Data processing method, computing device, computer readable storage medium and program product

Abstract

Embodiments of the invention provide a data processing method, a computing device, a computer readable storage medium and a program product. The data processing method comprises the steps of obtaining a to-be-processed short message and a prompt text corresponding to the to-be-processed short message; the to-be-processed short message and the prompt text are input into a short message evaluation model, an evaluation report output by the short message evaluation model is obtained, and the evaluation report comprises risk grades for grading the to-be-processed short message according to an auditing rule in the prompt text; according to the risk level in the evaluation report, determining an audit result corresponding to the to-be-processed short message; according to the technical scheme, the labor cost of overall auditing is remarkably reduced, the auditing efficiency is improved, the risk of missed auditing and false auditing caused by artificial fatigue can be reduced, and optimal balance of cost, efficiency and accuracy is achieved.

Description

Technical Field

[0001] The embodiments in this specification relate to the field of computer technology, and in particular to a data processing method. Background Technology

[0002] In the medical field, medical text messages must meet regulatory requirements. Therefore, text message service providers need to review the text messages submitted by merchants. When reviewing based on keywords, it is easy to miss the review due to untimely keyword updates. On the other hand, when reviewing by humans, there are problems such as high cost, low efficiency, and missed reviews due to fatigue. Summary of the Invention

[0003] In view of the above, embodiments of this specification provide a data processing method. One or more embodiments of this specification also relate to a data processing apparatus, a computing device, a computer-readable storage medium, and a computer program product, to address the technical deficiencies existing in the prior art.

[0004] According to a first aspect of the embodiments of this specification, a data processing method is provided, comprising: Obtain the pending SMS messages and the corresponding prompt text for the pending SMS messages; The pending SMS message and the prompt text are input into the SMS evaluation model to obtain the evaluation report output by the SMS evaluation model. The evaluation report includes the risk level of the pending SMS message based on the review rules in the prompt text. Based on the risk level stated in the assessment report, the review result corresponding to the pending SMS message is determined.

[0005] According to a second aspect of the embodiments of this specification, a data processing apparatus is provided, comprising: The SMS acquisition module is configured to acquire the SMS messages to be processed and the corresponding prompt text of the SMS messages to be processed; The SMS evaluation module is configured to input the SMS to be processed and the prompt text into the SMS evaluation model to obtain an evaluation report output by the SMS evaluation model. The evaluation report includes a risk level classification of the SMS to be processed based on the review rules in the prompt text. The result acquisition module is configured to determine the review result corresponding to the SMS message to be processed based on the risk level in the assessment report.

[0006] According to a third aspect of the embodiments of this specification, a computing device is provided, comprising: Memory and processor; The memory is used to store computer programs / instructions, and the processor is used to execute the computer programs / instructions, which, when executed by the processor, implement the steps of the above-described data processing method.

[0007] According to a fourth aspect of the embodiments of this specification, a computer-readable storage medium is provided that stores a computer program / instructions that, when executed by a processor, implement the steps of the data processing method described above.

[0008] According to a fifth aspect of the embodiments of this specification, a computer program product is provided, including a computer program / instructions that, when executed by a processor, implement the steps of the above-described data processing method.

[0009] This specification provides a data processing method in one embodiment, which acquires a text message to be processed and its corresponding prompt text. By determining the prompt text corresponding to the text message to be processed, targeted guidance is provided for subsequent review. The text message to be processed and the prompt text are input into a text message evaluation model to obtain an evaluation report output by the model. Natural language processing based on the text message evaluation model enables semantic understanding of the text message to be processed, thereby avoiding deep-seated violations that may be missed by human reviewers due to fatigue or negligence. Based on the risk level in the evaluation report, the review result corresponding to the text message to be processed is determined. This method significantly reduces the overall labor cost of review, improves review efficiency, and reduces the risk of missed or incorrect reviews due to human fatigue, achieving an optimal balance between cost, efficiency, and accuracy. Attached Figure Description

