Intelligent content production method and system based on user needs

Abstract

The embodiment of the present invention discloses a method and system for intelligent content production based on user needs, wherein the method includes: processing case judgment documents through a Seq2Seq model to generate corresponding case summaries; selecting different intelligent content production sub-modules based on the length of the case fact description file input by the user, that is, a first content production sub-module based on a twin neural network model or a second content production sub-module based on a recurrent neural network model RNN, and finally outputting intelligently matched laws and corresponding case summaries and case judgment documents to improve user experience and content generation accuracy.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a method and system for producing intelligent content based on user needs. Background Art

[0002] Artificial intelligence is an important part of the intelligent discipline. It attempts to understand the essence of intelligence and produce a new intelligent machine that can respond in a similar way to human intelligence. Artificial intelligence is a very broad science, including robotics, language recognition, image recognition, natural language processing, expert systems, machine learning, computer vision, etc.

[0003] In the field of legal consultation and retrieval, artificial intelligence has also gradually played an important role. Applying artificial intelligence technology to the legal field helps meet the society's needs for high-quality legal services. However, most of the existing legal content-related systems are retrieval systems, and users generally search based on titles or keywords. This method cannot accurately obtain the applicable laws or similar case judgments when users search for specific cases, which is not convenient for users to accurately search and make relevant predictions. Therefore, building an intelligent content production system based on user legal consultation needs is a technical problem that needs to be solved urgently. Summary of the invention

[0004] In view of the defects in the prior art, the present invention provides an intelligent content production method based on user needs, which can apply natural language processing and machine learning technologies in the field of artificial intelligence to the field of intelligent production of legal auxiliary content, thereby improving user experience and content generation accuracy. To achieve the above purpose, the present invention provides the following technical solutions:

[0005] A method for producing intelligent content based on user needs, characterized in that the method comprises the following steps:

[0006] (1) The case database collects and stores case judgment documents, processes the case judgment documents through the Seq2Seq model, and then generates corresponding case summaries;

[0007] (2) The user input module receives the user's description file of the facts of the legal case and forwards it to the temporary file cache module;

[0008] (3) The temporary file cache module converts the case fact description file into a text file;

[0009] (4) The resource utilization rate judgment module calculates the current system resource utilization rate. When the system resource utilization rate is less than or equal to the first threshold, the process directly proceeds to step (5); when the system resource utilization rate is greater than the first threshold, the control system releases part of the system resources so that the updated system resource utilization rate is lower than the first threshold, and the process proceeds to step (5);

[0010] (5) The file character length judgment module calculates the character length L of the text file, and when L is less than a second threshold, controls the temporary file cache module to forward the text file to the first content production submodule of the content production module; when L is greater than or equal to the second threshold, controls the temporary file cache module to forward the text file to the second content production submodule of the content production module;

[0011] (6) The first content production submodule or the second content production submodule of the content production module outputs intelligently matched legal provisions and corresponding case summaries and case judgment documents.

[0012] Another object of the present invention is to provide an intelligent content production system, characterized in that the system comprises:

[0013] A case database, which is used to collect and store case judgment documents, process the case judgment documents through a Seq2Seq model, and generate corresponding case summaries;

[0014] A user input module, used to receive a user's description file of the facts of a legal case and forward it to a temporary file cache module;

[0015] A temporary file cache module, used for converting the case fact description file into a text file;

[0016] A resource utilization rate judgment module, used to calculate the current system resource utilization rate; when the system resource utilization rate is greater than a first threshold, the control system releases part of the system resources so that the updated system resource utilization rate is lower than the first threshold;

[0017] A file character length judgment module, used for calculating the character length L of the text file, and when L is less than a second threshold, controlling the temporary file cache module to forward the text file to the first content production submodule of the content production module; when L is greater than or equal to the second threshold, controlling the temporary file cache module to forward the text file to the second content production submodule of the content production module;

[0018] The content production module includes a first content production sub-module and a second content production sub-module, and the content production module is used to output intelligently matched legal provisions and corresponding case summaries and case judgment documents through the first content production sub-module or the second content production sub-module.

