Record text processing method, device and computer readable storage medium

Abstract

The present invention discloses a method, device and computer-readable storage medium for processing recorded text. The method comprises: obtaining a target recorded text, extracting text features of the target recorded text, obtaining target text features, and predicting the program legitimacy of the target recorded text based on the target text features to obtain a prediction result. The present invention solves the technical problems in the related art that the recorded text processing work is highly repetitive and heavy due to manual review of recorded texts, and the recorded text processing efficiency and intelligence are low.

Description

Technical Field

[0001] The present invention relates to the field of text processing technology, and in particular to a method and device for processing recorded text, and a computer-readable storage medium. Background Art

[0002] In some application scenarios, the legality of the text needs to be reviewed. For example, in the case review process, the legality of the transcript needs to be reviewed based on the content recorded in the transcript.

[0003] At present, the content recorded in the record text is mainly identified through manual review, and the legality of the record text program is judged based on the content recognition results. Manual review of record texts will lead to high repetitiveness and heavy workload in record text processing, which in turn leads to low efficiency and intelligence in record text processing.

[0004] To address the above-mentioned problems, no effective solution has been proposed yet. Summary of the invention

[0005] The embodiments of the present invention provide a method, device and computer-readable storage medium for processing record texts, so as to at least solve the technical problems in the related art that the record text processing work is highly repetitive and heavy, and the record text processing efficiency and intelligence are low due to manual review of record texts.

[0006] According to one aspect of an embodiment of the present invention, a record text processing method is provided, comprising: obtaining a target record text; performing text feature extraction on the target record text to obtain target text features; and predicting the program legitimacy of the target record text based on the target text features to obtain a prediction result.

[0007] Optionally, the extracting text features of the target record text to obtain target text features includes: dividing the target record text into an input sentence sequence; performing a convolution layer operation on the input sentence sequence to obtain convolution features; performing a pooling operation on the obtained convolution features to obtain pooling features; and performing a fully connected layer operation on the pooling features to obtain the target text features.

[0008] Optionally, based on the target text features, the procedural legality of the target record text is predicted to obtain prediction results, including: based on the target text features, predicting a first probability that the parties are informed of the corresponding rights for the procedures included in the target record text; based on the target text features, predicting a second probability that the procedures included in the target record text are checked and confirmed by the parties; based on the target text features, predicting a third probability that there is a legality description of evidence sampling for the procedures included in the target record text; wherein the prediction results include at least one of the following probabilities: the first probability, the second probability, and the third probability.

[0009] Optionally, the program legitimacy of the target record text is predicted based on the target text features to obtain a prediction result, including: extracting a first program operation time for a target object in the target record text based on the target text features; obtaining a second program operation time of the target object in other record texts; and obtaining a prediction result of whether the program operation times in the target record text overlap based on a comparison of the first program operation time and the second program operation time.

[0010] Optionally, based on the target text features, extracting the first program operation time for the target object in the target record text includes: inputting the target text features into an entity extraction model to obtain a text-based first program operation time in the target record text; and standardizing the first program operation time to obtain a standardized first program operation time.

[0011] Optionally, the target record text includes at least one of the following: interrogation record, inquiry record.

[0012] According to another aspect of an embodiment of the present invention, a record text processing method is provided, comprising: displaying an input control on an interactive interface; receiving a target record text in response to an input operation on the input control; and displaying a prediction result obtained by predicting the program legitimacy of the target record text on the interactive interface, wherein the prediction result is predicted based on a target text feature of the target record text, and the target text feature is obtained by extracting a text feature of the target record text.

[0013] According to another aspect of an embodiment of the present invention, a record text processing device is provided, comprising: an acquisition module for acquiring a target record text; an extraction module for performing text feature extraction on the target record text to obtain target text features; and a prediction module for predicting the program legitimacy of the target record text based on the target text features to obtain a prediction result.

[0014] According to another aspect of an embodiment of the present invention, a record text processing device is provided, comprising: a first display module, used to display an input control on an interactive interface; a receiving module, used to receive a target record text in response to an input operation on the input control; and a second display module, used to display a prediction result obtained by predicting the program legitimacy of the target record text on the interactive interface, wherein the prediction result is predicted based on a target text feature of the target record text, and the target text feature is obtained by extracting a text feature of the target record text.

[0015] According to another aspect of an embodiment of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium includes a stored program, wherein when the program is executed, the device where the computer-readable storage medium is located is controlled to execute any one of the above-mentioned record text processing methods.

[0016] According to another aspect of an embodiment of the present invention, a computer device is provided, comprising: a memory and a processor, wherein the memory stores a computer program; the processor is used to execute the computer program stored in the memory, and when the computer program is run, the processor executes any one of the record text processing methods described above.

[0017] In the embodiment of the present invention, a method is adopted to extract text features of a target record text to obtain target text features, and based on the target text features, the program legitimacy of the target record text is predicted to obtain the prediction result of the program legitimacy of the target record text. The program legitimacy of the target record text is predicted based on the target text features of the target record text, and the prediction result of the program legitimacy of the target record text is obtained, thereby achieving the purpose of automatically processing the record text, reducing the repetitiveness of the record text processing work and the workload of the record text processing work, improving the efficiency and intelligence of the record text processing, and solving the technical problems in the related art that the record text processing work is highly repetitive and heavy, and the record text processing efficiency and intelligence are low due to manual review of the record text. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of this application. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0019] Figure 1 is a hardware structure block diagram of a computer terminal for implementing a method for processing recorded text according to an embodiment of the present invention;

