A method, apparatus, electronic device, and storage medium for detecting reach copy.

Abstract

This disclosure provides a method, apparatus, electronic device, and storage medium for detecting outreach text, belonging to the field of internet technology. The method includes: periodically acquiring target outreach text according to a preset outreach template; acquiring pre-constructed verification rules for multiple detection dimensions, and current pre-defined information for each detection dimension; constructing target prompt words based on the target outreach text, the verification rules for each detection dimension, and the pre-defined information; and calling a pre-trained language model to perform content analysis on the target outreach text based on the target prompt words, thereby determining the detection result of the target outreach text. Using this disclosure can improve detection accuracy.

Description

Technical Field

[0001] This invention relates to the field of Internet technology, and in particular to a method, apparatus, electronic device, and storage medium for detecting reach messages. Background Technology

[0002] Online outreach messages (such as push ads, SMS notifications, pop-up notifications, etc.) are an important way to push messages to users in the internet industry.

[0003] Currently, the main approach to detecting the reasonableness of online outreach copy in the internet industry is to use a sensitive word regular expression matching scheme. This scheme uses a pre-set sensitive word library and regular expressions to filter keywords, and performs mechanical matching by setting prohibited word lists and industry-restricted word lists.

[0004] However, the sensitive word regular expression matching scheme lacks semantic understanding capabilities, cannot identify the true meaning of words in the context, and has difficulty detecting hidden violations, resulting in low detection accuracy. Summary of the Invention

[0005] In view of this, embodiments of the present disclosure provide a method, apparatus, electronic device, and storage medium for detecting reachable text, so as to improve the accuracy of detection.

[0006] According to one aspect of this disclosure, a method for detecting outreach copy is provided, the method comprising: Based on the preset outreach template, periodically obtain the target outreach text; Obtain the pre-constructed verification rules for multiple detection dimensions, and the current prior information for each detection dimension; Based on the target reach text, the verification rules for each detection dimension, and the prior information, target prompt words are constructed; A pre-trained language model is invoked to perform content analysis on the target reach text based on the target prompt words, and the detection result of the target reach text is determined.

[0007] According to another aspect of this disclosure, a device for detecting reach text is provided, the device comprising: The acquisition unit is used to periodically acquire target outreach text according to a preset outreach template; acquire verification rules for multiple pre-built detection dimensions, and current pre-existing information for each detection dimension; The prompt word construction unit is used to construct target prompt words based on the target reach text, the verification rules for each of the detection dimensions, and the prior information; The model detection unit is used to call a pre-trained language model, perform content analysis on the target reach text based on the target prompt words, and determine the detection result of the target reach text.

[0008] According to another aspect of this disclosure, an electronic device is provided, comprising: Processor; and Stored program memory, The aforementioned program includes instructions that, when executed by the aforementioned processor, cause the processor to perform the aforementioned method for detecting the reach message.

[0009] According to another aspect of this disclosure, a non-transitory computer-readable storage medium storing computer instructions is provided, wherein the computer instructions are used to cause a computer to execute the aforementioned method for detecting the access message.

[0010] In this disclosure, target outreach text is periodically acquired based on a preset outreach template. Then, based on pre-built verification rules across multiple detection dimensions, current contextual information, and the target outreach text, target prompts are constructed to perform rationality checks on the target outreach text. A pre-trained language model is then invoked to perform content analysis on the target outreach text based on these prompts, determining the detection result. Furthermore, the semantic analysis capabilities of the pre-trained language model are applied to outreach text detection, improving accuracy. Finally, by associating the outreach content with the preset outreach template, automated detection of online outreach text is achieved. Attached Figure Description

[0011] Further details, features, and advantages of this disclosure are disclosed in the following description of exemplary embodiments in conjunction with the accompanying drawings, in which: Figure 1 A flowchart illustrating a method for detecting reach text according to an exemplary embodiment of this disclosure is shown; Figure 2 A schematic block diagram of a device for detecting reach text provided according to an exemplary embodiment of the present disclosure is shown; Figure 3 A structural block diagram of an exemplary electronic device that can be used to implement embodiments of the present disclosure is shown. Detailed Implementation

[0012] Embodiments of this disclosure will now be described in more detail with reference to the accompanying drawings. While some embodiments of this disclosure are shown in the drawings, it should be understood that this disclosure can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of this disclosure. It should be understood that the accompanying drawings and embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of protection of this disclosure.