[0019] Another object of the present invention is to provide an electronic device, comprising: a memory and a processor, wherein the memory and the processor are coupled; the memory stores program instructions, and when the program instructions are executed by the processor, the electronic device executes the above-mentioned intelligent content production method based on user needs.

[0020] Another object of the present invention is to provide a computer-readable storage medium storing one or more programs, wherein the one or more programs enable a computer to execute the above-mentioned intelligent content production method based on user needs.

[0021] In combination with all the above technical solutions, the present invention has the following advantages compared with the prior art:

[0022] (1) The case judgment document file is processed through the Seq2Seq model to generate a corresponding case summary. This is convenient for the subsequent output of the corresponding case summary for users to review. On the other hand, when the input file is a short text, it is also convenient for the first content production submodule to perform data processing, thereby improving the speed of law matching and the efficiency of content generation when short text questions are input.

[0023] (2) A resource utilization judgment module is set up. When the system resource utilization is greater than a first threshold, the control system releases part of the system resources to ensure the smooth progress of subsequent content production tasks, thereby preventing the failure of content production tasks due to insufficient system resources and causing a poor user experience.

[0024] (3) When the input file is a long text, the RNN network structure processing technology is combined to deeply explore the correlation between semantics, further improving the speed of legal provision matching and the efficiency of intelligent content generation.

[0025] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] The above and other purposes, features and advantages of the present invention will become more apparent by describing the embodiments of the present invention in more detail in conjunction with the accompanying drawings. The accompanying drawings are used to provide a further understanding of the embodiments of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0027] Figure 1 It is a flowchart of a method provided by an exemplary embodiment of the present invention.

[0028] Figure 2 It is a schematic diagram of the structure of a system provided by an exemplary embodiment of the present invention. DETAILED DESCRIPTION

[0029] Below, the exemplary embodiments according to the present invention will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments of the present invention, and it should be understood that the present invention is not limited to the exemplary embodiments described here.

[0030] It should be noted that the relative arrangement of components and steps, the numerical expressions and numerical values ​​set forth in these embodiments do not limit the scope of the present invention unless specifically stated otherwise.

[0031] Those skilled in the art can understand that the terms "first" and "second" in the embodiments of the present invention are only used to distinguish different steps, devices or modules, etc., and neither represent any specific technical meaning nor indicate the necessary logical order between them.

[0032] It should also be understood that, in the embodiments of the present invention, “plurality” may refer to two or more than two, and “at least one” may refer to one, two or more than two.

[0033] It should also be understood that any component, data or structure mentioned in the embodiments of the present invention can generally be understood as one or more, unless explicitly limited or otherwise indicated in the context.

[0034] The execution flow chart of the intelligent content production method based on user needs is shown in Figure 1 .like Figure 1 As shown, the present invention provides an intelligent content production method based on user needs, the method comprising the following steps:

[0035] (1) The case database collects and stores case judgment documents, processes the case judgment documents through the Seq2Seq model, and then generates corresponding case summaries;

[0036] (2) The user input module receives the user's description file of the facts of the legal case and forwards it to the temporary file cache module;

[0037] (3) The temporary file cache module converts the case fact description file into a text file;

[0038] (4) The resource utilization rate judgment module calculates the current system resource utilization rate. When the system resource utilization rate is less than or equal to the first threshold, the process directly proceeds to step (5); when the system resource utilization rate is greater than the first threshold, the control system releases part of the system resources so that the updated system resource utilization rate is lower than the first threshold, and the process proceeds to step (5);

[0039] (5) The file character length judgment module calculates the character length L of the text file, and when L is less than a second threshold, controls the temporary file cache module to forward the text file to the first content production submodule of the content production module; when L is greater than or equal to the second threshold, controls the temporary file cache module to forward the text file to the second content production submodule of the content production module;

[0040] (6) The first content production submodule or the second content production submodule of the content production module outputs intelligently matched legal provisions and corresponding case summaries and case judgment documents.