[0010] Figure 1 This is a schematic diagram illustrating a data processing method provided in one embodiment of this specification. Figure 2 This is a flowchart illustrating a data processing method provided in one embodiment of this specification; Figure 3 This is a flowchart illustrating the processing procedure of a data processing method provided in one embodiment of this specification. Figure 4 This is a schematic diagram of the structure of a data processing apparatus provided in one embodiment of this specification; Figure 5 This is a structural block diagram of a computing device provided in one embodiment of this specification. Detailed Implementation

[0011] Many specific details are set forth in the following description to provide a full understanding of this specification. However, this specification can be implemented in many other ways than those described herein, and those skilled in the art can make similar extensions without departing from the spirit of this specification. Therefore, this specification is not limited to the specific implementations disclosed below.

[0012] The terminology used in one or more embodiments of this specification is for the purpose of describing particular embodiments only and is not intended to be limiting of the one or more embodiments of this specification. The singular forms “a,” “described,” and “the” as used in one or more embodiments of this specification and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used in one or more embodiments of this specification refers to and includes any or all possible combinations of one or more associated listed items.

[0013] It should be understood that although the terms first, second, etc., may be used to describe various information in one or more embodiments of this specification, such information should not be limited to these terms. These terms are only used to distinguish information of the same type from one another. For example, first may also be referred to as second without departing from the scope of one or more embodiments of this specification, and similarly, second may also be referred to as first. Depending on the context, the word "if" as used herein may be interpreted as "when," "when," or "in response to a determination."

[0014] Furthermore, it should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in one or more embodiments of this specification are all information and data authorized by the user or fully authorized by all parties. Moreover, the collection, use and processing of related data must comply with the relevant laws, regulations and standards of the relevant countries and regions, and corresponding operation entry points are provided for users to choose to authorize or refuse.

[0015] This specification provides a data processing method, and also relates to a data processing apparatus, a computing device, a computer-readable storage medium, and a computer program product, which will be described in detail in the following embodiments.

[0016] See Figure 1 , Figure 1 A schematic diagram of a data processing method according to an embodiment of this specification is shown.

[0017] Specifically, the data processing method is applied to a data processing system, which includes an end-side device 102 and a server 104. The end-side device 102 is used to send an SMS verification request to the server 104, and the SMS verification request carries an SMS to be processed.

[0018] A text message evaluation model is trained in server 104. Server 104 acquires text messages to be processed and corresponding prompt text. The text messages to be processed and the prompt text are input into the text message evaluation model to obtain an evaluation report output by the model. The evaluation report includes a risk level classification of the text messages to be processed based on the review rules in the prompt text. The review result corresponding to the text messages to be processed is determined based on the risk level in the evaluation report. The review result is then returned to the terminal device 102.

[0019] The edge device 102 may include a browser, an app (application), or a web application such as an H5 (Hypertext Markup Language 5) application, a lightweight application (also known as a mini-program), or a cloud application. The edge device can be developed based on a software development kit (SDK) provided by the server, such as a real-time communication (RTC) SDK. The edge device can be deployed in an electronic device and depends on the device's operation or certain apps within the device to run. The electronic device may have a display screen and support information browsing, such as a personal mobile terminal like a mobile phone, tablet, or personal computer. Various other types of applications can also be configured in the electronic device, such as human-computer interaction applications, model training applications, data processing applications, web browser applications, shopping applications, search applications, instant messaging tools, email clients, and social media platform software.

[0020] Server 104 can be understood as a server providing various services, including physical servers and cloud servers. Examples include servers providing communication services to multiple clients, servers supporting backend training of models used on clients, and servers processing data sent by clients. It's important to note that Server 104 can be implemented as a distributed server cluster composed of multiple servers, or as a single server. Server 104 can also be a server in a distributed system, or a server integrated with blockchain. Server 104 can also be a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms, or an intelligent cloud computing server or intelligent cloud host with artificial intelligence technology.

