of the same type at different levels contain the same dimensions, the content of those dimensions will differ.

[0072] Therefore, in this embodiment of the invention, after receiving the uploaded judgment document, the type and level of the judgment document are determined first, that is, both the type and level of the judgment document are fully considered in order to implement the subsequent civil supervision method of the procuratorate based on text positioning technology.

[0073] 102. By finding the pre-defined mapping relationship between the combined information obtained from the category and the level of review and the dimension tree, the target dimension tree corresponding to the judgment document is obtained.

[0074] The target dimension tree consists of a root node with a tree structure and leaf nodes below it. Both the root node and leaf nodes are pre-bound with corresponding node labels.

[0075] In this embodiment of the invention, combined information is obtained based on the type and level of the judgment document, and then a corresponding dimension tree is pre-defined for the combined information, which is equivalent to pre-determining a corresponding dimension tree for a specific level of a certain type of judgment document.

[0076] It should be noted that the node labels in this dimension tree are pre-defined based on high-frequency browsing needs. For example, a human-computer interaction interface can be provided to users (such as prosecutors) to display the node labels of the dimension tree. Figure 2 The example of a human-computer interaction interface for using a dimensional tree to view court documents is as follows. Figure 2 The dimension tree for display exemplifies a "root node label" as "case information," and the "leaf node labels" under this "root node label" include: "case number," "judgment date," "trial process," and "trial team members."

[0077] 103. Use a preset document parsing model to parse the judgment document and obtain the dimensions contained in the judgment document and the paragraph information corresponding to the dimensions.

[0078] In this embodiment of the invention, the document segmentation problem is regarded as a multi-classification problem. Based on the hierarchical structure of legal documents and the fact that text paragraphs have a certain sequence relationship, a hierarchical classification model based on SR-RCNN is proposed to achieve document segmentation. Furthermore, this embodiment of the invention pre-establishes matching document parsing models for different types of judgment documents, so that the document parsing model is applicable to different levels of review of the same document type.

[0079] Specifically, by using this preset document parsing model to parse judicial documents, it is possible not only to clearly identify the paragraph information contained in the judicial documents, but also to determine the dimension that each paragraph information matches.

[0080] Furthermore, by dividing the labeled data into training and testing sets, we trained segmentation models for first-instance, second-instance, and retrial documents. The final segmentation model for first-instance documents achieved an average accuracy of 99.2%, the segmentation model for second-instance documents achieved an average accuracy of 98.7%, and the segmentation model for retrial documents achieved an average accuracy of 97.8%, which can support the application.

[0081] 104. Bind the paragraph information corresponding to the dimension to the node labels contained in the target dimension tree.

[0082] In this embodiment of the invention, after parsing the judgment document to obtain the dimensions contained in the judgment document and the paragraph information corresponding to the dimensions, the paragraph information is bound to the node label of the target dimension in combination with the previously obtained target dimension tree of the judgment document. The specific binding operation needs to follow the following rule: if the dimension corresponding to the paragraph information has the same meaning as the node label, then the paragraph information is bound to the node label.

[0083] 105. When a trigger is received to click on the target node label, locate the paragraph information corresponding to the target node label.

[0084] In this embodiment of the invention, after establishing the binding relationship between paragraph information and node tags, when a trigger is received to click a target node tag, the paragraph information corresponding to the target node tag is located. Thus, based on the node tags contained in the displayed dimension tree, the specified paragraph information position of the judgment document can be quickly located, thereby achieving efficient assistance in reviewing the specified content information of the judgment document.

[0085] This invention provides a method for civil supervision by procuratorates based on text positioning technology. The method involves obtaining a corresponding target dimension tree based on the type and level of the judgment document, parsing the judgment document using a preset document parsing model to obtain the dimensions contained in the judgment document and the corresponding paragraph information, and binding the paragraph information to the root node label or leaf node label of the target dimension tree. Therefore, when a target node label is clicked, the bound paragraph information can be located based on that node label. Compared to existing technologies, this method solves the technical problem of the high manpower and time costs associated with reviewing case files in procuratorate work. This invention can quickly locate the specified content information within a case file, thus efficiently assisting in the review of case files and greatly improving the work efficiency of procurators.