[0041] Preferably, in step (1), the case judgment document file is processed by a Seq2Seq model to generate a corresponding case summary; the processing steps include:

[0042] (1.1) Assumptions and They are the original text of a case judgment document and the corresponding abstract, expressed in the form of word sequence. , , and i>j, the training data is A collection of text summaries in the form of;

[0043] (1.2) The encoder of the Seq2Seq model encodes the source text in the encoding stage Encode and get The context vector , ;

[0044] (1.3) In the decoding phase, the decoder of the Seq2Seq model uses the historical information of the target text sequence and the context vector Decode to predict the output word at the current moment , the decoded output word is contained in the dictionary, through The function calculates the probability of each word in the dictionary and selects the word with the highest probability as the output word.

[0045] The present invention processes the case judgment document file through the Seq2Seq model to generate a corresponding case summary. This, on the one hand, is convenient for the subsequent output of the corresponding case summary for users to review. On the other hand, when the input file is a short text, it is also convenient for the first content production sub-module to perform data processing, thereby improving the speed of law matching and the efficiency of content generation when short text problems are input.

[0046] It should be noted that the case judgment documents can be collected from the China Judgments Online website.

[0047] Preferably, the format of the user's description file of the facts of the legal case in step (2) is one of voice format, PDF format, Word format or image format. The input files in the above multiple formats are processed by the temporary file cache module to generate a unified format, i.e., a text file, for subsequent processing.

[0048] Preferably, the system resource utilization in step (4) is calculated using one of CPU utilization, memory utilization or storage utilization; when the system resource utilization is represented by CPU utilization, the first threshold is set to 80%, when the system resource utilization is represented by memory utilization, the first threshold is set to 70%, and when the system resource utilization is represented by storage utilization, the first threshold is set to 60%.

[0049] The present invention sets up a resource utilization judgment module. When the system resource utilization is greater than a first threshold, the control system releases part of the system resources to ensure the smooth progress of subsequent content production tasks, thereby preventing the failure of content production tasks due to insufficient system resources and causing a poor user experience.

[0050] Preferably, the first content production submodule is implemented based on a twin neural network model, which is composed of two identical and parameter-shared convolutional neural network structures including a convolutional layer, an activation function, a pooling layer, and a loss function; the processing process of the first content production submodule is as follows:

[0051] (4.1.1) When matching similar cases, the two convolutional neural networks will perform text encoding on the input text file and the case summary of the case to be matched and calculate the similarity, and select the case summary of the case to be matched whose similarity meets the threshold according to the similarity ranking;

[0052] (4.1.2) Output the case summaries that meet the above similarity threshold, as well as the corresponding laws and corresponding case judgment documents.

[0053] Preferably, the second content production submodule is implemented based on a recurrent neural network model RNN; the recurrent neural network model RNN is implemented by sequentially connecting an input layer, an embedding layer, a semantic encoding layer, and a content output layer.

[0054] Preferably, the second content production submodule performs the following operation steps during specific implementation:

[0055] (4.2.1) The input layer serializes the received text file to generate a sequence describing the case facts , is the maximum length of the sequence, and the legal article label matched by the model is expressed as , is the number of legal article labels; the semantic matching relationship between the sequence characteristics of the case fact description and the legal article labels is obtained by pre-training the recurrent neural network model RNN, and the training set comes from the collected case judgment documents;

[0056] (4.2.2) The Embedding layer converts the sequence The sequence in is transformed into a matrix with distance meaning and small dimension, that is, the output A two-dimensional vector, Represents the dimension of embedding word embedding;

[0057] (4.2.3) The semantic encoding layer receives the output of the Embedding layer A two-dimensional vector, which is expressed as ; Each word describing the facts of the case is embedded into a dimension The word is used as the input of each unit of the RNN, and the output vector obtained by the last unit is used as the output of the RNN layer;

[0058] At time t, the input of the RNN is the tth word , the value of the hidden layer is , the output is expressed as ,

[0059] Said , ;

[0060] in is the activation function, , , is the weight vector, , are all bias vectors;

[0061] After the entire case fact description passes through the RNN structure, only the output vector of the last cell is retained , as the output of the RNN layer; used in the prediction stage The function obtains the matching results of the legal article prediction;

[0062] (4.2.4) The content output layer outputs the legal provisions corresponding to the matching results, the corresponding case summaries and the case judgment documents.