[0021] See Figure 2 , Figure 2 A flowchart of a data processing method provided in one embodiment of this specification is shown, which specifically includes the following steps.

[0022] Step 202: Obtain the SMS message to be processed and the corresponding prompt text for the SMS message to be processed.

[0023] Among them, the pending SMS messages can be understood as the original text information that needs to be reviewed and evaluated. In this embodiment, the pending SMS messages are SMS templates or specific SMS content; the prompt text can be understood as the text that guides the subsequent SMS evaluation model to conduct targeted review of the pending SMS messages.

[0024] Specifically, the strictness of the review rules may vary in different prompt texts. Therefore, when a text message to be processed is identified, the prompt text that matches the text message to be processed should be determined.

[0025] In one or more embodiments of this specification, the SMS message to be processed is obtained through an SMS review request sent by the client. In practical applications, the data processing method provided in the embodiments of this specification is applied to an SMS review platform. The user sends an SMS review request to the SMS review platform through the client, and the SMS content can only be sent after the review is approved. The specific implementation method is as follows: In response to an SMS review request sent by the client, the pending SMS message carried in the SMS review request is obtained; In this context, an SMS review request can be understood as a request initiated by the client that contains SMS messages to be processed; the client can be understood as an external system or application that needs to conduct SMS compliance review, such as a merchant client or a content publishing platform.

[0026] Specifically, by receiving SMS review requests from clients, the system obtains the original SMS content to be reviewed (i.e., SMS to be processed), and further matches the corresponding prompt text based on the SMS to be processed, thereby providing review rule guidance for subsequent automated review.

[0027] In one or more embodiments of this specification, when matching corresponding prompt text to a text message to be processed, entity recognition is first performed on the text message to be processed to obtain the target entities contained in the text message, and then prompt text is matched according to the category corresponding to the target entity. Specific implementation methods are described below: Obtain the pending SMS messages and the corresponding prompt text for the pending SMS messages, including: Entity extraction is performed on the SMS message to be processed to obtain the target entity; The target entity is classified, and the prompt text corresponding to the SMS to be processed is determined based on the classification result. The prompt text contains the review rules corresponding to the classification result, and the strictness of the review rules varies for different classification results.

[0028] Entity extraction can be understood as automatically identifying and extracting information units with specific semantics from unstructured text messages to be processed. Target entities can be understood as information units extracted from text messages to be processed. In this embodiment, entities can be drug names, medical terms, amounts, institution names, etc., without limitation.

[0029] The classification result can be understood as the result of classifying the extracted target entities according to a predefined category system. For example, the predefined category system covers categories such as "prescription drugs", "non-prescription drugs", "health products", "financial services", and "general product promotion". Therefore, the classification result is determined from this predefined category system.

[0030] For example, when a client sends a text message review request about the promotion of drug A, the pending text message carried in the text message review request is obtained; through entity extraction, it is determined that the target entity of the pending text message includes the name of drug A; by classifying the drug name as an entity, since drug A is a prescription drug, the classification result is determined to be the "prescription drug" category.

[0031] For drug advertisements in the "prescription drug" category, matching or generating corresponding prompt text can be used. This prompt text may include specific review rules and guidelines such as "It is necessary to focus on checking whether there are efficacy assertions or absolute terms" or "It is necessary to verify the indications in the drug instructions."

[0032] In practice, when a text message to be processed contains multiple target entities, the category of the text message to be processed is determined by comprehensively considering the classification results of the multiple target entities, and then the corresponding prompt text is determined based on the category of the text message to be processed.

[0033] Because the strictness of the review rules varies in different notification texts, determining an appropriate notification text for pending SMS messages allows for a reasonable level of review rigor. For example, pending SMS messages related to pharmaceuticals require stricter review rules, demanding compliance not only with advertising regulations but also with scrutiny of the drug's efficacy descriptions (e.g., avoiding absolute terms). Conversely, pending SMS messages promoting "ordinary products" only require compliance with advertising regulations.