[0086] To provide a more detailed explanation of the above embodiments, this invention also provides another method for civil supervision by procuratorates based on text positioning technology, such as... Figure 3As shown, the following specific steps are provided in this embodiment of the invention:

[0087] 201. Determine the types and levels of judicial documents.

[0088] In this embodiment of the invention, for the description of this step, please refer to step 101, which will not be repeated here.

[0089] 202. By finding the pre-defined mapping relationship between the combined information obtained from the category and the level of review and the dimension tree, the target dimension tree corresponding to the judgment document is obtained.

[0090] The target dimension tree consists of a root node with a tree structure and leaf nodes below it. Both the root node and leaf nodes are pre-bound with corresponding node labels.

[0091] In this embodiment of the invention, for the description of this step, please refer to step 102, which will not be repeated here.

[0092] 203. Use the preset document parsing model to parse the judgment document and obtain the dimensions contained in the judgment document and the paragraph information corresponding to the dimensions.

[0093] In this embodiment of the invention, this step is further described in detail below:

[0094] In this embodiment of the invention, document segmentation is treated as a multi-classification problem. Based on the hierarchical structure of legal documents and the sequential relationship between text paragraphs, a hierarchical classification model based on SR-RCNN is proposed to achieve document segmentation. The model's bottom layer uses word embedding to represent the document text as word vectors. At the sentence representation layer, an SR-RCNN model combining text-CNN and Bi-RNN captures the contextual semantic information and sequence information of sentences to obtain the document's sentence vector representation. The obtained sentence vectors are then input into a multi-layer bidirectional RNN model, leveraging its powerful long-sequence information modeling capabilities to represent the document text as paragraph vectors. The final document paragraph representation vectors are then input into a softmax layer and a CRF layer for sequence label prediction, outputting the dimension matching each paragraph within the document text.

[0095] In this embodiment of the invention, the preset document parsing model is used to parse the judgment document, thereby not only clearly identifying the paragraph information contained in the judgment document, but also determining the dimension that each paragraph information matches.

[0096] Furthermore, to simplify the mapping relationship between storage dimensions and paragraph information, it is also possible to record only the mapping relationship between storage dimensions and paragraph numbers in the document text where the paragraph information is located.

[0097] 204. Bind the paragraph information corresponding to the dimension to the node labels contained in the target dimension tree.

[0098] In this embodiment of the invention, this step is further described in detail below:

[0099] First, obtain the dimensions contained in the judgment document and the paragraph information corresponding to each dimension. Then, obtain the node labels contained in the target dimension tree.

[0100] Secondly, by comparing each dimension contained in the judgment document with the node labels contained in the target dimension tree, it is determined whether the similarity between the dimension and the node label is greater than a preset threshold.

[0101] In this embodiment of the invention, the similarity of words is used to determine which node label represents the same meaning as each dimension.

[0102] Finally, if a node label with the same dimension and meaning is found, the dimension and node label are matched. Based on the matching relationship between the dimension and node label, the paragraph information corresponding to the dimension can be bound to the node label.

[0103] 205. When a trigger is received to click the target node label, determine the paragraph information bound to the target node label.

[0104] 206. Locate and jump to the position of the paragraph information in the judgment document within the page display area, and display the paragraph information in a feature-based manner.

[0105] In this embodiment of the invention, based on the position of the paragraph information in the judgment document, the system locates and jumps to the corresponding target position within the page display area. At the target position, the paragraph information is displayed using a feature-based display method. Specifically, the feature-based display method may include, but is not limited to: displaying the paragraph information in a highlighted manner, or displaying the paragraph information in a combination of enlarged text and an additional pop-up box, or a display method obtained by superimposing multiple feature-based display methods.