[0020] Figure 2is a flow chart of an optional method for processing recorded text according to an embodiment of the present invention;

[0021] Figure 3 is a flow chart of an optional method for obtaining target text features according to an embodiment of the present invention;

[0022] Figure 4 is a flow chart of an optional method for predicting the legitimacy of a target recorded text program based on target text features according to an embodiment of the present invention;

[0023] Figure 5 is a flowchart of another optional method for processing recorded text according to an embodiment of the present invention;

[0024] Figure 6 is a flow chart of an optional transcript processing method according to an embodiment of the present invention;

[0025] Figure 7 is a schematic diagram of an optional model architecture for classifying target transcripts according to an embodiment of the present invention;

[0026] Figure 8 is a structural block diagram of an optional record text processing device according to an embodiment of the present invention;

[0027] Fig. 9 is a structural block diagram of another optional recording text processing device according to an embodiment of the present invention;

[0028] Fig.10 is a structural block diagram of an optional computer device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0029] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0030] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0031] First, some nouns or terms that appear in the description of the embodiments of the present application are subject to the following explanations:

[0032] Record text: A document used to record text content, including document content with a standard format or document content in a non-standard format. For example, document content with a standard format includes professional legal documents, etc.

[0033] Interrogation record: An interrogation record is a document made by case examiners when interrogating a criminal suspect. It records the confession or defense of the criminal suspect, or records the interrogation of the criminal suspect.

[0034] Interrogation record: The interrogation record is an investigation record made against the victim and witness, and is an evidentiary document that records the interrogation of the victim and witness.

[0035] Optical Character Recognition (OCR): A function used to convert non-editable text content into editable text. For example, it can recognize the text content on pictures and photos and directly convert it into editable text.

[0036] Text classification model: Text classification models are widely used. For example, they can be used to identify spam or malicious comments, classify the tendency of articles, classify positive and negative comments on products, etc. There are many methods for text classification, including traditional machine learning logistic regression, SVM (Support Vector Machines), Bayesian classification model, topic model, deep learning FastText (fast text classifier), classification models based on CNN (Convolutional Neural Network) or RNN (Recurrent Neural Network), BERT (Bidirectional Encoder Representation from Transformers), ELMo (Embedding from Language Models), GPT (Gererate Pre-Training Model), etc.

[0037] Pre-trained Language Model: A type of text classification model. In recent years, pre-trained language models such as BERT have been pre-trained based on a large amount of corpus through the language model loss function, and have achieved good results in a series of natural language processing (NLP) tasks.

[0038] Pre-trained language representation model (Bidirectional Encoder Representation from Transformers, BERT): BERT is a pre-trained language model, which belongs to a type of text classification model. The input of BERT is a linear sequence, which supports single sentence text and sentence pair text. The beginning of a sentence is represented by the symbol [CLS], and the end of a sentence is represented by the symbol [SEP]. If it is a sentence pair, the symbol [SEP] is added between sentences. The input features are composed of three components: Token vector, Segment vector and Position vector, which represent word information, sentence information and position information respectively. BERT uses two strategies, MLM (Masked Language Model) and NSP (Next Sentence Prediction) for model pre-training. Among them, when using the MLM model for model pre-training, 15% of the words in the input word sequence are randomly masked, and then the masked words are predicted. Compared with the traditional standard conditional language model that can only predict the target function in the left-to-right or right-to-left direction, MLM can predict the masked words from any direction. Many important downstream tasks such as QA (Question Answering) and NLI (Natural Language Inference) require language models to understand the relationship between two sentences. However, traditional language models do not consider the learning of sentence pairs during training. The NSP model is used to predict the next sentence and adds the task of predicting the relationship between sentences A and B. When making predictions, 50% of the time B is the next sentence of A and is classified as IsNext. The other 50% of the time, B is a randomly selected sentence and is not the next sentence of A and is classified as NotNext.

[0039] Text classification model (Text-Convolut ional Neural Network, TextCNN): is a model that uses convolutional neural network CNN to handle text classification problems. Among them, CNN performs feature detection through convolution operation to obtain multiple feature maps, and then screens the features and filters the noise through pooling operation to extract key information for classification.

[0040] Named Entity Recognition (NER): NER is a type of NLP sequence labeling task, which is used to identify entities with specific meanings or strong referentiality from the input text. It is the basis of machine translation, knowledge graph, relationship extraction and question-answering systems. Academically, the named entities of NER are divided into 3 categories and 7 subcategories. The 3 categories refer to entity category, time category and number category, and the 7 subcategories refer to names of people, places, names of organizations, time, date, currency and percentage. Because language has grammar and corpus follows a certain grammatical structure, probabilistic graph models such as CRF (Conditional Random Field Algorithm), HMM (Hidden Markov Model) and MEMM (Maximum Entropy Markov Model) can be used to analyze label transfer probabilities. In some deep learning models, a CRF layer is generally added to be responsible for sentence-level label prediction.

[0041] Example 1

[0042] According to an embodiment of the present invention, a method embodiment of a method for recording text processing is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0043] The method embodiment provided in Embodiment 1 of the present application can be executed in a mobile terminal, a computer terminal or a similar computing device. Figure 1 FIG. 1 shows a hardware structure block diagram of a computer terminal for implementing a method for processing recorded text. Figure 1 As shown, the computer terminal 10 may include one or more processors (102a, 102b, ..., 102n are used to illustrate the figure, and the processor may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA, etc.), a memory 104 for storing data, and a transmission device for communication functions. In addition, it may also include: a display, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports of the BUS bus), a network interface, a power supply and / or a camera. It can be understood by those skilled in the art that Figure 1 The structure shown is only for illustration and does not limit the structure of the above electronic device. Figure 1 More or fewer components as shown, or with Figure 1 Different configurations are shown.

