Script data processing method and device, computer equipment and storage medium

Abstract

The invention belongs to the technical field of artificial intelligence, and relates to a script data processing method, which comprises the following steps: receiving theme and demand information input by a user; performing script generation processing on the theme and the demand information based on a preset large language model to obtain a corresponding initial script; performing content adjustment processing on the initial script based on a preset multi-agent to obtain a corresponding first script; performing content enhancement processing on the first script based on a preset compliance optimization strategy to obtain a corresponding second script; based on a preset target engine, performing content adaptation processing on the second script by using the demand information to obtain a corresponding target script; and outputting the target script. The invention further provides a script data processing device, computer equipment and a storage medium. In addition, the target script can be stored in the block chain. The method can be applied to insurance script generation scenes in the fields of financial science and technology and medical health insurance, and the professional degree, accuracy and compliance of script generation are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of artificial intelligence technology, and can be applied to the fields of financial technology and medical health insurance, and in particular to a script data processing method and device, a computer device and a storage medium. BACKGROUND

[0002] In the existing insurance text generation field, large language models (LLM, such as the GPT series) have many problems when generating insurance scripts, and the professional degree and compliance are not good. Specifically, these models often fabricate insurance clauses, such as generating "life compound interest 30%" which is obviously not in line with the actual situation of the industry; they also confuse professional terms, such as confusing "cash value" with "account value", resulting in a compliance review rejection rate of more than 25%. Taking a certain health insurance script as an example, "waiting period" is mistakenly described as "30-day no reason refund", while the actual situation should be "180-day observation period", and this error has caused user complaints, seriously affecting the development of insurance business and customer experience.

[0003] Similar problems also exist in the financial field in the context of generating promotional texts for financial products. Some models may fabricate the yield calculation method of a financial product, such as claiming that a certain financial product "has a monthly fixed yield growth of 15%", which is far beyond the reasonable range of the market and does not take into account market fluctuations and other risk factors. This inaccurate and non-compliant text generation method not only damages users' trust in insurance and financial products, but also poses compliance risks and reputational losses for financial institutions.

[0004] Therefore, there is an urgent need for a method that can effectively improve the professional degree and compliance of insurance text generation to ensure that the generated insurance scripts are accurate and compliant, and to protect the rights and interests of users and the stable operation of financial institutions. SUMMARY

[0005] The purpose of the embodiments of the present application is to propose a script data processing method, device, computer device and storage medium to solve the technical problems of low professional degree and low compliance of the existing method of generating insurance scripts based on large language models.

[0006] In a first aspect, a script data processing method is provided, comprising:

[0007] receiving a topic and requirement information input by a user;

[0008] performing script generation processing on the topic and the requirement information based on a preset large language model to obtain a corresponding initial script;

[0009] The initial script is content-adjusted based on a preset multi-agent to obtain a corresponding first script; wherein the multi-agent includes a fact verification agent, a logical reasoning agent, and a compliance review agent;

[0010] The first script is content-enhanced based on a preset compliance optimization strategy to obtain a corresponding second script;

[0011] The second script is content-adapted based on a preset target engine using the requirement information to obtain a corresponding target script;

[0012] The target script is output processed.

[0013] In a second aspect, a script data processing apparatus is provided, comprising:

[0014] A receiving module is configured to receive a topic and requirement information input by a user;

[0015] A generating module is configured to generate an initial script based on a preset large language model using the topic and the requirement information;

[0016] A first processing module is configured to content-adjust the initial script based on a preset multi-agent to obtain a corresponding first script; wherein the multi-agent includes a fact verification agent, a logical reasoning agent, and a compliance review agent;

[0017] A second processing module is configured to content-enhance the first script based on a preset compliance optimization strategy to obtain a corresponding second script;

[0018] A third processing module is configured to content-adapt the second script based on a preset target engine using the requirement information to obtain a corresponding target script;

[0019] An output module is configured to output process the target script.

[0020] In a third aspect, a computer device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the script data processing method.

[0021] In a fourth aspect, a computer readable storage medium is provided, which stores a computer program, wherein the computer program is executable by a processor to implement the steps of the script data processing method.

[0022] In the above-mentioned scheme implemented by the script data processing method, apparatus, computer equipment, and storage medium, the user input of topic and requirement information is first received; then, based on a preset large language model, the topic and requirement information are processed to generate a script, resulting in a corresponding initial script; subsequently, based on a preset multi-agent system, the initial script is processed to adjust its content, resulting in a corresponding first script; wherein, the multi-agent system includes a fact-checking agent, a logical reasoning agent, and a compliance review agent; subsequently, based on a preset compliance optimization strategy, the first script is processed to enhance its content, resulting in a corresponding second script; further, based on a preset target engine, the requirement information is used to adapt the second script to its content, resulting in a corresponding target script; finally, the target script is output. After receiving the user input of topic and requirement information, this application first processes the topic and requirement information to generate an initial script based on a large language model, then processes the initial script to adjust its content based on a multi-agent system, resulting in a first script; then processes the first script to enhance its content based on a compliance optimization strategy, resulting in a second script; subsequently, based on a target engine, the requirement information is used to adapt the second script to its content, resulting in a target script; finally, the target script is output. After obtaining the initial script generated by the large language model, this application automatically and intelligently optimizes the content of the initial script based on multi-agent, compliance optimization strategies, and the collaborative use of the target engine. This effectively optimizes the content generation of the script and improves the professionalism, accuracy, and compliance of the generated target script. Attached Figure Description

[0023] To more clearly illustrate the solutions in this application, the accompanying drawings used in the description of the embodiments of this application will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0024] Figure 1 This is an exemplary system architecture diagram to which this application can be applied;

[0025] Figure 2 This is a flowchart of one embodiment of the script data processing method according to this application;

[0026] Figure 3 This is a schematic diagram of the structure of an embodiment of the script data processing apparatus according to this application;

[0027] Figure 4 This is a schematic diagram of the structure of one embodiment of the computer device according to this application. Detailed Implementation

[0028] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs; the terminology used in the description herein is for describing particular embodiments only and is not intended to be limiting upon the application; the terms "comprising," "including," and "having," and variations thereof, as used in enrolling and claims herein, are intended to be open-ended and to mean including, but not limited to; the terms "first," "second," and the like, as used in the description herein, are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs; the terminology used in the description herein is for describing particular embodiments only and is not intended to be limiting upon the application; the terms "comprising," "including," and "having," and variations thereof, as used in enrolling and claims herein, are intended to be open-ended and to mean including, but not limited to; the terms "first," "second," and the like, as used in the description herein, are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs; the terminology used in the description herein is for describing particular embodiments only and is not intended to be limiting upon the application; the terms "comprising," "including," and "having," and variations thereof, as used in enrolling and claims herein, are intended to be open-ended and to mean including, but not limited to; the terms "first," "second," and the like, as used in the description herein, are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order.