[0063] Aiming at case consultation situations where the input file is a long text, the present invention combines RNN network structure processing technology to deeply mine the correlation between semantics, further improving the speed of law matching and the efficiency of intelligent content generation.

[0064] The structural diagram of the intelligent content production system based on user needs is shown in Figure 2 .like Figure 2 As shown, the present invention provides an intelligent content production system based on user needs, the system comprising:

[0065] A case database, which is used to collect and store case judgment documents, process the case judgment documents through a Seq2Seq model, and generate corresponding case summaries;

[0066] A user input module, used to receive a user's description file of the facts of a legal case and forward it to a temporary file cache module;

[0067] A temporary file cache module, used for converting the case fact description file into a text file;

[0068] A resource utilization rate judgment module, used to calculate the current system resource utilization rate; when the system resource utilization rate is greater than a first threshold, the control system releases part of the system resources so that the updated system resource utilization rate is lower than the first threshold;

[0069] A file character length judgment module is used to calculate the character length L of the text file, and when L is less than a second threshold, control the temporary file cache module to forward the text file to the first content production submodule of the content production module; when L is greater than or equal to the second threshold, control the temporary file cache module to forward the text file to the second content production submodule of the content production module;

[0070] The content production module includes a first content production sub-module and a second content production sub-module, and the content production module is used to output intelligently matched legal provisions and corresponding case summaries and case judgment documents through the first content production sub-module or the second content production sub-module.

[0071] Correspondingly, the present invention also provides an electronic device, comprising: a memory and a processor, wherein the memory and the processor are coupled; the memory stores program instructions, and when the program instructions are executed by the processor, the electronic device executes the above-mentioned intelligent content production method based on user needs.

[0072] Correspondingly, the present invention also provides a computer-readable storage medium storing one or more programs, wherein the one or more programs enable a computer to execute the above-mentioned intelligent content production method based on user needs.

[0073] The basic principles of the present disclosure are described above in conjunction with specific embodiments. However, it should be noted that the advantages, strengths, effects, etc. mentioned in the present disclosure are only examples and not limitations, and it cannot be considered that these advantages, strengths, effects, etc. must be possessed by each embodiment of the present disclosure. In addition, the specific details disclosed above are only for the purpose of illustration and ease of understanding, and are not limiting. The above details do not limit the present disclosure to be implemented by adopting the above specific details.

[0074] Each embodiment in this specification is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the system embodiment, since it basically corresponds to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0075] The block diagrams of the devices, apparatuses, equipment, and systems involved in this disclosure are only illustrative examples and are not intended to require or imply that they must be connected, arranged, and configured in the manner shown in the block diagrams. As will be appreciated by those skilled in the art, these devices, apparatuses, equipment, and systems can be connected, arranged, and configured in any manner. Words such as "including," "comprising," "having," and the like are open words, referring to "including but not limited to," and can be used interchangeably therewith. The words "or" and "and" used herein refer to the words "and / or," and can be used interchangeably therewith, unless the context clearly indicates otherwise. The word "such as" used herein refers to the phrase "such as but not limited to," and can be used interchangeably therewith.

[0076] The method and apparatus of the present disclosure may be implemented in many ways. For example, the method and apparatus of the present disclosure may be implemented by software, hardware, firmware, or any combination of software, hardware, and firmware. The above order of steps for the method is for illustration only, and the steps of the method of the present disclosure are not limited to the order specifically described above, unless otherwise specifically stated. In addition, in some embodiments, the present disclosure may also be implemented as a program recorded in a recording medium, which includes machine-readable instructions for implementing the method according to the present disclosure. Therefore, the present disclosure also covers a recording medium storing a program for executing the method according to the present disclosure.