[0044] It should be noted that the one or more processors and / or other data processing circuits described above may generally be referred to herein as "data processing circuits". The data processing circuits may be embodied in whole or in part as software, hardware, firmware, or any other combination thereof. In addition, the data processing circuit may be a single independent processing module, or may be incorporated in whole or in part into any of the other components in the computer terminal 10. As described in the embodiments of the present application, the data processing circuit acts as a processor control (e.g., selection of a variable resistor terminal path connected to an interface).

[0045] The memory 104 can be used to store software programs and modules of application software, such as the program instructions / data storage device corresponding to the recorded text processing method in the embodiment of the present invention. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory 104, that is, realizing the vulnerability detection method of the above-mentioned application program. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 104 may further include a memory remotely arranged relative to the processor, and these remote memories may be connected to the computer terminal 10 via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0046] The transmission device is used to receive or send data via a network. The specific example of the above network may include a wireless network provided by a communication provider of the computer terminal 10. In one example, the transmission device includes a network adapter (Network Interface Controller, NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device can be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0047] The display may be, for example, a touch screen liquid crystal display (LCD) that enables a user to interact with a user interface of the computer terminal 10 (or mobile device).

[0048] It should be noted that, in some optional embodiments, the above Figure 1 The computer device (or mobile device) shown may include hardware elements (including circuits), software elements (including computer code stored on a computer-readable medium), or a combination of hardware elements and software elements. It should be noted that Figure 1This is merely one example of a particular embodiment and is intended to illustrate the types of components that may be present in the above-described computer device (or mobile device).

[0049] Under the above operating environment, this application provides Figure 2 The recorded text processing method shown. Figure 2 FIG. 1 is a flow chart of a method for processing recorded text according to Embodiment 1 of the present invention. Figure 2 As shown, the record text processing method provided in the embodiment of the present application can be implemented by the following steps:

[0050] Step S202, obtaining the target record text.

[0051] There may be many types of target record texts. As an optional embodiment, the target record texts may be some standard legal documents or some non-standard casual notes. For example, standard legal documents may be document contents with some standard formats. For example, the target record text may be an interrogation record made by the case examiner based on the interrogation of the parties, or may be an interrogation record made by the case examiner based on the interrogation of the victims or witnesses.

[0052] The target record text may be in various forms. As an optional embodiment, the target record text may be a paper document or an electronic document.

[0053] Step S204: extract text features from the target record text to obtain target text features.

[0054] As an optional embodiment, an OCR text extraction method may be used to extract text from the target text, and target text features may be acquired based on the text content of the extracted target text.

[0055] Step S206, predicting the program legitimacy of the target record text based on the target text features to obtain a prediction result.

[0056] As an optional embodiment, a text classification model may be used to predict the program legitimacy of the target record text.

[0057] There are various models for text classification. As an optional embodiment, the TextCNN model and the BERT model are used to predict the program legitimacy of the target record text.

[0058] Through the solution provided by the above-mentioned embodiments of the present application, the program legitimacy of the target record text is predicted based on the target text features of the target record text, and the prediction result of the program legitimacy of the target record text is obtained, thereby achieving the purpose of automatically processing the record text, reducing the repetitiveness and workload of the record text processing work, and improving the efficiency and intelligence of the record text processing, thereby solving the technical problems in the related technology that the record text processing work is highly repetitive and heavy, and the record text processing efficiency and intelligence are low due to manual review of the record text.

[0059] When predicting the program legitimacy of a target record text, it is necessary to obtain target text features of the target record text. There are multiple methods for obtaining target text features.

[0060] Figure 3 FIG. 1 is a flow chart of an optional method for obtaining target text features according to an embodiment of the present invention. Figure 3 As shown, the method for obtaining target text features provided in the embodiment of the present application can be implemented by the following steps:

[0061] Step S302: Divide the target record text into input sentence sequences.

[0062] Step S304, performing a convolution layer operation on the input sentence sequence to obtain convolution features.

[0063] Step S306: performing a pooling operation on the obtained convolutional features to obtain pooled features.

[0064] Step S308, performing a fully connected layer operation on the pooled features to obtain target text features.

[0065] As an optional embodiment, a text classification model TextCNN is used to obtain target text features. The input layer of the TextCNN model needs to input a fixed-length text sequence, so the target record text needs to be divided to obtain a fixed-length input sentence sequence as the input of the TextCNN model. After obtaining the input sentence sequence, the convolution layer and pooling layer in the TextCNN model use the local word order information to extract primary target text features from the sentence sequence of the given target record text, and combine the primary target text features to obtain advanced target text features (pooling layer features). The fully connected layer in the TextCNN model is equivalent to a classifier, which is used to input the pooling layer features extracted by the convolution layer and the pooling layer into the classifier for classification, and finally obtain the target text features.

[0066] In this optional embodiment, the target text is subjected to feature extraction through the operations of the convolution layer and the pooling layer, thereby eliminating the feature engineering step in traditional machine learning, improving the efficiency of target text feature extraction, and making the target text feature extraction process efficient and intelligent, thereby achieving the purpose of automatically and quickly acquiring the target text features.