[0034] The data processing method provided in the embodiments of this specification transforms the original text messages to be processed into structured target entities through entity extraction and classification. Based on the category to which the target entity belongs, the corresponding review rules are dynamically configured to achieve intelligent association and accurate matching between the review rules and the text messages to be processed, thereby greatly improving the targeting of the review and ensuring that the appropriate review rules can be used for reasonable review when the text messages to be processed are analyzed in the future.

[0035] Step 204: Input the SMS message to be processed and the prompt text into the SMS evaluation model to obtain the evaluation report output by the SMS evaluation model. The evaluation report includes the risk level of the SMS message to be processed based on the review rules in the prompt text.

[0036] Among them, the SMS assessment model can be understood as a large-scale model that has been trained and has the ability to understand natural language and detect compliance; the assessment report can be understood as the assessment content output by the SMS assessment model after comprehensive analysis and review of the SMS to be processed; the risk level can be understood as the level of the SMS to be processed after review and judgment based on the review rules in the prompt text, according to the severity of the violation.

[0037] Specifically, once the prompt text is determined, the SMS to be processed and the prompt text are sent to the SMS evaluation model for in-depth analysis. This SMS evaluation model not only understands the semantics of the SMS to be processed, but also combines the review rules in the prompt text to evaluate the compliance of the SMS to be processed, and quantifies the severity of the violation into a risk level.

[0038] For example, regarding a pending text message containing "Drug A," the review rules in the prompt text clearly state that "efficacy claims" require strict scrutiny. When analyzing the content of the pending text message, "For nourishing Yin and tonifying the kidneys, choose classic options; take Drug A," the text message assessment model identified that "For nourishing Yin and tonifying the kidneys, choose classic options" involves claims of drug efficacy. According to relevant advertising regulations, drug advertisements must not contain claims or guarantees of efficacy. Based on the severity of the violation, the text message assessment model classified its risk level as "medium risk" in the assessment report.

[0039] The data processing method provided in the embodiments of this specification, the SMS evaluation model based on the review rules in the prompt text can adapt to the review needs of different fields and different levels of strictness, thereby improving the applicability of SMS review, and when there is a risk level in the evaluation report output by the SMS evaluation model, it provides a basis for subsequent determination of the review result corresponding to the SMS to be processed.

[0040] In one or more embodiments of this specification, the SMS evaluation model will perform multi-dimensional detection on the SMS to be processed. In practice, the SMS to be processed is usually a promotional SMS during SMS review. Therefore, when reviewing the SMS to be processed, it is necessary not only to perform semantic detection on the SMS content for sensitive words, exaggerated descriptions, etc., but also to use advertising-related clauses to check the compliance of the SMS to be processed. The specific implementation method is as follows: The SMS message to be processed and the prompt text are input into the SMS evaluation model to obtain the evaluation report output by the SMS evaluation model, including: The SMS message to be processed and the prompt text are input into the SMS evaluation model. Based on the SMS evaluation model, the SMS message to be processed is detected in multiple dimensions to obtain multi-dimensional detection results. Based on the detection results of the multiple dimensions, the evaluation report is obtained and output.

[0041] Among them, multiple dimensions can be understood as reviewing the content of the SMS to be processed from different perspectives. Multiple dimensions usually include semantic detection, clause matching, and level classification. The detection results of multiple dimensions can be understood as the conclusions or scores drawn by the SMS evaluation model after analyzing the SMS to be processed in each independent dimension. The evaluation report can be understood as the output document formed by the SMS evaluation model after comprehensively summarizing and standardizing the detection results of multiple dimensions.

[0042] Specifically, guided by the review rules corresponding to the prompt text, the SMS assessment model detects the SMS to be processed from multiple dimensions, determines the detection results under each dimension, and integrates the detection results from multiple dimensions to generate a comprehensive assessment report that includes the final risk determination.