[0013] It should be understood that the steps described in the method embodiments of this disclosure may be performed in different orders and / or in parallel. Furthermore, the method embodiments may include additional steps and / or omit the steps shown. The scope of this disclosure is not limited in this respect.

[0014] The term "comprising" and its variations as used herein are open-ended, meaning "including but not limited to". The term "based on" means "at least partially based on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments". Definitions of other terms will be given in the description below. It should be noted that the concepts of "first", "second", etc., used in this disclosure are only used to distinguish different devices, modules, or units, and are not intended to limit the order of functions performed by these devices, modules, or units or their interdependencies.

[0015] It should be noted that the terms "one" and "more" used in this disclosure are illustrative rather than restrictive, and those skilled in the art should understand that, unless otherwise expressly indicated in the context, they should be understood as "one or more".

[0016] The names of messages or information exchanged between multiple devices in the embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of such messages or information.

[0017] This disclosure provides a method for detecting reach text, which can be performed by a terminal, a server, and / or other devices with processing capabilities. The method provided in the embodiments of this disclosure can be performed by any of the aforementioned devices, or by multiple devices working together; this disclosure does not limit this approach.

[0018] The following will refer to Figure 1 The flowchart shown illustrates the method for detecting reach copy, and the method is described below. The method includes the following steps 101-104.

[0019] Step 101: Periodically obtain target outreach text based on the preset outreach template.

[0020] In one possible implementation, the system can periodically scan the business database for pre-set outreach templates (including SMS templates, official account message templates, pop-up templates, etc.) using a pre-defined automated scheduling module, covering all online outreach channels. It then dynamically retrieves a real-time dataset of text to be detected based on these pre-set outreach templates, using each text as the target outreach text to be detected. The acquired text content is preprocessed: the user's real information is hidden, and different verification rules are matched based on dynamic parameters such as different limits and dates.

[0021] Step 102: Obtain the pre-built verification rules for multiple detection dimensions, and the current background information for each detection dimension.

[0022] The detection dimensions can include at least: financial figures, date figures, contact information figures, signature text, legal compliance text, culturally sensitive text, misleading statements, and logical consistency. The prerequisite information for financial figures includes: legal annualized interest rate and legal daily interest rate; the prerequisite information for date figures includes: the current date.

[0023] In one possible implementation, an online outreach copy library can store a large number of outreach copy pieces. These can be analyzed for their reasonableness, and historical experience in reasonableness judgment can be combined to summarize characteristics prone to reasonableness issues. These characteristics can be categorized into types, and multiple detection dimensions can be constructed based on these categorized types. For example, when numbers appear in the outreach copy, the focus should be on whether the value of the number is reasonable and legal: numbers representing amounts cannot be less than 0, numbers representing credit limits cannot be non-integer hundreds, numbers representing days or minutes must be positive integers, and numbers representing annual / daily interest rates must be within legal limits, etc. When dates appear in the copy, the focus should be on whether the date format is correct and whether the date is earlier than the current date, etc. When SMS signatures appear in the copy, it should be checked whether the signature represents the company, and when official customer service phone numbers appear in the copy, it should be checked whether the provided phone number is a genuine official customer service number, etc.

[0024] Furthermore, at least one verification rule can be set for each detection dimension based on the characteristics of the reasonableness issues obtained from the analysis. For example, for financial numbers, verification rules may include: verification of amount greater than 0, verification of limit greater than 0, verification of missing limit, verification of limit as a whole hundred, etc.; for date numbers, verification rules may include: date format verification, verification of expiration date later than the current date, etc.

[0025] To improve the accuracy of plausibility detection, pre-trained language models can be provided with prior information related to plausibility detection for each detection dimension. This prior information may change over time, and the system can acquire the latest prior information through engineering capabilities.

[0026] Step 103: Based on the target reach copy and the verification rules and prerequisite information for each detection dimension, construct the target prompt words.