[0067] As an optional embodiment, based on the target text characteristics, the procedural legality of the target record text is predicted to obtain a prediction result, including at least one of the following: based on the target text characteristics, predicting a first probability that the parties are informed of the corresponding rights for the procedures included in the target record text; based on the target text characteristics, predicting a second probability that the procedures included in the target record text are verified and confirmed by the parties; based on the target text characteristics, predicting a third probability that there is a description of the legality of evidence sampling for the procedures included in the target record text.

[0068] It should be understood that the procedural content recorded in the target record text should include the procedure for the parties to be informed of their corresponding rights. After obtaining the first probability that the parties are informed of their corresponding rights, the procedural legality of the target record text can be judged based on the first probability. When the first probability is less than the first probability threshold, it indicates that the parties have not been informed of their litigation rights, so it can be determined that the procedure of the target record text is illegal, and an error warning can be issued for the illegal result of the procedure. When the first probability is greater than or equal to the first probability threshold, it indicates that the parties are informed of their litigation rights, so it can be determined that the process is correct, that is, the procedure for informing the parties of their litigation rights is legal.

[0069] It should be understood that after the target record text is produced, the parties are required to check and confirm the target record text. After obtaining the second probability that the target record text has been checked and confirmed by the parties, the procedural legality of the target record text can be judged based on the second probability. When the second probability is less than the second probability threshold, it indicates that after the target record text is produced, the parties did not check and confirm the target record text, so it can be determined that the program of the target record text is illegal, and an error warning can be issued for the illegal result of the program. When the second probability is greater than or equal to the second probability threshold, it indicates that after the target record text is produced, the parties checked and confirmed the target record text, so it can be determined that the procedure for the parties to check and confirm the target record text is legal.

[0070] It should be understood that there should be a description of the legality of evidence sampling in the target record text. After obtaining the third probability that there is a description of the legality of evidence sampling in the target record text, the legality of the program of the target record text can be determined based on the third probability. When the third probability is less than the third probability threshold, it indicates that there is no description of the legality of evidence sampling in the target record text, so it can be determined that the program of the target record text is illegal, and an error warning can be issued for the illegal result of the program. When the third probability is greater than or equal to the third probability threshold, it indicates that the legality of evidence sampling is explained in the target record text, so it can be determined that the program with the description of the legality of evidence sampling in the target record text is legal.

[0071] When the prediction result includes one of the first probability, the second probability or the third probability, if the probability is less than the corresponding probability threshold, it can be determined that the program of the target record text is illegal; if the probability is greater than or equal to the corresponding probability threshold, it can be determined that the program of the target record text is legal.

[0072] When the prediction results include multiple of the first probability, the second probability or the third probability, when any one of the probabilities is less than the corresponding probability threshold, it can be determined that the program of the target record text is illegal; when each probability is greater than or equal to the corresponding probability threshold, it can be determined that the program of the target record text is legal.

[0073] In this optional embodiment, based on the target text features, the program legitimacy of the target record text is predicted to obtain at least one of the first probability, the second probability and the third probability, and the program legitimacy of the target record text is predicted based on the obtained probability, thereby realizing automatic processing of the record text, reducing the duplication of the record text processing work and the workload of the record text processing work, and improving the efficiency and intelligence of the record text processing.

[0074] There are many methods for predicting the procedural legitimacy of the target record text based on the target text features. Optionally, based on the fact that the legitimacy elements in the record text are relatively concentrated, the TextCNN model is used to classify the target text to obtain the prediction result.

[0075] The TextCNN model is divided into four layers: input layer, convolution layer, pooling layer and fully connected softmax output layer. The detailed process of using the TextCNN model for text classification is as follows: For a sentence with an input length of n and a word vector dimension of d, there is a one-dimensional convolution layer with filter size filter_size = (2,3,4). Each filter has 2 channels, which is equivalent to extracting two 2-gram, 3-gram and 4-gram features respectively. Through the convolution operation, the feature map c is obtained. Then, the maximum value in the feature map c is extracted through the maximum pooling operation, and different filters obtain different n-gram features. Finally, the classification result is obtained through the softmax function. Then, through three independent fully connected layers, the legal probabilities of the three programs can be obtained respectively, that is, through three independent fully connected layers, the first probability, the second probability and the third probability can be obtained respectively, and then the program legality prediction of the target record text is performed based on the obtained probabilities.

[0076] Figure 4 FIG. 1 is a flow chart of an optional method for predicting the legitimacy of a target recorded text program based on target text features according to an embodiment of the present invention. Figure 4 As shown, based on the target text features, the program legitimacy of the target record text is predicted to obtain a prediction result, which may include the following steps:

[0077] Step S402, extracting the first program operation time of the target object in the target record text based on the target text feature;

[0078] Step S404, obtaining the second program operation time of the target object in other record texts;

[0079] Step S406: Based on the comparison between the first program operation time and the second program operation time, a prediction result of whether the program operation time in the target record text overlaps is obtained.

[0080] In this optional embodiment, the target record text may include multiple record texts. Based on the target text features, the first program operation time for the target object in the target record text can be extracted, which can be the program operation time in multiple record texts. The second program operation time of the target object in other record texts can also be obtained in multiple copies. When comparing the first program operation time and the second program operation time, if the program operation time in the target record text overlaps, it means that the program of the target record text is illegal, and an error warning can be issued for the illegal program result. Otherwise, it means that the program of the target record text is legal, and the process of the target record text is correct.

[0081] In this optional embodiment, the first program operation time and the second program operation time of the target object in the target record text are compared to obtain a prediction result of whether the program operation times in the target record text overlap, and a determination is made as to whether the target record text program is legal based on the prediction result, thereby realizing automatic processing of the record text, reducing the repetitiveness and workload of the record text processing work, and improving the efficiency and intelligence of the record text processing.