[0106] In this embodiment of the invention, steps 205-206 above provide a specific execution process for quickly locating specified content information within a judgment document based on the search requirements. This process can be combined with... Figure 4 The following is a detailed explanation of the example of the human-computer interaction interface for assisting civil procuratorial supervision. Figure 4 The left side of the table contains the judgment document, and the right side contains the dimension tree. The judgment document and the dimension tree are in a corresponding relationship. That is, the dimensions and their corresponding paragraph information in the judgment document obtained after parsing are bound to the node labels contained in the dimension tree. See steps 201-204 for details, which will not be repeated here.

[0107] So, for Figure 4The example human-computer interaction interface shows a dimension tree on the right with the "root node label" set to "Party Information". This "root node label" contains "leaf node labels" set to "Applicant" and "Respondent". Figure 4 As shown, when a target node label "Applicant" is clicked, the paragraph information bound to that target node label can be determined. Based on the position of the paragraph information in the judgment document, the system navigates to the corresponding target location within the page display area. At the target location, the paragraph information is displayed using a feature-based display method, such as highlighting the paragraph information. Figure 4 The portion selected on the left is displayed in this way for this feature. Figure 4 This is merely one example of an implementation method. Alternatively, the paragraph information can be displayed using a combination of enlarged text and an additional pop-up box. This embodiment of the invention does not limit the specific implementation method. However, it should be understood that the purpose of this feature display is to highlight the paragraph information that has been quickly located so that it can be clearly viewed.

[0108] Furthermore, as a response to the above Figure 1 , Figure 2 The present invention provides a civil supervision device for procuratorates based on text positioning technology to implement the method shown. This device embodiment corresponds to the aforementioned method embodiment. For ease of reading, this device embodiment will not repeat the details of the aforementioned method embodiment, but it should be understood that the device in this embodiment can implement all the contents of the aforementioned method embodiment. This device is used to assist in efficiently reviewing the content of judgment documents, specifically as follows... Figure 5 As shown, the device includes:

[0109] Unit 31 is used to determine the type and level of the judgment document;

[0110] The acquisition unit 32 is used to obtain the target dimension tree corresponding to the judgment document by searching the preset mapping relationship between the combined information obtained by the type and the level of review and the dimension tree. The target dimension tree is composed of a root node with a tree structure and leaf nodes thereunder. The root node and the leaf nodes are pre-bound with corresponding node labels.

[0111] The parsing unit 33 is used to parse the judgment document using a preset document parsing model to obtain the dimensions contained in the judgment document and the paragraph information corresponding to the dimensions;

[0112] Binding unit 34 is used to bind the paragraph information corresponding to the dimension parsed by parsing unit 33 with the node labels contained in the target dimension tree obtained by acquisition unit 32;

[0113] The positioning unit 35 is used to locate the paragraph information corresponding to the target node label when a trigger click on the target node label is received.

[0114] Furthermore, such as Figure 6 As shown, the parsing unit 33 includes:

[0115] Processing module 331 is used to process the text of the judgment document to obtain the word vector representation contained therein by utilizing the underlying model of the preset document parsing model;

[0116] The processing module 331 is further configured to use the model sentence representation layer to process the document text after word vector representation processing to obtain the sentence vector representation corresponding to the document text;

[0117] The processing module 331 is further configured to input the sentence vector represented by the sentence vector into a multi-layer bidirectional recurrent neural network model for processing, so as to obtain the segment vector representation corresponding to the document text;

[0118] The output module 332 is used to input the segment vectors represented by the segment vectors obtained by the processing module 331 into the softmax layer and the CRF layer for sequence label prediction, and output the dimension matched by each paragraph in the document text.

[0119] Furthermore, such as Figure 6 As shown, the binding unit 34 includes:

[0120] The acquisition module 341 is used to acquire the dimensions contained in the judgment document and the paragraph information corresponding to the dimensions;

[0121] The acquisition module 341 is further configured to acquire the node labels contained in the target dimension tree;

[0122] The judgment module 342 is used to determine whether the similarity between the dimension and the node label is greater than a preset threshold by comparing each dimension contained in the judgment document with the node label contained in the target dimension tree.