[0043] In specific implementation, the SMS message to be processed is detected from multiple dimensions based on the SMS evaluation model to obtain multi-dimensional detection results, including: Based on the SMS evaluation model and the review rules in the prompt text, the SMS to be processed is subjected to semantic detection, clause matching and level classification to obtain semantic detection results, compliance detection results and risk levels, wherein the risk levels include first level, second level and third level.

[0044] Semantic detection can be understood as the SMS evaluation model using natural language understanding technology to analyze the deeper meaning, intent, and sentiment of the SMS to be processed, in order to identify violations that are not easily detected by keywords, such as prohibited or sensitive words, exaggerated promises, and hidden violations.

[0045] Clause matching can be understood as the SMS assessment model associating and matching the SMS to be processed with relevant clauses, industry standards, and other specific clauses in the knowledge base to confirm whether the content of the SMS to be processed is compliant. Risk level can be understood as the SMS assessment model classifying the overall violation risk of the SMS to be processed based on the detection results of multiple dimensions. In fact, the first level, the second level, and the third level usually correspond to different levels of risk severity from high to low. For example, the first level, the second level, and the third level correspond to high risk, medium risk, and low risk, respectively.

[0046] Specifically, when evaluating a drug promotion SMS, the SMS evaluation model can detect whether the SMS to be processed contains prohibited or sensitive words such as "national level" or "absolutely safe" in the semantic detection dimension, and can also detect the emotional tendency of the SMS to be processed (such as exaggerated promises) and hidden violations (such as variant expressions of "cures all diseases"); in the terms matching dimension, it can match whether the content of the SMS to be processed matches the relevant terms of the advertisement (for example, the SMS to be processed matches Article 17 "medical terminology shall not contain assertions of efficacy").

[0047] For example, when reviewing a product advertisement SMS claiming "exclusive formula, effective in three days, completely say goodbye to hair loss," the SMS evaluation model performs semantic analysis on the advertisement. It identifies "completely say goodbye" as an absolute promise and "effective in three days" as implying an unrealistic timeframe for efficacy. These statements contain exaggerated and misleading intent, resulting in a semantic analysis result. Regarding clause matching, the assertion "completely say goodbye to hair loss" is associated with the clause in the advertising terms that "medical terminology must not contain assertions of efficacy"; "effective in three days" is associated with the provision in the rights protection clause that "false or misleading advertising is prohibited," resulting in a specific compliance analysis result. Given that the product advertisement SMS contains explicit absolute assertions and exaggerated guarantees of efficacy, the violation is relatively serious. Therefore, the SMS evaluation model determines the risk level to "Level 1" (i.e., high risk).

[0048] In fact, multiple dimensions can be dynamically expanded. For example, if there is a purchase link in the pending SMS, the link security dimension can be used to check whether the purchase link provided in the pending SMS is a credible official channel. Based on the dynamically expanded multiple dimensions, the SMS evaluation model can improve its ability to identify risks in pending SMS, and improve the comprehensiveness and accuracy of the review.

[0049] For example, a pending text message might read, "Double 11 pre-sale! Choose classic tonifying and kidney-nourishing medicine, take A medicine, a course of treatment for as low as 50 yuan / box! >link.page for details, beware of scams, reply R to refuse." The corresponding evaluation report for this pending text message could include the following: Compliance of Drug Advertising: The phrase "Nourishing Yin and Tonifying Kidneys: Choose Classic Drugs" involves advertising the efficacy of a drug. According to Article 15 of the relevant advertising regulations, drug advertisements must not contain assertions or guarantees of efficacy (such as "classic" or "chosen"). If the instructions for "Drug A" do not clearly state the indication of "nourishing Yin and tonifying kidneys," or if it promotes the drug in a way that implies therapeutic effects, it may be suspected of false advertising.