[0082] As an optional embodiment, based on the target text feature, extracting the first hearing time of the target object in the target record text may include the following steps:

[0083] Inputting the target text features into the entity extraction model to obtain the first program operation time based on the text in the target record text;

[0084] The first program operation time is standardized to obtain a standardized first program operation time.

[0085] There are many methods for extracting the first program operation time on the target object in the target record text based on the target text features. For example, a text entity extraction model can be used to extract the first program operation time on the target object in the target record text.

[0086] As an optional embodiment, a named entity recognition (NER) method is used to extract the program operation time of the target object in the target record text. Specifically, a pre-trained model BERT is used as an entity extraction model to extract the program operation time in the record text, wherein the pre-trained model BERT based on a deep bidirectional Transformer uses a Transformer encoder part as the feature extractor, and the pre-trained model BERT is divided into two stages: pre-training and downstream task fine-tuning.

[0087] In this optional embodiment, the target record text is processed based on the entity extraction model, and the program operation time in the target record text can be automatically obtained, thereby realizing automatic processing of the record text, reducing the repetitiveness and workload of the record text processing work, and improving the efficiency and intelligence of the record text processing.

[0088] In one embodiment, the pre-trained model BERT is used to extract the program operation time of the target text, and the obtained program operation time information is "1:00 p.m. on September 1, 21". The time information is standardized to obtain the standardized program operation time of "2021-09-01 13:00". The obtained program operation time information is standardized to facilitate cross-comparison of the program operation time in each recorded text, thereby improving the efficiency of overlapping prediction of the program operation time in each recorded text and improving the efficiency of recorded text processing.

[0089] According to an embodiment of the present invention, another embodiment of a recorded text processing method is also provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0090] It should be noted that the method embodiment provided in Example 2 of the present application can be used in Figure 1 The process is executed in the computer terminal 10 shown or a similar computing device.

[0091] Figure 5 is a flowchart of another optional method for processing recorded text according to an embodiment of the present invention, such as Figure 5 As shown, the record text processing method provided in the embodiment of the present application can be implemented by the following method steps:

[0092] Step S502, displaying input controls on the interactive interface;

[0093] Step S504, receiving target record text in response to an input operation on the input control;

[0094] Step S506, displaying a prediction result of the program legality of the target record text on the interactive interface, wherein the prediction result is predicted based on a target text feature of the target record text, and the target text feature is obtained by extracting text features from the target record text.

[0095] Through the scheme provided by the above-mentioned embodiment of the present application, the target record text is received through the input operation of the display control, and the prediction result obtained by predicting the program legitimacy of the target record text is displayed on the interactive interface, wherein the prediction result is obtained by predicting the program legitimacy of the target record text based on the target text features of the target record text. The scheme achieves the purpose of automatically processing the record text, reduces the repetitiveness and workload of the record text processing work, improves the efficiency and intelligence of the record text processing, and thus solves the technical problems in the related technology that the record text processing work is highly repetitive and heavy, and the record text processing efficiency and intelligence are low due to manual review of the record text. In addition, by displaying the process of intelligently processing the record text on the interactive interface, the process and results of the record text processing can be intuitively displayed, thereby effectively improving the user experience.

[0096] Based on the above embodiments and optional embodiments, an optional implementation is provided, which is described in detail below.

[0097] In this optional implementation mode, the target record text is taken as an example of criminal interrogation record, and the procedural legality of the interrogation record is reviewed based on whether the parties were informed of their litigation rights in the first interrogation, whether the parties were asked to check and confirm the interrogation record after it was made, whether there is a legality statement on the evidence sampling in the interrogation record, and whether there is overlap in the hearing times of multiple interrogation records obtained from multiple interrogations of the same target object.

[0098] In the relevant technology, the procedural legality of interrogation records is reviewed through manual review, which results in heavy and repetitive work for grassroots case reviewers, and a low level of intelligence in the record review work, which affects the quality and efficiency of case review by case reviewers.

[0099] In view of this, in the embodiments of the present disclosure, a method for processing interrogation records is provided to realize intelligent processing of interrogation records, technically reduce the repetitive work of record processing, improve the efficiency of record processing and the degree of intelligent processing, complete the automated record form review, thereby improving the efficiency of reviewing evidence, alleviating the contradiction of "too many cases and too few people", reducing legal deviations, improving the fairness of case reviewers, and improving the quality and efficiency of case review by case reviewers.

[0100] Figure 6 FIG. 1 is a flow chart of another optional method for recording and processing according to an embodiment of the present invention. Figure 6 As shown, the transcript processing method provided in the embodiment of the present application can be implemented by the following steps:

[0101] Get the original transcript.

[0102] The OCR text extraction method is used to extract text from the original transcript document to obtain the target transcript (equivalent to the target record text in the aforementioned embodiment).

[0103] The TextCNN model is used as a text classification model to classify the target transcript, and the probability of whether the party is informed of the litigation rights during the first interrogation (equivalent to the first probability in the aforementioned embodiment), the probability of whether the party is asked to check and confirm the transcript after the interrogation transcript is made (equivalent to the second probability in the aforementioned embodiment), and the probability of whether there is a description of the legality of evidence sampling in the interrogation transcript (equivalent to the third probability in the aforementioned embodiment) is obtained, and the corresponding prediction results are obtained according to the probability. If the party is informed of the litigation rights during the first interrogation, the party is asked to check and confirm the transcript after the interrogation transcript is made, and there is a description of the legality of evidence sampling in the interrogation transcript, it means that the target transcript procedure is legal and the process of the target transcript is correct; otherwise, it means that the target transcript procedure is illegal, and an error warning is issued for illegal results.