[0029] Reference herein to "an embodiment" means that a particular feature, structure, or characteristic described in connection with the embodiment can be included in at least one embodiment of the application. The appearances of the phrase in various places in the specification are not necessarily all referring to the same embodiment, nor are they necessarily mutually exclusive or alternative embodiments. It is expressly understood that the embodiments described herein are merely examples from a whole class of comparable embodiments which those skilled in the art will readily appreciate. It is also expressly understood that the description herein and the claims that follow are intended to cover all such variations as will become apparent to those in the art to which this application is related.

[0030] In order to make the technical personnel in the art better understand the scheme of the application, the technical solutions in the embodiments of the application will be clearly and completely described below with reference to the drawings.

[0031] As shown in Figure 1 The system architecture 100 can include a terminal device 101, a network 102, and a server 103. The terminal device 101 can be a notebook computer 1011, a tablet computer 1012, or a mobile phone 1013. The network 102 is a medium for providing a communication link between the terminal device 101 and the server 103. The network 102 can include various connection types, such as wired, wireless communication links, or optical fiber cables, etc.

[0032] A user can use the terminal device 101 to interact with the server 103 through the network 102 to receive or send messages, etc. Various communication client applications can be installed on the terminal device 101, such as web browser applications, shopping applications, search applications, instant messaging tools, email clients, social platform software, etc.

[0033] The terminal device 101 can be various electronic devices with a display screen and supporting web browsing, in addition to the notebook computer 1011, the tablet computer 1012 or the mobile phone 1013, the terminal device 101 can also be an electronic book reader, an MP3 player (Moving Picture Experts Group Audio Layer III), an MP4 player (Moving Picture Experts Group Audio Layer IV), a laptop computer and a desktop computer, etc.

[0034] The server 103 can be a server providing various services, for example, a background server providing support for a page displayed on the terminal device 101.

[0035] It should be noted that the script data processing method provided by the embodiments of the present application is generally executed by a server / terminal device, and accordingly, the script data processing apparatus is generally provided in the server / terminal device.

[0036] It should be understood that Figure 1 The number of terminal devices, networks and servers in

[0037] With reference to Figure 2 , a flowchart of one embodiment of the script data processing method according to the present application is shown. The order of the steps in the flowchart can be changed according to different needs, and some steps can be omitted. The script data processing method provided by the embodiments of the present application can be applied to any scenario requiring script data processing, and then the script data processing method can be applied to products in these scenarios, for example, insurance script generation scenarios in the fields of financial technology and medical health insurance. The script data processing method includes the following steps:

[0038] Step S201, receiving theme and requirement information input by a user.

[0039] In this embodiment, the electronic device (for example Figure 1The server / terminal device shown) can obtain the theme and demand information input by the user through wired or wireless connection. It should be noted that the wireless connection can include but is not limited to 3G / 4G / 5G connection, WiFi connection, Bluetooth connection, WiMAX connection, Zigbee connection, UWB (ultra wide band) connection, and other now known or future developed wireless connection. The subject of the present application is a script processing system, which can be referred to as a system. The present application can be applied to the generation of insurance scripts in the field of financial technology and medical health insurance. Insurance scripts are commonly used and widely understood terms in the context of insurance business-related fields and around insurance product content creation and dissemination. For example, insurance scripts can be widely used in industry internal communication, cross-field cooperation, professional literature and data business scenarios. Among them, industry internal communication: within the insurance industry, from product planning, promotion (such as annuity product promotion video, critical illness insurance promotion video, etc.) to sales training, etc. Different personnel (such as product managers, marketing personnel, training instructors, etc.) will frequently use the term "insurance script" when communicating insurance product-related content creation. For example, product managers and marketing teams discuss how to optimize insurance product promotion copy scripts to better attract target customers; training instructors prepare sales scripts for sales personnel to improve sales skills. Cross-field cooperation: when insurance companies cooperate with external agencies (such as advertising companies, video production teams) to carry out promotional activities, "insurance scripts" are also commonly used communication vocabulary. Advertising companies need to design promotional posters and shoot promotional videos according to the requirements of insurance scripts; video production teams will shoot and edit videos according to the video script. Professional literature and data: in the insurance industry's professional books, research reports, academic papers, etc. The term "insurance script" is often used. These materials will discuss how to optimize insurance scripts to improve the market competitiveness of insurance products, customer satisfaction, etc.

[0040] In addition, the theme and demand information is the information input by the user according to the actual needs of the insurance script generation. For example, the theme can be "annuity product explanation", and the demand information can be video length (such as 30 seconds) and compliance requirements (such as "must include hesitation period prompt"). The theme and demand information are used to clarify the direction, length limit and key compliance points of script generation, and provide a basic framework for subsequent generation.

[0041] Step S202, based on a preset large language model, the theme and the demand information are processed to obtain a corresponding initial script.

[0042] In the embodiment, the selection of the large language model is not specifically limited, and can be determined according to the business selection. For example, an existing LLM model (such as the GPT series) can be used. The system preloads a knowledge base containing an insurance rule base and a term atlas into the insurance script generation process. The insurance rule base contains 583 clauses, covering various regulations and standards in the insurance industry; the term atlas establishes the association between professional terms, such as “premium exemption → trigger condition → application process”. These knowledge bases provide accurate and professional basis for subsequent script generation, ensuring that the generated script meets the requirements of insurance industry standards and term usage.

[0043] Specifically, the theme and demand information can be input into the selected large language model, and the large language model can infer the theme and demand information according to the use of the knowledge base, and generate an initial script that meets the requirements.

[0044] In step S203, the initial script is content-adjusted based on a preset multi-agent to obtain a corresponding first script; wherein the multi-agent includes a fact verification agent, a logic reasoning agent, and a compliance review agent.

[0045] In the embodiment, the specific implementation process of the content adjustment of the initial script based on the preset multi-agent to obtain the corresponding first script will be further described in detail in the subsequent specific embodiments, and will not be described here.

[0046] In step S204, the first script is content-enhanced based on a preset compliance optimization strategy to obtain a corresponding second script.

[0047] In the embodiment, the specific implementation process of the content enhancement of the first script based on the preset compliance optimization strategy to obtain the corresponding second script will be further described in detail in the subsequent specific embodiments, and will not be described here.

[0048] In step S205, the second script is content-adapted based on a preset target engine using the demand information to obtain a corresponding target script.