[0027] In one possible implementation, a cue word template can be pre-set to provide sufficient context for the pre-trained language model, helping it understand the background of the problem or task. The content of the cue word template may include task definition, role assignment, and output specification. The task definition is used to indicate the task of performing a reasonableness check on the target message, the role assignment is used to assign the pre-trained language model the role of a professional technician for checking the message (i.e., indicating the technical field of the target message), and the output specification is used to indicate the content and format of the output.

[0028] By combining the target message, the validation rules for each detection dimension, and the prerequisite information into the prompt word template, you can obtain the target prompt words used to detect the target message.

[0029] Optionally, target cue words can be used to indicate the rule score for each validation rule in the target reach copy, including: If the target message does not meet the detection type of the current validation rule, then the rule score of the current validation rule is the first score; If the target message matches the detection type of the current validation rule and passes the validation rule, then the rule score of the current validation rule is the second score. If the target message matches the detection type of the current validation rule but fails the validation rule, then the rule score of the current validation rule is the third score.

[0030] As a concrete example, the standardized output of the target prompts can indicate the rule score for each validation rule in the target reach copy, and output the final score. The specific content could be as follows: "You need to calculate the score of the SMS content according to the following principles: Initialize the score variable score=0, and validate the SMS content one by one according to the rules in the rule enumeration. If the text in the SMS involves the relevant rule type, it is considered to have hit the rule; otherwise, it is considered to have missed the rule. If the rule is missed: score-0; if the rule is hit and the validation passes: score-0; if the rule is hit but the validation fails: score-1."

[0031] I will give you a text message, some prerequisite knowledge, and a list of rules. Please iterate through the rules in the rule enumeration, analyze the text message content one by one, list the validation result of each rule, and list the current score value for each rule. The specific values ​​of the first score, the second score, and the third score can be set according to actual needs, and this embodiment does not limit them.

[0032] Step 104: Call the pre-trained language model to perform content analysis on the target reach text based on the target prompt words, and determine the detection results of the target reach text.

[0033] Among them, pre-trained language models can refer to ultra-large-scale pre-trained language models based on deep learning (such as GPT, BERT, etc.), which have cross-domain semantic understanding, contextual reasoning and text generation capabilities, and the parameter scale usually exceeds 1 billion.

[0034] In one possible implementation, the API (Application Program Interface) of the pre-trained language model can be invoked, the target prompt words can be input into the pre-trained language model, the model can be asked to perform the task of analyzing the content, and the output format of the model can be limited to integral system.

[0035] The processing of the detection results for determining the target reach copy may include: Determine the final score after detecting the target reach copy based on each verification rule; If the final score is lower than the preset threshold, the target message will be marked as abnormal. Generate alerts for abnormal text, which serve as the detection results for the target text.

[0036] In one possible implementation, after the pre-trained language model outputs the final score of the target outreach copy, the system can perform a threshold judgment on the final score. If the score is greater than or equal to a preset threshold, it indicates that the target outreach copy has passed the reasonableness check. If the final score is lower than the preset threshold, it indicates that the target outreach copy has failed the reasonableness check and can be marked as abnormal copy.

[0037] Optionally, the processing of the alarm information for generating abnormal text may include: The rule scoring indicates at least one target rule and its content that is deemed unacceptable. Retrieve the exception fields corresponding to the target rule from the exception text; Based on at least one target rule and its rule content, abnormal fields, final score, abnormal text content, and the template identifier of the corresponding preset reach template, generate alarm information for abnormal text.

[0038] In one possible implementation, since target prompts can be used to indicate the rule score for each validation rule in the target reach text, when an abnormal text is detected, the pre-trained language model can retain the number of the failed target rule and its rule content, as well as the abnormal fields in the target reach text that failed the detection.

[0039] Based on the aforementioned failed target rules and their content, determine the error type of the abnormal field. Obtain the template identifier of the preset outreach template corresponding to the abnormal text. Assemble the error type, abnormal field, final score, text content of the abnormal text, and the template identifier of the corresponding preset outreach template to generate an alarm message containing these elements.