[0104] Figure 7 is a schematic diagram of an optional model architecture for classifying target transcripts according to an embodiment of the present invention, referring to Figure 7 As shown in the figure, when the target transcript is classified by the text classification model, the target transcript is divided into an input sentence sequence, and then the input sentence sequence is subjected to a convolution layer operation to obtain a convolution feature. The obtained convolution feature is then subjected to a pooling operation to obtain a pooling feature. The pooling feature is subjected to a fully connected layer operation to obtain the target text feature. Based on the target text feature and three independent fully connected layers, the prediction results of whether the parties were informed of their litigation rights in the first interrogation, whether the parties were asked to check and confirm the interrogation transcript after it was made, and whether there was a description of the legality of evidence sampling in the interrogation transcript are obtained, and the corresponding procedural legality prediction is performed based on the prediction results.

[0105] Use the BERT model as a text entity extractor to extract the time of each interrogation record in the target record, obtain the hearing time in each interrogation record (i.e., the program operation time in the aforementioned embodiment), and standardize each hearing time to obtain a standardized hearing time. Then compare each standardized hearing time to determine whether the hearing times overlap. If so, the procedure of the record is determined to be illegal, and an error warning is issued for the illegal result. Otherwise, the procedure of the record is determined to be legal, that is, the process of the record is correct.

[0106] It should be noted that, for the above-mentioned method embodiments, for the sake of simplicity, they are all described as a series of action combinations, but those skilled in the art should know that the present invention is not limited by the described action sequence, because according to the present invention, certain steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present invention.

[0107] Through the description of the above implementation methods, those skilled in the art can clearly understand that the recorded text processing method according to the above embodiment can be implemented by means of software plus a necessary general hardware platform, and of course it can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a computer-readable storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for enabling a terminal device (which can be a mobile phone, computer, server, or network device, etc.) to execute the methods of various embodiments of the present invention.

[0108] Example 2

[0109] According to an embodiment of the present invention, a record text processing device for implementing the above record text processing method is also provided. Figure 8 is a structural block diagram of an optional recording text processing device according to an embodiment of the present invention, such as Figure 8 As shown, the device comprises:

[0110] The acquisition module 802 is used to acquire the target record text; the extraction module 804 is connected to the above acquisition module 802 and is used to extract text features of the target record text to obtain target text features; the prediction module 806 is connected to the above extraction module 804 and is used to predict the program legitimacy of the target record text based on the target text features to obtain a prediction result.

[0111] It should be noted that the acquisition module 802, extraction module 804 and prediction module 806 correspond to steps S202 to S206 in Example 1, and the three modules and the corresponding steps implement the same examples and application scenarios, but are not limited to the contents disclosed in the above-mentioned Example 1. It should be noted that the above-mentioned modules, as part of the device, can be run in the computer terminal 10 provided in Example 1.

[0112] According to an embodiment of the present invention, another record text processing device for implementing the above record text processing method is also provided. Fig. 9is a structural block diagram of another optional recording text processing device according to an embodiment of the present invention, such as Fig. 9 As shown, the device includes: a first display module 902, used to display input controls on an interactive interface; a receiving module 904, connected to the above-mentioned first display module 902, used to receive target record text in response to an input operation on the input control; a second display module 906, connected to the above-mentioned receiving module 904, used to display a prediction result obtained by predicting the program legitimacy of the target record text on the interactive interface, wherein the prediction result is predicted based on a target text feature of the target record text, and the target text feature is obtained by extracting text features from the target record text.

[0113] It should be noted that the first display module 902, the receiving module 904 and the second display module 906 correspond to steps S502 to S506 in Example 2, and the three modules and the corresponding steps implement the same examples and application scenarios, but are not limited to the contents disclosed in the above-mentioned Example 2. It should be noted that the above-mentioned modules as part of the device can be run in the computer terminal 10 provided in Example 1.

[0114] The record text processing device provided by this optional embodiment can be used to implement the above-mentioned record text processing method, receive the target record text through the input operation of the display control, and display the prediction result obtained by predicting the program legitimacy of the target record text on the interactive interface, wherein the prediction result is obtained by predicting the program legitimacy of the target record text based on the target text features of the target record text. The scheme achieves the purpose of automatically processing the record text, reduces the repetitiveness and workload of the record text processing work, improves the efficiency and intelligence of the record text processing, and thus solves the technical problems in the related technology that the record text processing work is highly repetitive and heavy, and the record text processing efficiency and intelligence are low due to manual review of the record text.

[0115] Example 3

[0116] The embodiment of the present invention further provides a computer-readable storage medium. Optionally, in this embodiment, the computer-readable storage medium can be used to store the program code executed by the record text processing method provided in the above embodiment 1.

[0117] Optionally, in this embodiment, the computer-readable storage medium may be located in any one of the computer terminals in a computer terminal group in a computer network, or in any one of the mobile terminals in a mobile terminal group.

[0118] Optionally, in this embodiment, the computer-readable storage medium is configured to store program code for executing the following steps: obtaining a target record text; performing text feature extraction on the target record text to obtain target text features; and predicting the program legitimacy of the target record text based on the target text features to obtain a prediction result.

[0119] Optionally, in this embodiment, the computer-readable storage medium is configured to store program code for executing the following steps: performing text feature extraction on the target record text to obtain target text features, including: dividing the target record text into an input sentence sequence; performing a convolution layer operation on the input sentence sequence to obtain convolution features; performing a pooling operation on the obtained convolution features to obtain pooling features; performing a fully connected layer operation on the pooling features to obtain target text features.