[0123] The matching module 343 is used to perform matching on the dimension and the node label when the judgment module 342 determines that the similarity between the dimension and the node label is greater than a preset threshold.

[0124] The binding module 344 is used to bind the paragraph information corresponding to the dimension to the node label according to the matching relationship between the dimension and the node label obtained by the matching module 343.

[0125] Furthermore, such as Figure 6 As shown, the positioning unit 35 includes:

[0126] The determination module 351 is used to determine the paragraph information bound to the target node label when a trigger click on the target node label is received;

[0127] The positioning module 352 is used to locate and jump to the position of the paragraph information determined by the determining module 351 in the judgment document within the page display area;

[0128] Display module 353 is used to display the paragraph information in a feature-based display manner.

[0129] Furthermore, such as Figure 6 As shown, the display module 353 is further used for:

[0130] The paragraph information is displayed in a highlighted manner; and / or,

[0131] The paragraph information is displayed using a combination of enlarged text and an additional pop-up window.

[0132] In summary, this invention provides a method and apparatus for civil supervision by procuratorates based on text positioning technology. This invention obtains a corresponding target dimension tree based on the type and level of the judgment document, parses the judgment document using a preset document parsing model to obtain the dimensions contained in the judgment document and the corresponding paragraph information, and binds the paragraph information corresponding to the dimensions to the root node label or leaf node label of the target dimension tree. Therefore, when a target node label is clicked, the bound paragraph information can be located based on that node label. Compared with existing technologies, this invention solves the technical problem of the high manpower and time costs associated with reviewing case files in procuratorate work. This invention can quickly locate the specified content information location in a case file, thereby efficiently assisting in the review of case files and greatly improving the work efficiency of prosecutors.

[0133] The procuratorate's civil supervision device based on text positioning technology includes a processor and a memory. The aforementioned determining unit, acquiring unit, parsing unit, binding unit, and positioning unit are all stored in the memory as program units. The processor executes the aforementioned program units stored in the memory to achieve the corresponding functions.

[0134] The processor contains a kernel, which retrieves the corresponding program units from memory. One or more kernels can be configured, and by adjusting kernel parameters, the processor can quickly locate specific information within case files according to the needs of document retrieval, thereby efficiently assisting in document retrieval and greatly improving the prosecutor's work efficiency.

[0135] This invention provides a storage medium storing a program that, when executed by a processor, implements the procuratorate's civil supervision method based on text positioning technology.

[0136] This invention provides a processor for running a program, wherein the program executes the procuratorate civil supervision method based on text positioning technology.

[0137] This invention provides an electronic device 40, such as... Figure 7 As shown, the device includes at least one processor 401, and at least one memory 402 and bus 403 connected to the processor 401; wherein the processor 401 and the memory 402 communicate with each other through the bus 403; the processor 401 is used to call program instructions in the memory 402 to execute the above-mentioned civil supervision method of the procuratorate based on text positioning technology.

[0138] The devices mentioned in this article can be servers, PCs, tablets, mobile phones, etc.

[0139] This application also provides a computer program product, which, when executed on a data processing device, is suitable for executing a program that initializes the following method steps:

[0140] A method for civil supervision by procuratorates based on text localization technology includes: determining the type and level of a judgment document; obtaining a target dimension tree corresponding to the judgment document by finding a preset mapping relationship between the combined information obtained from the type and level of the judgment document and a dimension tree, wherein the target dimension tree consists of a root node with a tree structure and leaf nodes thereunder, and the root node and the leaf nodes are pre-bound with corresponding node labels; parsing the judgment document using a preset document parsing model to obtain the dimensions contained in the judgment document and the paragraph information corresponding to the dimensions; binding the paragraph information corresponding to the dimensions with the node labels contained in the target dimension tree; and locating the paragraph information corresponding to the target node label when a trigger click on the target node label is received.