[0040] Subsequently, the business affiliation of the abnormal text can be determined based on the alarm information, and the alarm information can be pushed to the relevant personnel responsible for that business.

[0041] In one possible implementation, the system can interface with an enterprise monitoring platform (such as Lark, DingTalk, etc.). Based on the aforementioned alarm information (such as template identifiers), the system can query the business affiliation and responsible person of the preset reach template in the enterprise monitoring platform. The alarm information is then sent as a push message to the relevant responsible person, facilitating targeted modifications.

[0042] Optionally, after marking the target message as abnormal, the following processing can also be performed: If the abnormal text is determined to be a false alarm, the reason for the false alarm is obtained based on the target prompt words analysis; The target prompts and validation rules were adjusted based on the reasons for false positives.

[0043] In one possible implementation, after detecting abnormal copy, the accuracy of the detection result for the target reach copy can be checked. If the detection result is determined to be inaccurate, i.e., a false alarm, the target prompt words can be analyzed to determine the cause of the false alarm. Furthermore, based on the cause of the false alarm, the target prompt words and validation rules can be adjusted in a targeted manner to continuously optimize the prompt words and validation rules, achieving positive feedback.

[0044] This embodiment can achieve the following beneficial effects: (1) Based on the preset outreach template, target outreach texts are periodically acquired. Then, based on the pre-built verification rules for multiple detection dimensions, the current prior information, and the target outreach text, target prompts are constructed to perform rationality checks on the target outreach texts. This allows the pre-trained language model to be invoked to perform content analysis on the target outreach texts based on the target prompts, thus determining the detection results. On this basis, the semantic analysis capabilities of the pre-trained language model are applied to the outreach text detection field, improving the accuracy of the detection. Furthermore, by associating outreach content with the preset outreach template, automated detection of online outreach texts is achieved.

[0045] (2) Whenever an abnormal text is detected, the corresponding alarm information can be pushed to the relevant personnel.

[0046] (3) Whenever a false alarm occurs, the target prompt words and verification rules can be adjusted based on the cause of the false alarm to further improve the accuracy of detection.

[0047] This disclosure provides an apparatus for detecting outreach text, which is used to implement the aforementioned method for detecting outreach text. For example... Figure 2 As shown, the detection device 200 for reaching text includes: an acquisition unit 201, a prompt word construction unit 202, and a model detection unit 203.

[0048] The acquisition unit 201 is used to periodically acquire target outreach text according to a preset outreach template; acquire verification rules for multiple pre-built detection dimensions, and current pre-existing information for each detection dimension; The prompt word construction unit 202 is used to construct target prompt words based on the target reach text, the verification rules of each detection dimension, and the pre-existing information; The model detection unit 203 is used to call a pre-trained language model to perform a content analysis task on the target reach text based on the target prompt words, and determine the detection result of the target reach text.

[0049] Optionally, the model detection unit 203 is used for: Determine the final score after detecting the target reach copy based on each of the aforementioned verification rules; If the final score is lower than a preset threshold, the target message will be marked as abnormal. An alarm message for the abnormal text is generated, which serves as the detection result of the target reach text.

[0050] Optionally, the model detection unit 203 is further configured to: Based on the alarm information, the business affiliation of the abnormal text is determined, and the alarm information is pushed to the relevant personnel of the business affiliation.

[0051] Optionally, the target prompt words are used to indicate the rule score for each of the verification rules for the target reach copy; The model detection unit 203 is used for: The rule score indicates at least one target rule and its content that are deemed unqualified. Obtain the exception field corresponding to the target rule from the exception text; Based on at least one target rule and its rule content, the abnormal field, the final score, the text content of the abnormal text, and the template identifier of the corresponding preset reach template, an alarm message for the abnormal text is generated.

[0052] Optionally, the target prompt words are used to instruct the determination of a rule score for each of the validation rules for the target reach copy, including: If the target message does not conform to the detection type of the current verification rule, then the rule score of the current verification rule is the first score; If the target message matches the detection type of the current verification rule and passes the verification rule, then the rule score of the current verification rule is the second score. If the target message matches the detection type of the current verification rule but fails the verification rule, then the rule score of the current verification rule is the third score.