[0050] Price authenticity: The claim of "as low as 50 yuan / box" needs to be verified to determine whether the original price is fabricated or a false discount is false. If this offer does not actually exist, it may violate Article 14 of the Price Regulations and Article 8 of the Competition Regulations regarding price fraud.

[0051] Link Security: Links to ">link.page" must be verified to ensure they are officially authorized by Tmall. Links that point to illegal websites or mislead users into phishing pages may violate Article 12 of the cybersecurity regulations.

[0052] Other risks: Although the "Beware of Scams" and "Reply R to Refuse" messages are standard formats, it is necessary to ensure that there are no hidden charges that induce users to reply (such as not clearly informing users of SMS reply charges).

[0053] In other words, the above-mentioned multiple dimensions include not only semantic detection and clause matching, but also price authenticity and link security.

[0054] Each dimension can generate corresponding detection results and risk confidence scores. By summarizing these detection results and risk confidence scores, an assessment report is formed. This assessment report can not only determine the risk level of the SMS to be processed, but also list in detail the specific problems found under each dimension, the relevant clauses, etc.

[0055] The data processing method provided in the embodiments of this specification detects SMS messages to be processed from multiple dimensions, which can largely avoid omissions in the review process. Furthermore, the detection results under each dimension can make the final output evaluation report more interpretable, facilitating quick identification of problems during subsequent manual review.

[0056] In one or more embodiments of the specification, since this embodiment is for reviewing SMS messages to be processed, manual review is usually required to obtain accurate review results. However, if an SMS evaluation model is used to detect the SMS messages to be processed, review suggestions can be given based on the detection results. This facilitates rapid review during manual review based on an assessment report that includes semantic detection results, compliance detection results, review suggestions, and risk levels. Specific implementation methods are described below: Based on the detection results from the multiple dimensions, the evaluation report is obtained and output, including: Based on the semantic detection results and the compliance detection results, an audit suggestion is generated; Obtain and output an assessment report that includes the semantic detection results, the compliance detection results, the audit recommendations, and the risk level.

[0057] The review suggestions can be understood as reference-based processing guidelines generated based on the violation intent identified by semantic detection and the specific clauses matched by clause matching. For example, the review suggestion could be: "There is a potential risk of violation. Further verification of the authenticity of the drug advertisement content, the legality of the price, and the security of the link is required. If the drug advertisement exceeds the approved scope or the price is false, it is non-compliant. If the drug indication is consistent with the advertisement, the price is true, and the link is legal, it is basically compliant."

[0058] Specifically, the SMS assessment model can utilize detection results from multiple dimensions. In this embodiment, the main focus is on semantic detection results and compliance detection results to generate targeted suggestions that provide review guidance. The problem description (semantic detection result), the supporting terms (compliance detection result), the review processing plan (review suggestions), and the overall risk level evaluation (risk level) of the SMS to be processed are integrated into a complete assessment report.

[0059] The data processing method provided in the embodiments of this specification provides the evaluation report with information such as the violation issues, the relevant clauses, and review suggestions for the SMS to be processed, in the case of subsequent manual review. The manual reviewer can rely on the explanatory information in the evaluation report without having to re-analyze the SMS to be processed, which greatly improves the review efficiency.

[0060] Step 206: Determine the review result corresponding to the SMS message to be processed based on the risk level in the assessment report.

[0061] Specifically, based on the different risk levels in the assessment report, different processing strategies are used to process the text messages to be processed, and the final review result for the text messages to be processed is determined.

[0062] In practice, the review result corresponding to the SMS message to be processed is determined based on the risk level in the assessment report, including: If the risk level in the assessment report is Level 1, the assessment report will be sent to the review end. If the review end determines that the semantic detection result and the compliance detection result are correct, the review result corresponding to the pending SMS will be determined as "not approved". If the risk level in the assessment report is Level 2, the assessment report is sent to the review end so that the review end can review the pending SMS message according to the review recommendations in the assessment report and determine the review result corresponding to the pending SMS message, wherein the review result is pass or fail. If the risk level in the assessment report is Level 3, the review result for the pending SMS message is determined to be passed.