[0049] In the embodiment, the specific implementation process of the content adaptation of the second script based on the preset target engine using the demand information to obtain the corresponding target script will be further described in detail in the subsequent specific embodiments, and will not be described here.

[0050] In step S206, the target script is output.

[0051] In the embodiment, the specific implementation process of the output processing of the target script will be further described in detail in subsequent specific embodiments, and will not be described here.

[0052] For example, taking the optimization of an annuity insurance benefit demonstration script as an example, the process of generating an optimized script by the script processing method proposed in the application includes: an initial script generated originally: “This product provides life annuity, the annual yield can reach 8%, the insurance amount grows with compound interest, and the wealth is passed down from generation to generation.” System correction process: 1. Fact verification: detecting that “annual yield 8%” violates “New Personal Insurance Product Information Disclosure Management Method”, triggering replacement strategy; 2. Term correction: change to “non-guaranteed bonus part historical annual yield rate reference value 4.2% (2000-2023 data)”; 3. Logic verification: find that “generation to generation” is not associated with Article 42 of “Insurance Law”, automatically insert “the death insurance will be distributed to the designated beneficiary according to the contract”. The final compliant script generated: “This product provides life annuity, the non-guaranteed bonus part historical annual yield rate reference value is 4.2% (2000-2023 data), and the specific actual operation will be subject to the actual operation. The death insurance will be distributed to the designated beneficiary according to the contract, and the life insurance (no fixed termination age) will be guaranteed.

[0053] First, the application receives the theme and demand information input by the user; then, based on the preset large language model, the theme and the demand information are processed to generate an initial script; then, based on the preset multi-agent, the content of the initial script is adjusted to obtain a first script; wherein the multi-agent includes a fact verification agent, a logic reasoning agent and a compliance review agent; subsequently, based on the preset compliance optimization strategy, the content of the first script is enhanced to obtain a second script; further based on the preset target engine, the demand information is used to adapt the content of the second script to obtain a target script; finally, the target script is output. After receiving the theme and demand information input by the user, the application will first generate an initial script based on the use of a large language model, then adjust the content of the initial script based on the use of a multi-agent to obtain a first script, then enhance the content of the first script based on the use of a compliance optimization strategy to obtain a second script, then adapt the content of the second script based on a target engine using demand information to obtain a target script, and finally output the target script. After obtaining the initial script generated by the large language model, the application will automatically and intelligently optimize the content of the initial script based on the collaborative use of the multi-agent, the compliance optimization strategy and the target engine, effectively optimizing the content generation of the script and improving the professionalism, accuracy and compliance of the generated target script.

[0054] In some optional implementations, step S203 comprises the following steps:

[0055] Based on the fact verification agent, the initial script is subjected to fact verification using a preset insurance rule base, and the initial script is subjected to corresponding content modification based on the obtained fact verification result, to obtain a corresponding first generated script.

[0056] In the present embodiment, the insurance rule base is a pre-constructed database covering various regulations and standards of the insurance industry, and the insurance rule base is pre-loaded into the script generation process. The fact verification agent detects professional terms in the initial script in real time according to the loaded insurance rule base. For example, it checks whether the “death benefit payment conditions of whole life insurance” are accurately stated, and when it detects that the script content in the initial script contains abnormal content that does not conform to the facts, it automatically modifies the first abnormal content according to the insurance rule base to obtain a modified first generated script, so as to ensure that the first generated script does not contain fictitious or misleading information, and to ensure the authenticity of the script content.

[0057] Based on the logic reasoning agent, the first generated script is subjected to logic verification using a preset verification model, and the first generated script is subjected to corresponding content modification based on the obtained logic verification result, to obtain a corresponding second generated script.

[0058] In the present embodiment, the verification model can specifically be an LSTM network. The logic reasoning agent can use the verification model to verify the time sequence logic of the first generated script, for example, to check whether the logical relationship “the insured age is 20 years old, and the effective period of the insurance is 30 days after the insurance is purchased” is reasonable, and when it detects that the script content in the first generated script contains second abnormal content that is logically unreasonable, it automatically corrects the second abnormal content to obtain a modified second generated script, so as to ensure that the script content is logically coherent and free of contradictions.

[0059] Based on the compliance review agent, the second script is subjected to term correction using a preset regulatory rule base, to obtain a corresponding third generated script.

[0060] In the present embodiment, the regulatory rule base can specifically be a third-party regulatory rule base. Based on the compliance review agent, the script content of the second script is matched with the regulatory rule base, and the non-compliant expressions are marked and deleted, for example, once a non-compliant expression such as “guaranteed return” is detected, it is immediately marked and subjected to term correction (i.e. the non-compliant expression is deleted), to obtain a modified third generated script, so as to ensure that the obtained third generated script complies with regulatory requirements.

[0061] The third generated script is taken as the first script.

[0062] In this embodiment, the obtained first script can also be preliminarily integrated in a logical order to form a coherent script framework. For example, the content such as the insurance age, the effective period of the insurance, the waiting period, the insurance range, the benefit description, and the like is arranged in a reasonable order to ensure that the script content is smooth and natural.

[0063] The application verifies the initial script based on the fact verification intelligent agent, uses a preset insurance rule library to perform fact verification on the initial script, and performs corresponding content modification on the initial script based on the obtained fact verification result to obtain a corresponding first generated script. Then, based on the logic reasoning intelligent agent, a preset verification model is used to perform logic verification on the first generated script, and corresponding content modification is performed on the first generated script based on the obtained logic verification result to obtain a corresponding second generated script. Then, based on the compliance review intelligent agent, a preset regulatory rule library is used to perform term modification on the second script to obtain a corresponding third generated script. Subsequently, the third generated script is taken as the first script. Through the cooperation of the fact verification intelligent agent, the logic reasoning intelligent agent, and the compliance review intelligent agent, the application can efficiently and accurately adjust the content of the initial script to obtain the required first script, effectively optimize the generation of the script, and improve the professionalism and compliance of the script.

[0064] In some optional implementation manners of this embodiment, step S204 includes the following steps:

[0065] Based on the preset compliance rule library, a high-risk expression existing in the first script is identified.

[0066] In this embodiment, the system can scan the script content of the first script generated by the above-mentioned multiple intelligent agents through the built-in compliance rule library to identify high-risk expressions that may cause compliance risks. For example, if the expression "guaranteed income" is detected, the system will immediately mark it out.

[0067] The high-risk expression in the first script is replaced to obtain a corresponding fourth generated script.