[0053] Optionally, the model detection unit 203 is further configured to: If the abnormal text is determined to be a false alarm, the reason for the false alarm is obtained based on the analysis of the target prompt words; The target prompt words and the verification rules are adjusted based on the reasons for the false alarms.

[0054] Optionally, the detection dimensions include at least: financial numbers, date numbers, contact information numbers, signature text, legal compliance text, culturally sensitive text, misleading statements, and logical consistency; The preceding information corresponding to the financial figures includes: legal annualized interest rate and legal daily interest rate; The preceding information corresponding to the date-type numbers includes: the current date.

[0055] This embodiment can achieve the following beneficial effects: Based on a pre-set outreach template, target outreach texts are periodically acquired. Using pre-built validation rules across multiple detection dimensions, current contextual information, and the target outreach text, target prompts are constructed to perform rationality checks on the outreach texts. A pre-trained language model is then invoked to perform content analysis on the target outreach texts based on these prompts, determining the detection results. Furthermore, the semantic analysis capabilities of the pre-trained language model are applied to outreach text detection, improving accuracy. By associating outreach content with pre-set outreach templates, automated online outreach text detection is achieved.

[0056] Exemplary embodiments of this disclosure also provide an electronic device, including: at least one processor; and a memory communicatively connected to the at least one processor. The memory stores a computer program executable by the at least one processor, the computer program being executed by the at least one processor to cause the electronic device to perform a method according to an embodiment of this disclosure.

[0057] Exemplary embodiments of this disclosure also provide a non-transitory computer-readable storage medium storing a computer program, wherein the computer program, when executed by a computer's processor, is used to cause the computer to perform a method according to embodiments of this disclosure.

[0058] Exemplary embodiments of this disclosure also provide a computer program product, including a computer program, wherein, when executed by a processor of a computer, the computer program is used to cause the computer to perform a method according to an embodiment of this disclosure.

[0059] refer to Figure 3 The present invention describes a structural block diagram of an electronic device 300 that can serve as a server or client of the present disclosure, which is an example of a hardware device that can be applied to various aspects of the present disclosure. The electronic device is intended to represent various forms of digital electronic computer devices, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0060] like Figure 3 As shown, the electronic device 300 includes a computing unit 301, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 302 or a computer program loaded from a storage unit 308 into a random access memory (RAM) 303. The RAM 303 may also store various programs and data required for the operation of the device 300. The computing unit 301, ROM 302, and RAM 303 are interconnected via a bus 304. An input / output (I / O) interface 305 is also connected to the bus 304.

[0061] Multiple components in electronic device 300 are connected to I / O interface 305, including: input unit 306, output unit 307, storage unit 308, and communication unit 309. Input unit 306 can be any type of device capable of inputting information to electronic device 300. Input unit 306 can receive input digital or text information and generate key signal inputs related to user settings and / or function control of electronic device. Output unit 307 can be any type of device capable of presenting information and may include, but is not limited to, a display, speaker, video / audio output terminal, vibrator, and / or printer. Storage unit 308 may include, but is not limited to, disk and optical disk. Communication unit 309 allows electronic device 300 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks, and may include, but is not limited to, modems, network cards, infrared communication devices, wireless communication transceivers, and / or chipsets, such as Bluetooth devices, Wi-Fi devices, Wi-Fi devices, cellular communication devices, and / or the like.

[0062] The computing unit 301 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 301 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 301 performs the various methods and processes described above. For example, in some embodiments, the above-described method for detecting the reach message can be implemented as a computer software program tangibly contained in a machine-readable medium, such as storage unit 308. In some embodiments, part or all of the computer program can be loaded and / or installed on the electronic device 300 via ROM 302 and / or communication unit 309. In some embodiments, the computing unit 301 can be configured to perform the above-described method for detecting the reach message by any other suitable means (e.g., by means of firmware).

[0063] The program code used to implement the methods of this disclosure may be written in any combination of one or more programming languages. This program code may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing apparatus, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code may be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0064] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. Machine-readable media can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0065] As used in this disclosure, the terms "machine-readable medium" and "computer-readable medium" refer to any computer program product, device, and / or apparatus (e.g., disk, optical disk, memory, programmable logic device (PLD)) for providing machine instructions and / or data to a programmable processor, including machine-readable media that receive machine instructions as machine-readable signals. The term "machine-readable signal" refers to any signal for providing machine instructions and / or data to a programmable processor.