[0077] It should also be noted that in the apparatus, equipment and method of the present disclosure, each component or each step can be decomposed and / or recombined. These decompositions and / or recombinations should be regarded as equivalent schemes of the present disclosure. The above description of the disclosed aspects is provided to enable any technician in the field to make or use the present disclosure. Various modifications to these aspects are very obvious to those skilled in the art, and the general principles defined herein can be applied to other aspects without departing from the scope of the present disclosure. Therefore, the present disclosure is not intended to be limited to the aspects shown here, but to the widest scope consistent with the principles and novel features disclosed herein.

[0078] The above description has been given for the purpose of illustration and description. In addition, this description is not intended to limit the embodiments of the present disclosure to the forms disclosed herein. Although multiple example aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, changes, additions and sub-combinations thereof.

Claims

1. An intelligent content production method based on user needs, characterized in that: The method comprises the following steps: (1) The case database collects and stores case judgment documents, processes the case judgment documents through the Seq2Seq model, and then generates corresponding case summaries; (2) The user input module receives the user's description file of the facts of the legal case and forwards it to the temporary file cache module; (3) The temporary file cache module converts the case fact description file into a text file; (4) The resource utilization rate judgment module calculates the current system resource utilization rate. When the system resource utilization rate is less than or equal to the first threshold, the process directly proceeds to step (5); when the system resource utilization rate is greater than the first threshold, the control system releases part of the system resources so that the updated system resource utilization rate is lower than the first threshold, and the process proceeds to step (5); (5) The file character length judgment module calculates the character length L of the text file, and when L is less than a second threshold, controls the temporary file cache module to forward the text file to the first content production submodule of the content production module; when L is greater than or equal to the second threshold, controls the temporary file cache module to forward the text file to the second content production submodule of the content production module; wherein the first content production submodule is implemented based on a twin neural network model, and the second content production submodule is implemented based on a recurrent neural network model RNN; (6) The first content production submodule or the second content production submodule of the content production module outputs the intelligently matched legal provisions and the corresponding case summaries and case judgment documents; The processing process of the first content production submodule is as follows: (4.1.1) When matching similar cases, the two convolutional neural networks in the twin neural network model will perform text encoding on the input text file and the case summary of the case to be matched and calculate the similarity, and select the case summary of the case to be matched whose similarity meets the threshold according to the similarity ranking; (4.1.2) Output the case summaries that meet the above similarity threshold, as well as the corresponding laws and corresponding case judgment documents.

2. The intelligent content production method according to claim 1, characterized in that: The method of processing the case judgment document file through the Seq2Seq model to generate the corresponding case summary includes the following processing steps: (1.1) Assumptions Text and Summary They are the original text of a case judgment document and the corresponding abstract, expressed in the form of word sequence. , , and i>j, the training data is A collection of text summaries in the form of; (1.2) The encoder of the Seq2Seq model encodes the source text in the encoding stage Text Encode and get Text The context vector C , ; (1.3) In the decoding phase, the decoder of the Seq2Seq model uses the historical information of the target text sequence and the context vector C Decode to predict the output word at the current moment , the decoded output word is contained in the dictionary, through soft The max function calculates the probability of each word in the dictionary and selects the word with the highest probability as the output word.

3. The intelligent content production method according to claim 1, characterized in that: The system resource utilization is calculated using one of CPU utilization, memory utilization or storage utilization; when the system resource utilization is expressed using CPU utilization, the first threshold is set to 80%; when the system resource utilization is expressed using memory utilization, the first threshold is set to 70%; when the system resource utilization is expressed using storage utilization, the first threshold is set to 60%.

4. The intelligent content production method according to claim 1, characterized in that: The format of the user's description file of the facts of the legal case is one of voice format, PDF format, Word format or image format.