[0068] In this embodiment, once the high-risk expression is identified, the system will automatically replace it with a more accurate and compliant expression. For example, replace "guaranteed returns" with "historical annualized returns of 3.5% (2023 data)" and insert a risk prompt "actual returns subject to actual business conditions" after it. This not only avoids the risk of false promises, but also ensures the transparency of information. After completing the replacement of high-risk expressions, the system can again verify the compliance of the replaced content to ensure that the new expression fully complies with regulatory requirements and does not cause user complaints or compliance risks.

[0069] determining whether the fourth generated script contains a key clause.

[0070] In this embodiment, the content matching process of the key clause can be performed on the fourth generated script to determine whether the fourth generated script contains a key clause. The clause settings of the key clause can be determined according to actual business needs.

[0071] If yes, the fourth generated script is supplemented with corresponding content based on the key clause to obtain a corresponding fifth generated script.

[0072] In this embodiment, when the fourth generated script is identified as containing a key clause, the system automatically associates relevant laws and regulations, such as relevant clauses in the Insurance Law. Further, the system automatically supplements necessary legal content according to the associated laws and regulations. For example, when referring to "major illness protection", supplement "the death insurance benefit will be distributed to the designated beneficiaries according to the contract" and other content to ensure that the script content not only complies with product clauses, but also complies with legal requirements. In addition, after completing the content supplement process of the fourth generated script, the system will perform legal compliance audit to ensure that all supplemented content is accurate and consistent with laws and regulations.

[0073] The fifth generated script is used as the second script.

[0074] The present application identifies high-risk expressions in the first script based on a pre-set compliance rule library; then replaces the high-risk expressions in the first script to obtain a corresponding fourth generated script; then determines whether the fourth generated script contains a key clause; if yes, the fourth generated script is supplemented with corresponding content based on the key clause to obtain a corresponding fifth generated script; subsequently, the fifth generated script is used as the second script. The present application replaces high-risk expressions and supplements content associated with clauses in the first script, thereby efficiently and accurately completing the content enhancement process of the first script, and effectively improving the compliance of the first script.

[0075] In some optional implementations, step S205 includes the following steps:

[0076] Based on the target engine, the corresponding accommodated text amount is determined based on the requirement information.

[0077] In this embodiment, the above is a pre-constructed dynamic compliance adaptation engine with a content adaptation processing function for scripts. According to actual business requirements, the maximum accommodated text amount corresponding to different video lengths is pre-defined. Specifically, the requirement information can include a target video length (such as 45 seconds) and compliance requirements (such as "a hesitation period prompt is required"). According to the mapping relationship between the video length and the maximum accommodated text amount, the accommodated text amount corresponding to the target video length in the above requirement information can be determined. For example, a 45-second video can need to control the script content within 180 words.

[0078] The accommodated text amount corresponding to the requirement information calculated and generated has the following effects: 1. Provides a text capacity limit basis: The maximum accommodated text amount corresponding to different video lengths is defined, which sets a clear boundary for subsequent script generation and optimization work. It ensures that the generated insurance script does not exceed the information capacity that the video can carry, avoiding problems such as video rhythm confusion and unclear information transmission due to too much text. For example, under a 45-second video length, a text amount of about 180 words is determined, which helps the system know that the content amount should not exceed this limit when generating the script, which helps to control the size of the script as a whole. 2. Guide script content planning: Helps the system have a preliminary direction for content planning before generating the script. After knowing the maximum text amount, the system can more targetedly select the insurance product information to be presented, preferentially retain key content, and reasonably allocate the length of each information point, so that the script content is more compact and efficient.

[0079] Based on the accommodated text amount, the second script is optimized in content using a preset greedy algorithm to obtain a corresponding sixth generated script.

[0080] In this embodiment, the system uses a greedy algorithm to optimize the script content of the generated second script according to the obtained accommodated text amount. Specifically, high-confidence professional terms (such as "waiting period 180 days" and "death insurance") are retained, which are crucial for accurately conveying key information of the insurance product.

[0081] The calculated maximum text amount (i.e., the accommodated text amount) determines the upper limit of the script content, and the greedy algorithm optimization further screens the content within this upper limit. The greedy algorithm ensures that the most important insurance information is conveyed within the limited text space by retaining high-confidence professional terms such as "waiting period 180 days" and "death benefit". For example, under the text amount limit of 180 characters, the greedy algorithm will prioritize retaining these key terms because they are crucial for accurately conveying the core terms and features of the insurance product.

[0082] In addition, the maximum text amount limit corresponding to the accommodated text amount provides an operating space for the greedy algorithm, which optimizes the content within this space. The combined effect of the two allows the script to meet the length requirement while improving the professionalism. By retaining professional terms, the script can more accurately introduce the insurance product and enhance the user's understanding of the product.

[0083] Based on the accommodated text amount, the sixth generated script is subjected to a redundant content deletion process to obtain a corresponding seventh generated script.

[0084] In this embodiment, the system will retain key terms while deleting redundant modifiers (such as "absolutely safe and reliable" and "absolutely trustworthy") based on the obtained accommodated text amount. These words may increase the attractiveness of the script, but they often lack substantive content and may cause confusion.

[0085] The calculation of the maximum text amount (i.e., the accommodated text amount) allows the system to clearly identify which content is redundant and exceeds the capacity, providing a clear target for deleting redundant content. After determining that a 45-second video corresponds to 180 characters of text, the system can compare the generated script content to identify parts that exceed this number of characters and delete redundant modifiers (such as "absolutely safe and reliable" and "absolutely trustworthy") as the deletion target.

[0086] In addition, the maximum text amount limit and the deletion of redundant content are both aimed at ensuring the simplicity and clarity of the script. The maximum text amount controls the length of the script from the total amount, and the deletion of redundant content further simplifies the script, making the information more focused and prominent. For example, under the premise of not exceeding 180 characters, deleting redundant modifiers can make the script more concise and highlight the key points, which is consistent with the characteristics of video dissemination and the user's need for quick information acquisition.

[0087] The seventh generated script is used as the target script.

[0088] In the embodiment, after the content adaptation processing on the second script is completed and the seventh generated script is obtained, the seventh generated script can also be subjected to brevity verification, and the seventh generated script passing the brevity verification is taken as the final target script, so that it can be ensured that the generated target script meets the time length requirement while the content is concise and clear and the key points are highlighted, and the core information of the insurance product can be effectively conveyed.

[0089] Based on the target engine, the application determines the corresponding accommodated text amount according to the demand information, and then uses a preset greedy algorithm to perform content optimization on the second script based on the accommodated text amount, to obtain a corresponding sixth generated script. Then, the sixth generated script is subjected to a redundant content deletion processing based on the accommodated text amount, to obtain a corresponding seventh generated script. Subsequently, the seventh generated script is taken as the target script. The application performs the processing of calculating the accommodated text amount, the greedy algorithm optimization and the deletion of redundant content on the second script based on the use of the target engine, so that the content adaptation processing on the second script can be efficiently and accurately completed, and it is effectively ensured that the generated target script meets the requirements of the demand information while having high professionalism and brevity.