[0066] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0067] The systems and technologies described herein can be implemented in computing systems that include back-end components (e.g., as a data server), or computing systems that include middleware components (e.g., an application server), or computing systems that include front-end components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with implementations of the systems and technologies described herein), or any combination of such back-end, middleware, or front-end components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., a communication network). Examples of communication networks include local area networks (LANs), wide area networks (WANs), and the Internet.

[0068] Computer systems can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. Client-server relationships are created by computer programs running on the respective computers and having a client-server relationship with each other.

Claims

1. A method for detecting reach copy, characterized in that, The method includes: Based on the preset outreach template, periodically obtain the target outreach text; Obtain the pre-constructed verification rules for multiple detection dimensions, and the current prior information for each detection dimension; Based on the target reach text, the verification rules for each detection dimension, and the prior information, target prompt words are constructed; A pre-trained language model is invoked to perform content analysis on the target reach text based on the target prompt words, and the detection result of the target reach text is determined.

2. The method according to claim 1, characterized in that, The determination of the detection result of the target reach copy includes: Determine the final score after detecting the target reach copy based on each of the aforementioned verification rules; If the final score is lower than a preset threshold, the target message will be marked as abnormal. An alarm message for the abnormal text is generated, which serves as the detection result of the target reach text.

3. The method according to claim 2, characterized in that, After generating the alarm information for the abnormal text, the method further includes: Based on the alarm information, the business affiliation of the abnormal text is determined, and the alarm information is pushed to the relevant personnel of the business affiliation.

4. The method according to claim 2, characterized in that, The target prompt words are used to indicate the rule score for each of the verification rules for the target reach text; The alarm information for generating the abnormal text includes: Determining the rule score indicates at least one target rule and its content that are deemed unacceptable; Obtain the exception field corresponding to the target rule from the exception text; Based on at least one target rule and its rule content, the abnormal field, the final score, the text content of the abnormal text, and the template identifier of the corresponding preset reach template, an alarm message for the abnormal text is generated.

5. The method according to claim 1, characterized in that, The target prompt words are used to instruct the determination of a rule score for each of the validation rules for the target reach copy, including: If the target message does not conform to the detection type of the current verification rule, then the rule score of the current verification rule is the first score; If the target message matches the detection type of the current verification rule and passes the verification rule, then the rule score of the current verification rule is the second score. If the target message matches the detection type of the current verification rule but fails the verification rule, then the rule score of the current verification rule is the third score.

6. The method according to claim 2, characterized in that, After marking the target message as abnormal, the method further includes: If the abnormal text is determined to be a false alarm, the reason for the false alarm is obtained based on the analysis of the target prompt words; The target prompt words and the verification rules are adjusted based on the reasons for the false alarms.

7. The method according to claim 1, characterized in that, The detection dimensions include at least: financial numbers, date numbers, contact information numbers, signature text, legal compliance text, culturally sensitive text, misleading statements, and logical consistency; The preceding information corresponding to the financial figures includes: legal annualized interest rate and legal daily interest rate; The preceding information corresponding to the date-type numbers includes: the current date.

8. A device for detecting reach text, characterized in that, The device includes: The acquisition unit is used to periodically acquire target outreach text according to a preset outreach template; acquire verification rules for multiple pre-built detection dimensions, and current pre-existing information for each detection dimension; The prompt word construction unit is used to construct target prompt words based on the target reach text, the verification rules for each of the detection dimensions, and the prior information; The model detection unit is used to call a pre-trained language model, perform content analysis on the target reach text based on the target prompt words, and determine the detection result of the target reach text.

9. An electronic device, comprising: processor; as well as Stored program memory, The program includes instructions that, when executed by the processor, cause the processor to perform the method according to any one of claims 1-7.

10. A non-transitory computer-readable storage medium storing computer instructions, wherein, The computer instructions are used to cause the computer to perform the method according to any one of claims 1-7.

Images