[0120] Optionally, in this embodiment, the computer-readable storage medium is configured to store program code for executing the following steps: based on the target text features, predicting the procedural legality of the target record text to obtain prediction results, including: when the target record text includes an interrogation transcript, based on the target text features, predicting a first probability that the parties are informed of the corresponding rights for the procedures included in the target record text; based on the target text features, predicting a second probability that the procedures included in the target record text are verified and confirmed by the parties; based on the target text features, predicting a third probability that there is a description of the legality of evidence sampling for the procedures included in the target record text; wherein the prediction results include at least one of the following probabilities: a first probability, a second probability, and a third probability.

[0121] Optionally, in this embodiment, the computer-readable storage medium is configured to store program code for executing the following steps: based on the target text features, predicting the program legality of the target record text to obtain a prediction result, including: based on the target text features, extracting the first review time of the target object in the target record text; obtaining the second program operation time of the target object in other record texts; based on the comparison of the first program operation time and the second program operation time, obtaining a prediction result of whether the program operation time in the target record text overlaps.

[0122] Optionally, in this embodiment, the computer-readable storage medium is configured to store program code for executing the following steps: based on the target text features, extracting the first program operation time for the target object in the target record text, including: inputting the target text features into the entity extraction model to obtain the text-based first program operation time in the target record text; standardizing the first program operation time to obtain a standardized first program operation time.

[0123] Optionally, in this embodiment, the computer-readable storage medium is configured to store program code for executing the following steps: obtaining a target record text including at least one of the following: an interrogation transcript, an inquiry transcript.

[0124] Optionally, in this embodiment, the computer-readable storage medium is configured to store program code for executing the following steps: displaying an input control on an interactive interface; receiving a target record text in response to an input operation on the input control; and displaying a prediction result obtained by predicting the program legitimacy of the target record text on the interactive interface, wherein the prediction result is predicted based on a target text feature of the target record text, and the target text feature is obtained by extracting text features from the target record text.

[0125] The embodiment of the present invention may provide a computer device, which may be any computer terminal device in a computer terminal group. Optionally, in this embodiment, the computer device may also be replaced by a terminal device such as a mobile terminal.

[0126] Optionally, in this embodiment, the computer terminal may be located in at least one network device among a plurality of network devices of the computer network.

[0127] In this embodiment, the above-mentioned computer device can execute the program code of the following steps in the record text processing method of the application: obtaining the target record text; performing text feature extraction on the target record text to obtain the target text feature; based on the target text feature, predicting the program legitimacy of the target record text to obtain a prediction result.

[0128] Fig.10 A block diagram of an optional computer device according to an embodiment of the present invention. Fig.10 As shown, the computer device may include: one or more (only one is shown in the figure) processors 1002, a memory 1004, etc.

[0129] The memory 1004 may be used to store software programs and modules, such as program instructions / modules corresponding to the recorded text processing method and device in the embodiment of the present invention. The processor 1002 executes various functional applications and data processing by running the software programs and modules stored in the memory 1004, that is, the recorded text processing method described above is realized. The memory 1004 may include a high-speed random access memory 1004, and may also include a non-volatile memory 1004, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory 1004. In some examples, the memory 1004 may further include a memory 1004 remotely arranged relative to the processor 1002, and these remote memories 1004 may be connected to the computer terminal via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0130] The processor 1002 can call the information and application programs stored in the memory 1004 through the transmission device to execute the following steps: obtain the target record text; extract text features of the target record text to obtain target text features; based on the target text features, predict the program legitimacy of the target record text to obtain a prediction result.

[0131] Optionally, the processor 1002 may also execute the following program code steps: extracting text features from the target record text to obtain target text features, including: dividing the target record text into an input sentence sequence; performing a convolution layer operation on the input sentence sequence to obtain convolution features; performing a pooling operation on the obtained convolution features to obtain pooling features; performing a fully connected layer operation on the pooling features to obtain target text features.

[0132] Optionally, the processor 1002 may also execute program code for the following steps: based on the target text features, predict the program legality of the target record text to obtain prediction results, including: based on the target text features, predicting a first probability that the parties are informed of the corresponding rights for the programs included in the target record text; based on the target text features, predicting a second probability that the programs included in the target record text are verified and confirmed by the parties; based on the target text features, predicting a third probability that there is a description of the legality of evidence sampling for the programs included in the target record text; wherein the prediction results include at least one of the following probabilities: a first probability, a second probability, and a third probability.

[0133] Optionally, the processor 1002 may also execute program code for the following steps: based on the target text features, predict the program legitimacy of the target record text to obtain a prediction result, including: based on the target text features, extracting the first program operation time for the target object in the target record text; obtaining the second program operation time of the target object in other record texts; and based on the comparison between the first program operation time and the second program operation time, obtaining a prediction result of whether the program operation time in the target record text overlaps.

[0134] Optionally, the processor 1002 may also execute program code for the following steps: based on target text features, extracting the first program operation time for the target object in the target record text, including: inputting the target text features into an entity extraction model to obtain a text-based first program operation time in the target record text; and normalizing the first program operation time to obtain a standardized first program operation time.

[0135] Optionally, the processor 1002 may also execute the program code of the following steps: obtaining a target record text including at least one of the following: an interrogation record text, an inquiry record text.