[0141] Furthermore, the step of parsing the judgment document using a preset document parsing model to obtain the dimensions contained in the judgment document and the paragraph information corresponding to the dimensions includes: using the lower layer of the preset document parsing model to process the document text corresponding to the judgment document to obtain the word vector representations contained therein; using the sentence representation layer of the model to process the document text after word vector representation processing to obtain the sentence vector representations corresponding to the document text; inputting the sentence vectors after sentence vector representation into a multi-layer bidirectional recurrent neural network model for processing to obtain the segment vector representations corresponding to the document text; inputting the segment vectors after segment vector representation into a softmax layer and a CRF layer for sequence label prediction, and outputting the dimensions matched by each paragraph contained in the document text.

[0142] Furthermore, the step of binding the paragraph information corresponding to the dimension with the node tags contained in the target dimension tree includes: obtaining the dimensions contained in the judgment document and the paragraph information corresponding to the dimensions; obtaining the node tags contained in the target dimension tree; comparing each dimension contained in the judgment document with the node tags contained in the target dimension tree to determine whether the similarity between the dimension and the node tags is greater than a preset threshold; if so, matching the dimension and the node tags; and binding the paragraph information corresponding to the dimension with the node tags according to the matching relationship between the dimension and the node tags.

[0143] Furthermore, the step of locating the paragraph information corresponding to the target node tag when a trigger click on the target node tag is received includes: when a trigger click on the target node tag is received, determining the paragraph information bound to the target node tag; locating and jumping to the position of the paragraph information in the judgment document within the page display area, and displaying the paragraph information in a feature display manner.

[0144] Furthermore, displaying the paragraph information in a feature-based manner includes: displaying the paragraph information in a highlighted manner; and / or, displaying the paragraph information in a combination of enlarged text and an additional pop-up box.

[0145] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products 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.

[0146] In a typical configuration, the device includes one or more processors (CPUs), memory, and a bus. The device may also include input / output interfaces, network interfaces, etc.

[0147] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, like read-only memory (ROM) or flash RAM, and memory includes at least one memory chip. Memory is an example of computer-readable media.

[0148] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0149] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0150] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. 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 product 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.

[0151] The above are merely embodiments of this application and are not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A method for civil supervision by procuratorates based on text positioning technology, characterized in that, The method includes: The types and levels of judgment documents shall be determined, and the types of judgment documents shall include at least civil judgment documents, criminal judgment documents, and administrative judgment documents; By finding the pre-defined mapping relationship between the combined information obtained from the type and level of review and the dimension tree, the target dimension tree corresponding to the judgment document is obtained. The target dimension tree consists of a root node with a tree structure and leaf nodes below it. The root node and the leaf nodes are pre-bound with corresponding node labels. Among them, the node label bound to the root node is case information, and the node labels bound to the leaf nodes under the root node are at least: case number, judgment date, trial process, and members of the trial team. The judgment document is parsed using a pre-defined document parsing model to obtain the dimensions contained in the judgment document and the paragraph information corresponding to the dimensions. This includes: using the lower layer of the pre-defined document parsing model to process the document text corresponding to the judgment document to obtain word vector representations; using the sentence representation layer of the model to process the document text after word vector representation processing to obtain sentence vector representations corresponding to the document text; inputting the sentence vectors into a multi-layer bidirectional recurrent neural network model for processing to obtain segment vector representations corresponding to the document text; and inputting the segment vectors into a softmax layer and a CRF layer for sequence label prediction, outputting the dimensions matched by each paragraph within the document text. The paragraph information corresponding to the dimension is bound to the node label contained in the target dimension tree. The binding operation must follow the following rule: if the dimension corresponding to the paragraph information has the same meaning as the node label, then the paragraph information is bound to the node label. When a trigger is received to click on a target node label, the paragraph information corresponding to the target node label is located.