5. The intelligent content production method according to claim 1, characterized in that: The twin neural network model consists of two identical and parameter-sharing convolutional neural network structures including convolutional layers, activation functions, pooling layers, and loss functions; the processing process of the first content production submodule is as follows: (4.1.1) When matching similar cases, the two convolutional neural networks will perform text encoding on the input text file and the case summary of the case to be matched and calculate the similarity, and select the case summary of the case to be matched whose similarity meets the threshold according to the similarity ranking; (4.1.2) Output the case summaries that meet the above similarity threshold, as well as the corresponding laws and corresponding case judgment documents.

6. The intelligent content production method according to claim 1, characterized in that: The recurrent neural network model RNN is implemented by sequentially connecting an input layer, an embedding layer, a semantic encoding layer, and a content output layer.

7. The intelligent content production method according to claim 6, characterized in that: The second content production submodule performs the following operation steps during specific implementation: (4.2.1) The input layer serializes the received text file to generate a sequence describing the case facts , n is the maximum length of the sequence, and the legal article label matched by the model is expressed as , m is the number of legal article labels; the semantic matching relationship between the sequence features of the case fact description and the legal article labels is obtained by pre-training the recurrent neural network model RNN, and the training set comes from the collected case judgment documents; (4.2.2) The Embedding layer converts the sequence x The sequence in is transformed into a matrix with distance meaning and small dimension, that is, the output n × b A two-dimensional vector, b Represents the dimension of embedding word embedding; (4.2.3) The semantic encoding layer receives the output of the Embedding layer n × b A two-dimensional vector, which is expressed as ; Each word describing the facts of the case is embedded into a dimension b The word is used as the input of each unit of the RNN, and the output vector obtained by the last unit is used as the output of the RNN layer; At time t, the input of the RNN is the tth word x t , the value of the hidden layer is S t , the output is expressed as h t , Said , ; in sigmoid is the activation function, U , W e , V e is the weight vector, p , c are all bias vectors; After the entire case fact description passes through the RNN structure, only the output vector of the last cell is retained h , as the output of the RNN layer; used in the prediction stage soft The max function obtains the matching results of legal article prediction; (4.2.4) The content output layer outputs the legal provisions corresponding to the matching results, the corresponding case summaries and the case judgment documents.

8. An intelligent content production system, characterized in that: The system comprises: A case database, which is used to collect and store case judgment documents, process the case judgment documents through a Seq2Seq model, and generate corresponding case summaries; A user input module, used to receive a user's description file of the facts of a legal case and forward it to a temporary file cache module; A temporary file cache module, used for converting the case fact description file into a text file; A resource utilization rate judgment module, used to calculate the current system resource utilization rate; when the system resource utilization rate is greater than a first threshold, the control system releases part of the system resources so that the updated system resource utilization rate is lower than the first threshold; A file character length judgment module is used to calculate the character length L of the text file, and when L is less than a second threshold, control the temporary file cache module to forward the text file to a first content production submodule of a content production module; when L is greater than or equal to the second threshold, control the temporary file cache module to forward the text file to a second content production submodule of the content production module; wherein the first content production submodule is implemented based on a twin neural network model, and the second content production submodule is implemented based on a recurrent neural network model RNN; A content production module, including a first content production submodule and a second content production submodule, wherein the content production module is used to output intelligently matched legal provisions and corresponding case summaries and case judgment documents through the first content production submodule or the second content production submodule; The processing process of the first content production submodule is as follows: (4.1.1) When matching similar cases, the two convolutional neural networks in the twin neural network model will perform text encoding on the input text file and the case summary of the case to be matched and calculate the similarity, and select the case summary of the case to be matched whose similarity meets the threshold according to the similarity ranking; (4.1.2) Output the case summaries that meet the above similarity threshold, as well as the corresponding laws and corresponding case judgment documents.

9. An electronic device, characterized in that: The electronic device comprises: a memory and a processor, wherein the memory and the processor are coupled; the memory stores program instructions, and when the program instructions are executed by the processor, the electronic device executes the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The method comprises a computer program, which, when executed on an electronic device, enables the electronic device to execute the method according to any one of claims 1 to 7.

Images