[0090] In some optional implementations, step S206 includes the following steps:

[0091] A preset script auditing strategy is obtained.

[0092] In the embodiment, the policy content of the above-mentioned script auditing strategy includes professional degree auditing, compliance auditing and logicality auditing. The professional degree auditing includes auditing the script to ensure that all professional terms are used accurately and the clause description is correct. For example, whether the "waiting period is 180 days" conforms to the actual clauses of the company's product and whether the major illness protection range is complete. The compliance auditing includes auditing the script to ensure that all content meets the regulatory requirements of the China Banking and Insurance Regulatory Commission and there is no sensitive or illegal expression. For example, whether there is illegal content such as "guaranteed return" and whether the risk prompt is in place. The logicality auditing includes auditing the script to ensure that the script content is logically coherent and has no contradictions. For example, whether the relationship between the insured age and the protection effective period is reasonable and whether the relationship between the waiting period and the protection responsibility is clear.

[0093] The target script is subjected to auditing processing based on the script auditing strategy, to obtain a corresponding auditing result.

[0094] In the embodiment, the auditing process on the target script can be performed based on the processing steps included in the policy content of the script auditing policy, and the corresponding auditing result is generated. The auditing result includes passing or failing. Specifically, only when the target script passes the professional degree auditing, the compliance auditing, and the logic auditing, the target script is determined to pass the auditing, and the corresponding passing auditing result is generated.

[0095] If the auditing result is passing, a preset script output mode is obtained.

[0096] In the embodiment, the selection of the script output mode is not specifically limited, and can be determined according to the actual needs of the user, for example, any one of the modes such as email sending, interface display, and message sending can be used.

[0097] The target script is output based on the script output mode.

[0098] In the embodiment, the target script can be sent to the user according to the selected script output mode, so as to complete the output processing of the target script.

[0099] The application obtains a preset script auditing policy, then performs auditing processing on the target script based on the script auditing policy to obtain a corresponding auditing result, if the auditing result is passing, a preset script output mode is obtained, and subsequently, the target script is output based on the script output mode. After the target script is generated, the application automatically performs auditing processing on the target script by using the script auditing policy, so as to further check the accuracy and compliance of the target script. When it is detected that the generated auditing result is passing, the target script is intelligently output based on the obtained script output mode, so that the accuracy and compliance of the output target script can be effectively ensured, and the use experience of the user is improved.

[0100] In some optional implementation manners of the embodiment, after step S206, the electronic device can further perform the following steps:

[0101] A pre-created dynamic reward function is obtained.

[0102] In the embodiment, the construction process of the dynamic reward function will be further described in detail in the subsequent specific embodiments, and will not be described here.

[0103] A preset target reinforcement learning algorithm is called.

[0104] In the embodiment, the target reinforcement learning algorithm can specifically adopt a PPO algorithm. The PPO algorithm is a reinforcement learning algorithm based on policy gradient, which updates the behavior policy of an agent (Agent) through policy gradient according to the reward value given by the dynamic reward function. In simple terms, the agent knows that taking action in what situation can obtain a higher reward.

[0105] Based on the target reinforcement learning algorithm, the behavior policy of the multi-agent is optimized using the dynamic reward function.

[0106] In the embodiment, based on the use of the target reinforcement learning algorithm, the behavior policy of the multi-agent is updated through policy gradient according to the reward value given by the dynamic reward function. Specifically, if a certain behavior of the agent (such as generating content of accurate reference clauses) obtains a higher positive reward, the PPO algorithm will increase the probability of the agent taking this behavior in the future; on the contrary, if a certain behavior (such as generating illusion content) obtains a negative punishment, the PPO algorithm will reduce the probability of the agent taking this behavior in the future. By continuously adjusting the behavior policy of the agent according to the dynamic reward function, the PPO algorithm can make the agent pay more attention to compliance and accuracy in the subsequent generation process.

[0107] The present application obtains a pre-created dynamic reward function, then calls a preset target reinforcement learning algorithm, and subsequently optimizes the behavior policy of the multi-agent using the dynamic reward function based on the target reinforcement learning algorithm. Through the mutual cooperation of the dynamic reward function and the target reinforcement learning algorithm, the dynamic reward function sets a clear optimization target and direction for the multi-agent, and the target reinforcement learning algorithm adjusts the behavior policy of the multi-agent according to these targets. The two work together to enable the multi-agent to continuously improve the generation strategy and generate more high-quality and compliant script data, thereby effectively improving the ability of the multi-agent to generate scripts and improving the quality of the generated scripts.

[0108] In some optional implementations of the embodiment, the obtaining of the pre-created dynamic reward function includes the following steps:

[0109] Obtaining a preset positive incentive rule.

[0110] In the embodiment, the setting of the positive incentive rule is not specifically limited, and can be determined according to actual business needs. For example, an accurate reference to insurance clauses is rewarded with +3 points, a logical closed loop is rewarded with +2 points, and a compliant expression is rewarded with +1 point. These incentive measures encourage multi-agents to generate high-quality and compliant script content. Among them, the positive incentive provides a clear goal orientation for multi-agents to generate high-quality and compliant script content. These positive rewards are equivalent to “good reviews” for multi-agents. When generating scripts, multi-agents will tend to adopt behaviors that can obtain these rewards.

[0111] The preset negative punishment rule is obtained.

[0112] In the embodiment, the setting of the negative punishment rule is not specifically limited, and can be determined according to actual business needs. Specifically, inaccurate and non-compliant content can be negatively punished. For example, hallucination content (fictional insurance clauses) is deducted -5 points, and term confusion is deducted -2 points. Through this punishment method, multi-agents are encouraged to continuously improve their generation strategies. Among them, the negative punishment is equivalent to “bad reviews” for multi-agents. Multi-agents will try to avoid these punished behaviors.

[0113] Based on the positive incentive rule and the negative punishment rule, a function construction process is performed to obtain a corresponding target function.

[0114] In the embodiment, a corresponding reward function can be designed according to the positive incentive rule and the negative punishment rule to obtain a target function containing the positive incentive rule and the negative punishment rule, and serve as a required dynamic reward function.

[0115] The target function is used as the dynamic reward function.