2. The method according to claim 1, characterized in that, The step of binding the paragraph information corresponding to the dimension with the node labels contained in the target dimension tree includes: Obtain the dimensions contained in the judgment document and the paragraph information corresponding to the dimensions; Retrieve the node labels contained in the target dimension tree; By comparing each dimension contained in the judgment document with the node labels contained in the target dimension tree, it is determined whether the similarity between the dimension and the node label is greater than a preset threshold. If so, then perform a matching operation between the dimension and the node label; Based on the matching relationship between the dimension and the node label, the paragraph information corresponding to the dimension is bound to the node label.

3. The method according to claim 1, characterized in that, The step of locating the paragraph information corresponding to the target node label when a trigger click on the target node label is received includes: When a trigger is received to click on a target node label, the paragraph information bound to the target node label is determined; The page displays the location of the paragraph information within the judgment document and displays the paragraph information using a feature-based display method.

4. The method according to any one of claims 1 to 3, characterized in that, Displaying the paragraph information using a feature-based display method includes: The paragraph information is displayed in a highlighted manner; and / or, The paragraph information is displayed using a combination of enlarged text and an additional pop-up window.

5. A civil supervision device for procuratorates based on text positioning technology, characterized in that, The device includes: The unit of determination is used to identify the type and level of the judgment document. The acquisition unit is used to obtain the target dimension tree corresponding to the judgment document by finding a preset mapping relationship between the combined information obtained by the type and the level of review and the dimension tree. The target dimension tree is composed of a root node with a tree structure and leaf nodes below it. The root node and the leaf nodes are pre-bound with corresponding node labels. The node label bound to the root node is case information, and the node labels bound to the leaf nodes under the root node are at least: case number, judgment date, trial process, and members of the trial organization. The parsing unit is used to parse the judgment document using a preset document parsing model to obtain the dimensions contained in the judgment document and the paragraph information corresponding to the dimensions; The parsing unit includes: a processing module, used to process the document text corresponding to the judgment document using the underlying layer of a preset document parsing model to obtain the word vector representations contained therein; the processing module is also used to process the document text after word vector representation processing using the sentence representation layer of the model to obtain the sentence vector representation corresponding to the document text; the processing module is also used to input the sentence vectors after sentence vector representation into a multi-layer bidirectional recurrent neural network model for processing to obtain the segment vector representation corresponding to the document text; and an output module, used to input the segment vectors obtained by the processing module into a softmax layer and a CRF layer for sequence label prediction, and output the dimension matched by each paragraph in the document text. The binding unit is used to bind the paragraph information corresponding to the dimension parsed by the parsing unit to the node label contained in the target dimension tree obtained by the acquisition unit. The rule to be followed for the binding operation is: if the dimension corresponding to the paragraph information has the same meaning as the node label, then the paragraph information and the node label are bound together. The positioning unit is used to locate the paragraph information corresponding to the target node label when a trigger click on the target node label is received.

6. The apparatus according to claim 5, characterized in that, The binding unit includes: The acquisition module is used to acquire the dimensions contained in the judgment document and the paragraph information corresponding to the dimensions; The acquisition module is also used to acquire the node labels contained in the target dimension tree; The judgment module is used to determine whether the similarity between the dimensions contained in the judgment document and the node labels contained in the target dimension tree is greater than a preset threshold by comparing each dimension with the node labels contained in the target dimension tree. The matching module is used to perform matching on the dimension and the node label when the judgment module determines that the similarity between the dimension and the node label is greater than a preset threshold. The binding module is used to bind the paragraph information corresponding to the dimension to the node label based on the matching relationship between the dimension and the node label obtained by the matching module.

7. A storage medium, characterized in that, The storage medium includes a stored program, wherein, when the program is executed, it controls the device where the storage medium is located to perform the procuratorate civil supervision method based on text positioning technology as described in any one of claims 1-4.

8. An electronic device, characterized in that, The device includes at least one processor, and at least one memory and bus connected to the processor; The processor and the memory communicate with each other via the bus. The processor is used to invoke program instructions in the memory to execute the procuratorate civil supervision method based on text positioning technology as described in any one of claims 1-4.

Images