[0136] The processor 1002 can call the information and application programs stored in the memory 1004 through the transmission device to perform the following steps: displaying input controls on the interactive interface; receiving target record text in response to input operations on the input controls; and displaying the prediction results obtained by predicting the program legitimacy of the target record text on the interactive interface, wherein the prediction results are predicted based on target text features of the target record text, and the target text features are obtained by extracting text features from the target record text.

[0137] By adopting the embodiment of the present invention, a scheme of a computer device is provided. Based on the target text features of the target record text, the program legitimacy of the target record text is predicted, and the prediction result of the program legitimacy of the target record text is obtained, thereby achieving the purpose of automatically processing the record text, reducing the repetitiveness of the record text processing work and the workload of the record text processing work, and improving the efficiency and intelligence of the record text processing, thereby solving the technical problems in the related technology that the record text processing work is highly repetitive and heavy, and the record text processing efficiency and intelligence are low due to manual review of the record text. A person of ordinary skill in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing the hardware related to the terminal device through a program, and the program can be stored in a computer-readable storage medium, and the computer-readable storage medium can include: a flash drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0138] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0139] In the above embodiments of the present invention, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0140] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic, for example, the division of units is only a logical function division, and there may be other division methods in actual implementation, for example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0141] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0142] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0143] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a computer-readable storage medium, including a number of instructions for a computer device (which can be a personal computer, a server or a network device, etc.) to perform all or part of the steps of the methods of each embodiment of the present invention. The aforementioned computer-readable storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, magnetic disk or optical disk and other media that can store program codes.

[0144] The above are only preferred embodiments of the present invention. It should be pointed out that, for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for processing recorded text, characterized in that: include: Get the target record text; Performing text feature extraction on the target record text to obtain target text features; Based on the target text features, predict at least one of a first probability that the party is informed of the corresponding rights, a second probability that the party has verified and confirmed, and a third probability that the legality of the evidence sampling is explained; Based on the first probability, the second probability, and at least one of the third probability, the program legitimacy of the target record text is predicted to obtain a prediction result.

2. The method according to claim 1, characterized in that The step of extracting text features from the target record text to obtain target text features includes: Dividing the target record text into an input sentence sequence; Performing a convolution layer operation on the input sentence sequence to obtain convolution features; Performing a pooling operation on the obtained convolutional features to obtain pooled features; A fully connected layer operation is performed on the pooled features to obtain the target text features.

3. The method according to claim 1, characterized in that Based on the target text feature, predicting at least one of a first probability that the party is informed of the corresponding rights, a second probability that the party has verified and confirmed, and a third probability of the legality of the evidence sampling, includes at least one of the following: Based on the target text features, predicting a first probability that the party concerned is informed of the corresponding rights for the procedure included in the target record text; Based on the target text feature, predict a second probability of the program included in the target record text, which has been checked and confirmed by the party concerned; Based on the target text features, a third probability of the existence of a legality statement of the evidence sampling is predicted for the program included in the target record text.

4. The method according to claim 1, characterized in that: The process of predicting the program legitimacy of the target record text based on the target text feature to obtain a prediction result includes: Extracting the first program operation time of the target object in the target record text based on the target text feature; Obtaining a second program operation time of the target object in other record texts; Based on the comparison between the first program operation time and the second program operation time, a prediction result of whether the program operation times in the target record text overlap is obtained.

5. The method according to claim 4, characterized in that The step of extracting the first program operation time of the target object in the target record text based on the target text feature includes: Inputting the target text feature into an entity extraction model to obtain a text-based first program operation time in the target record text; The first program operation time is standardized to obtain a standardized first program operation time.

6. The method according to any one of claims 1 to 5, characterized in that The target record text includes at least one of the following: interrogation record, inquiry record.

7. A method for processing recorded text, characterized in that: include: Display input controls on the interactive interface; In response to an input operation on the input control, receiving a target record text; A prediction result of the procedural legality of the target record text is displayed on the interactive interface, wherein the prediction result is obtained based on at least one of a first probability that the party is informed of the corresponding rights, a second probability verified and confirmed by the party, and a third probability of the legality description of the evidence sampling, and the first probability, the second probability, and at least one of the third probability are predicted based on the target text features of the target record text, and the target text features are obtained by extracting text features from the target record text.

8. A recorded text processing device, characterized in that: include: The acquisition module is used to obtain the target record text; An extraction module, used for extracting text features from the target record text to obtain target text features; A first prediction module is used to predict at least one of a first probability that the party is informed of the corresponding rights, a second probability that the party has verified and confirmed, and a third probability of the legality of the evidence sampling based on the target text features; The second prediction module is used to predict the program legitimacy of the target record text based on the first probability, the second probability, and at least one of the third probability to obtain a prediction result.

9. A recorded text processing device, characterized in that: include: A first display module, used to display input controls on the interactive interface; A receiving module, configured to receive a target record text in response to an input operation on the input control; The second display module is used to display the prediction result of the procedural legality of the target record text on the interactive interface, wherein the prediction result is obtained based on at least one of a first probability that the party is informed of the corresponding rights, a second probability confirmed by the party, and a third probability of the legality description of the evidence sampling, and the first probability, the second probability, and at least one of the third probability are predicted based on the target text features of the target record text, and the target text features are obtained by extracting text features from the target record text.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored program, wherein when the program is executed, the device where the computer-readable storage medium is located is controlled to execute the recorded text processing method according to any one of claims 1 to 7.

11. A computer device, characterized in that: include: Memory and processor, The memory stores a computer program; The processor is used to execute the computer program stored in the memory, and when the computer program is running, the processor executes the record text processing method described in any one of claims 1 to 7.

Images