[0116] In the embodiment, when optimizing the behavior of multi-agents, the target reinforcement learning algorithm needs to determine the good and bad of the current behavior of multi-agents according to a certain signal, and the reward value generated by the dynamic reward function is just such a signal. Each behavior (generating script content) of multi-agents corresponds to a reward value, and the target reinforcement learning algorithm evaluates whether the behavior strategy of multi-agents is effective according to these reward values.

[0117] Among them, in such a complex environment of insurance script generation, the dynamic reward function can flexibly adjust the reward rules according to different business needs and compliance requirements, and the target reinforcement learning algorithm can quickly adapt to these changes by optimizing the strategy of multi-agents to meet new requirements. This synergistic effect enables the entire system to better cope with various complex situations and continuously improve the ability to generate scripts.

[0118] The application obtains a preset positive incentive rule and a preset negative punishment rule, and then performs function construction on the positive incentive rule and the negative punishment rule, to obtain a corresponding target function. The application obtains the preset positive incentive rule and the preset negative punishment rule, and then performs function construction on the positive incentive rule and the negative punishment rule, so that the required dynamic reward function can be efficiently and accurately constructed, the construction efficiency of the dynamic reward function is improved, and the mutual cooperation of the dynamic reward function and the target reinforcement learning algorithm is facilitated, so that the multi-agent can continuously improve the generated strategy, and generate more high-quality and compliant script data, thereby effectively improving the ability of the multi-agent to generate scripts and improving the quality of the generated scripts.

[0119] In some optional implementations, the obtained user information seeks user consent and complies with relevant laws and relevant policies.

[0120] In addition, the non-company software tools or components appearing in the embodiments of the application are only examples and do not represent actual use.

[0121] It should be understood that the size of the serial number of each step in the above embodiments does not mean the order of execution, and the execution order of each process should be determined according to its function and inherent logic, and should not constitute any limitation on the implementation process of the embodiments of the application.

[0122] It should be emphasized that, in order to further ensure the privacy and security of the above target script, the above target script can also be stored in a node of a blockchain.

[0123] The blockchain referred to in the application is a new application mode of distributed data storage, peer-to-peer transmission, consensus mechanism, encryption algorithm and other computer technologies. Blockchain, in essence, is a decentralized database, a series of data blocks associated using cryptographic methods, each data block containing information about a batch of network transactions, used to verify the validity (anti-fraud) of the information and generate the next block. The blockchain can include a blockchain underlying platform, a platform product service layer, and an application service layer.

[0124] The embodiments of the application can acquire and process related data based on artificial intelligence technology. Artificial intelligence (AI) is the use of digital computers or computer-controlled machines to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.

[0125] The artificial intelligence basic technology generally includes technologies such as sensors, special artificial intelligence chips, cloud computing, distributed storage, big data processing technology, operation / interaction system, mechatronics, etc. The artificial intelligence software technology mainly includes computer vision technology, robot technology, biometric identification technology, speech processing technology, natural language processing technology, and machine learning / deep learning, etc.

[0126] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by computer readable instructions instructing related hardware, and the computer readable instructions can be stored in a computer readable storage medium. When the program is executed, it can include the processes of the above-mentioned embodiment methods. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only memory (ROM) and other non-volatile storage media, or a random access memory (RAM) and the like.

[0127] It should be understood that, although each step in the flowchart of the accompanying drawings is shown in sequence according to the direction of the arrow, these steps are not necessarily executed in sequence according to the direction of the arrow. Unless otherwise specified herein, the execution of these steps is not strictly limited in sequence, and they can be executed in other orders. Moreover, at least part of the steps in the flowchart of the accompanying drawings can include multiple sub-steps or multiple stages, which are not necessarily executed at the same time, but can be executed at different times, and the execution sequence is not necessarily sequential, but can be alternately executed with at least part of other steps or sub-steps or stages of other steps.

[0128] Further referring to Figure 3 , as an implementation of the method shown in Figure 2 , the present application provides an embodiment of a script data processing device, which corresponds to the method embodiment shown in Figure 2 , and the device can be specifically applied to various electronic devices.

[0129] As shown in Figure 3 , the script data processing device 300 described in the embodiment includes a receiving module 301, a generating module 302, a first processing module 303, a second processing module 304, a third processing module 305, and an output module 306. Among them:

[0130] The receiving module 301 is configured to receive a topic and demand information input by a user;

[0131] The generating module 302 is configured to perform script generation processing on the topic and the demand information based on a preset large language model, to obtain a corresponding initial script;

[0132] The first processing module 303 is configured to perform content adjustment processing on the initial script based on a preset multi-agent to obtain a corresponding first script; wherein the multi-agent includes a fact verification agent, a logic reasoning agent, and a compliance review agent.

[0133] The second processing module 304 is configured to perform content enhancement processing on the first script based on a preset compliance optimization strategy to obtain a corresponding second script.

[0134] The third processing module 305 is configured to perform content adaptation processing on the second script based on a preset target engine using the requirement information to obtain a corresponding target script.

[0135] The output module 306 is configured to perform output processing on the target script.

[0136] In the embodiment, the operations performed by the above modules or units correspond to the steps of the script data processing method of the foregoing embodiments one by one, and will not be described here.

[0137] In some optional implementations of the embodiment, the first processing module 303 includes:

[0138] The first processing submodule is configured to perform fact verification on the initial script using a preset insurance rule base based on the fact verification agent, and perform corresponding content modification on the initial script based on the obtained fact verification result to obtain a corresponding first generated script.

[0139] The second processing submodule is configured to perform logic verification on the first generated script using a preset verification model based on the logic reasoning agent, and perform corresponding content modification on the first generated script based on the obtained logic verification result to obtain a corresponding second generated script.

[0140] The third processing submodule is configured to perform term correction on the second script using a preset supervision rule base based on the compliance review agent to obtain a corresponding third generated script.

[0141] The first determination submodule is configured to take the third generated script as the first script.

[0142] In the embodiment, the operations performed by the above modules or units correspond to the steps of the script data processing method of the foregoing embodiments one by one, and will not be described here.

[0143] In some optional implementations of the embodiment, the second processing module 304 includes:

[0144] The identification submodule is configured to identify high-risk expressions existing in the first script based on a preset compliance rule library;

[0145] The replacement submodule is configured to perform expression replacement processing on the high-risk expressions in the first script to obtain a corresponding fourth generated script;

[0146] The judgment submodule is configured to judge whether a key clause exists in the fourth generated script;

[0147] The supplement submodule is configured to, if yes, perform corresponding content supplement processing on the fourth generated script based on the key clause to obtain a corresponding fifth generated script;

[0148] The second determination submodule is configured to take the fifth generated script as the second script.

[0149] In the embodiment, the modules or units are respectively used for operations corresponding to the steps of the script data processing method of the foregoing embodiments, and thus will not be described here.

[0150] In some optional implementations of the embodiment, the third processing module 305 includes:

[0151] The third determination submodule is configured to determine a corresponding accommodated text amount based on the target engine and the requirement information.

[0152] The optimization submodule is configured to perform content optimization on the second script based on the accommodated text amount and using a preset greedy algorithm to obtain a corresponding sixth generated script.

[0153] The deletion submodule is configured to perform redundancy content deletion processing on the sixth generated script based on the accommodated text amount to obtain a corresponding seventh generated script.

[0154] The fourth determination submodule is configured to take the seventh generated script as the target script.

[0155] In the embodiment, the modules or units are respectively used for operations corresponding to the steps of the script data processing method of the foregoing embodiments, and thus will not be described here.

[0156] In some optional implementations of the embodiment, the output module 306 includes:

[0157] The first acquisition submodule is configured to acquire a preset script auditing strategy.

[0158] The auditing submodule is configured to perform auditing processing on the target script based on the script auditing strategy to obtain a corresponding auditing result.

[0159] The second obtaining sub-module is configured to obtain a preset script output mode if the audit result is passed.

[0160] The output sub-module is configured to perform output processing on the target script based on the script output mode.

[0161] In the embodiment, the modules or units are respectively used for performing operations corresponding to the steps of the script data processing method of the foregoing embodiments, and thus will not be described here.

[0162] In some optional implementation forms of the embodiment, the script data processing apparatus further includes:

[0163] The obtaining module is configured to obtain a pre-created dynamic reward function.

[0164] The calling module is configured to call a preset target reinforcement learning algorithm.

[0165] The optimization module is configured to perform optimization processing on a behavior policy of the multiple agents based on the target reinforcement learning algorithm and using the dynamic reward function.

[0166] In the embodiment, the modules or units are respectively used for performing operations corresponding to the steps of the script data processing method of the foregoing embodiments, and thus will not be described here.

[0167] In some optional implementation forms of the embodiment, the obtaining module includes:

[0168] The third obtaining sub-module is configured to obtain a preset positive incentive rule.

[0169] The fourth obtaining sub-module is configured to obtain a preset negative punishment rule.

[0170] The constructing sub-module is configured to perform function constructing processing based on the positive incentive rule and the negative punishment rule, to obtain a corresponding target function.

[0171] The fifth determining sub-module is configured to use the target function as the dynamic reward function.

[0172] In the embodiment, the modules or units are respectively used for performing operations corresponding to the steps of the script data processing method of the foregoing embodiments, and thus will not be described here.

[0173] To solve the foregoing technical problems, an embodiment of the present application further provides a computer device. For details, please refer to Figure 4 , Figure 4 The following is a basic structure block diagram of the computer device of the embodiment.

[0174] The computer device 4 includes a memory 41, a processor 42, and a network interface 43, which are communicatively connected via a system bus. It should be noted that only the computer device 4 with components 41-43 is shown in the figure, but it should be understood that not all of the shown components are required to be implemented, and more or fewer components can be alternatively implemented. Among them, those skilled in the art can understand that the computer device herein is a device capable of automatically performing numerical calculation and / or information processing according to pre-set or stored instructions, and its hardware includes but is not limited to microprocessors, application specific integrated circuits (ASICs), field programmable gate arrays (FPGAs), digital signal processors (DSPs), embedded devices, etc.

[0175] The computer device can be a desktop computer, a notebook computer, a palm computer, a cloud server, and the like. The computer device can interact with the user through a keyboard, a mouse, a remote controller, a touchpad, a voice control device, and the like.

[0176] The memory 41 includes at least one type of readable storage medium, including a flash memory, a hard disk, a multimedia card, a card-type memory (e.g., an SD or DX memory, etc.), a random access memory (RAM), a static random access memory (SRAM), a read-only memory (ROM), an electrically erasable programmable read-only memory (EEPROM), a programmable read-only memory (PROM), a magnetic memory, a magnetic disk, an optical disk, and the like. In some embodiments, the memory 41 can be an internal storage unit of the computer device 4, such as a hard disk or a memory of the computer device 4. In other embodiments, the memory 41 can also be an external storage device of the computer device 4, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, and the like. Of course, the memory 41 can also include both the internal storage unit and the external storage device of the computer device 4. In the present embodiment, the memory 41 is generally used to store an operating system and various application software installed in the computer device 4, such as computer readable instructions of the script data processing method, and the like. In addition, the memory 41 can also be used to temporarily store various data that have been output or will be output.

[0177] The processor 42 may, in some embodiments, be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chip. The processor 42 is generally used to control the overall operation of the computer device 4. In the present embodiment, the processor 42 is configured to run computer-readable instructions stored in the memory 41 or process data, such as computer-readable instructions for running the processing method of the script data.

[0178] The network interface 43 may include a wireless network interface or a wired network interface, and is generally used to establish a communication connection between the computer device 4 and other electronic devices.

[0179] Compared with the prior art, the present application has the following beneficial effects:

[0180] In the present application, first, the theme and requirement information input by the user are received; then, the theme and the requirement information are subjected to script generation processing based on a preset large language model, to obtain a corresponding initial script; thereafter, the initial script is subjected to content adjustment processing based on a preset multi-agent, to obtain a corresponding first script; wherein the multi-agent includes a fact verification agent, a logic reasoning agent, and a compliance review agent; subsequently, the first script is subjected to content enhancement processing based on a preset compliance optimization strategy, to obtain a corresponding second script; further, the second script is subjected to content adaptation processing based on a preset target engine using the requirement information, to obtain a corresponding target script; finally, the target script is subjected to output processing. After receiving the theme and requirement information input by the user, the present application first subjects the theme and requirement information to script generation processing based on the use of a large language model to obtain an initial script, then subjects the initial script to content adjustment processing based on the use of a multi-agent to obtain a first script, thereafter subjects the first script to content enhancement processing based on the use of a compliance optimization strategy to obtain a second script, subsequently subjects the second script to content adaptation processing based on a target engine using the requirement information to obtain a target script, and finally subjects the target script to output processing. After obtaining the initial script generated by the large language model, the present application automatically and intelligently subjects the initial script to content optimization processing based on the collaborative use of a multi-agent, a compliance optimization strategy, and a target engine, effectively optimizes the content generation of the script, and improves the professionalism, accuracy, and compliance of the generated target script.

[0181] The present application also provides another implementation, namely providing a computer-readable storage medium storing computer-readable instructions executable by at least one processor to cause the at least one processor to perform the steps of the processing method of the script data as described above.

[0182] Compared with the prior art, the embodiments of the present application have the following beneficial effects:

[0183] In the embodiments of the present application, first, the theme and requirement information input by the user are received; then, based on a preset large language model, script generation processing is performed on the theme and the requirement information to obtain a corresponding initial script; thereafter, based on a preset multi-agent, content adjustment processing is performed on the initial script to obtain a corresponding first script; wherein, the multi-agent includes a fact verification agent, a logic reasoning agent and a compliance review agent; subsequently, based on a preset compliance optimization strategy, content enhancement processing is performed on the first script to obtain a corresponding second script; further, based on a preset target engine, content adaptation processing is performed on the second script using the requirement information to obtain a corresponding target script; finally, output processing is performed on the target script. After receiving the theme and requirement information input by the user, the present application will first perform script generation processing on the theme and requirement information based on the use of a large language model to obtain an initial script, then perform content adjustment processing on the initial script based on the use of a multi-agent to obtain a first script, thereafter perform content enhancement processing on the first script based on the use of a compliance optimization strategy to obtain a second script, subsequently perform content adaptation processing on the second script using the requirement information based on a target engine to obtain a target script, and finally perform output processing on the target script. After obtaining the initial script generated by the large language model, the present application will automatically and intelligently perform content optimization processing on the initial script based on the collaborative use of the multi-agent, the compliance optimization strategy and the target engine, effectively optimizing the content generation of the script and improving the professionalism, accuracy and compliance of the generated target script.

[0184] From the above description of the embodiments, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software and a general hardware platform as required, and of course, they can also be implemented by hardware, but in many cases, the former is a better embodiment. Based on this understanding, the technical solutions of the present application can be embodied in the form of a software product, which is stored in a storage medium (such as a ROM / RAM, a magnetic disk, an optical disk), and includes a plurality of instructions for causing a terminal device (which can be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) to execute the methods described in the various embodiments of the present application.

[0185] Obviously, the above-described embodiments are only some embodiments but not all the embodiments of the present application, the preferred embodiments of the present application are shown in the drawings, but do not limit the patent scope of the present application. The present application can be implemented in many different forms, and conversely, the purpose of providing these embodiments is to make the disclosure of the present application more thorough and comprehensive. Although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing specific embodiments, or make equivalent replacements to some technical features therein. Any equivalent structure made by using the content of the specification and drawings, directly or indirectly applied to other related technical fields, is also within the patent protection scope of the present application.

Claims

1. A method for processing script data, characterized in that, Includes the following steps: Receive user input regarding the topic and requirements; Based on a pre-defined large language model, the topic and the requirement information are processed to generate a script, resulting in a corresponding initial script. The initial script is adjusted based on a preset multi-agent system to obtain a corresponding first script; wherein the multi-agent system includes a fact verification agent, a logical reasoning agent, and a compliance review agent. The first script is enhanced based on a preset compliance optimization strategy to obtain the corresponding second script; Based on the preset target engine, the second script is adapted using the required information to obtain the corresponding target script. The target script is then processed for output.

2. The method for processing script data according to claim 1, characterized in that, The step of adjusting the content of the initial script based on a preset multi-agent system to obtain the corresponding first script specifically includes: Based on the fact-verification agent, the initial script is fact-verified using a preset insurance rule base, and the content of the initial script is modified accordingly based on the obtained fact-verification results to obtain the corresponding first generated script. Based on the logical reasoning agent, the first generated script is logically verified using a preset verification model, and the corresponding content of the first generated script is modified based on the obtained logical verification result to obtain the corresponding second generated script. Based on the aforementioned compliance review agent, the second script is modified using a preset regulatory rule base to obtain the corresponding third generated script; The third generated script is used as the first script.

3. The method for processing script data according to claim 1, characterized in that, The step of performing content enhancement processing on the first script based on a preset compliance optimization strategy to obtain the corresponding second script specifically includes: Based on a pre-defined compliance rule base, high-risk statements in the first script are identified; The high-risk expressions in the first script are replaced to obtain the corresponding fourth generated script; Determine whether there are any key clauses in the fourth generated script; If so, based on the aforementioned key clauses, the fourth generated script is supplemented with corresponding content to obtain the corresponding fifth generated script; The fifth generated script is used as the second script.

4. The method for processing script data according to claim 1, characterized in that, The step of using the requirement information to perform content adaptation processing on the second script based on a preset target engine to obtain the corresponding target script specifically includes: Based on the target engine, the corresponding amount of text to be accommodated is determined using the requirement information; Based on the amount of text to be accommodated, the second script is optimized using a preset greedy algorithm to obtain the corresponding sixth generated script; Based on the amount of text to be accommodated, redundant content is removed from the sixth generated script to obtain the corresponding seventh generated script. The seventh generated script is used as the target script.

5. The method for processing script data according to claim 1, characterized in that, The step of outputting the target script specifically includes: Obtain the preset script review policy; The target script is reviewed and processed based on the script review strategy to obtain the corresponding review result; If the review result is "approved", then the preset script output method is obtained; The target script is output based on the aforementioned script output method.

6. The method for processing script data according to claim 1, characterized in that, After the step of outputting the target script, the method further includes: Get the pre-created dynamic reward function; Invoke the preset target reinforcement learning algorithm; Based on the target reinforcement learning algorithm, the dynamic reward function is used to optimize the behavioral policies of the multi-agent system.

7. The method for processing script data according to claim 6, characterized in that, The step of obtaining the pre-created dynamic reward function specifically includes: Obtain the preset positive incentive rules; Obtain the preset negative penalty rules; Based on the positive incentive rule and the negative penalty rule, a function construction process is performed to obtain the corresponding objective function; The objective function is used as the dynamic reward function.

8. A script data processing apparatus, characterized in that, include: The receiving module is used to receive the topic and request information input by the user; The generation module is used to perform script generation processing on the topic and the requirement information based on a preset large language model to obtain the corresponding initial script; The first processing module is used to perform content adjustment processing on the initial script based on a preset multi-agent system to obtain a corresponding first script; wherein the multi-agent system includes a fact verification agent, a logical reasoning agent, and a compliance review agent. The second processing module is used to perform content enhancement processing on the first script based on a preset compliance optimization strategy to obtain the corresponding second script; The third processing module is used to perform content adaptation processing on the second script based on the preset target engine and the requirement information to obtain the corresponding target script. The output module is used to process the output of the target script.

9. A computer device, characterized in that, The system includes a memory and a processor, wherein the memory stores computer-readable instructions, and the processor executes the computer-readable instructions to implement the steps of the script data processing method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-readable instructions, which, when executed by a processor, implement the steps of the script data processing method as described in any one of claims 1 to 7.

Images