[0063] The system is divided into three levels: Level 1 corresponds to high risk, which usually refers to SMS messages with clear and serious violations. The system tends to automatically reject the message, but to ensure accuracy, the assessment report still needs to be submitted to the reviewer for final confirmation; Level 2 corresponds to medium risk, which usually refers to SMS messages with potential violations or flawed wording. The system cannot make a completely automatic judgment and needs to rely on the reviewer to make a final review decision based on the assessment report; Level 3 corresponds to low risk, which usually refers to SMS messages that are basically compliant and have acceptable risk. The system can automatically approve the message.

[0064] The review terminal can be understood as a processing terminal specifically for manual review. The review result can be understood as the final result returned to the client that sent the SMS review request.

[0065] Specifically, when pending SMS messages are categorized by risk level, different processing methods are applied based on the risk level. For example, high-risk pending SMS messages are automatically intercepted and pushed to the review end to ensure that serious violations are not overlooked; medium-risk pending SMS messages are marked with review suggestions and judged based on the review end's decision-making ability, giving full play to the review end's decision-making ability for content that is uncertain in the SMS evaluation model; and low-risk pending SMS messages are directly approved to improve review efficiency.

[0066] For example, a text message deemed "Level 1" (high risk) and pending processing due to containing explicitly prohibited terms such as "cure" and "national-level" and involving unverified claims of medical efficacy will initially be rejected by the text message evaluation model. Simultaneously, a detailed evaluation report will be sent to the review stage. The human reviewer at the review stage will quickly verify the accuracy of the test results in the evaluation report. If there are no questions, the final review result will be "rejected."

[0067] Another text message was classified as "Level Two" (medium risk) due to the use of terms like "preferred" and "classic," which could constitute minor exaggeration in advertising. An assessment report containing specific review suggestions (e.g., the suggestion that there is a potential risk of violation and further verification of the drug advertisement's authenticity is needed; if the drug's advertising exceeds the approved scope, it is non-compliant; if the drug's indications are consistent with the advertising, it is basically compliant) was sent to the review panel. The human reviewers at the review panel, considering the context, determined whether these terms indeed constituted a violation and decided whether the review would pass or fail.

[0068] For a pending SMS message that is a routine activity notification and has no obvious violations, if the SMS assessment model determines its risk level to be "Level 3" (low risk), it will automatically give an "approved" review result without manual intervention.

[0069] In practice, the embodiments of this specification typically process a large number of pending SMS messages simultaneously. Given the risk level of each pending SMS message, multiple pending SMS messages can be sorted according to their risk level, thereby prioritizing the push of high-risk pending SMS messages to the review end for review.

[0070] The data processing method provided in the embodiments of this specification implements different processing strategies based on different risk levels, ensuring that the overall review efficiency is improved while strictly controlling high-risk content. Specifically, manual review is focused on medium- and high-risk pending SMS messages that require professional judgment, thus avoiding ineffective waste on a large amount of low-risk content.

[0071] In one or more embodiments of this specification, after determining the review result corresponding to the SMS message to be processed, the method further includes: returning the review result to the client.

[0072] The review result can be understood as the final judgment on the compliance of the SMS messages to be processed, usually a clear conclusion of "pass" or "fail".

[0073] Specifically, after performing model analysis and manual review on the SMS messages to be processed, a definite review result is obtained. This review result is then returned to the client that initiated the SMS review request as the output of the corresponding service of the SMS review platform, thus completing a complete service interaction.

[0074] In one or more embodiments of this specification, after reviewing the SMS message to be processed and obtaining the review result, a complete service interaction is completed. The data involved in this service interaction is stored in the review log so that the SMS evaluation model can be iteratively trained based on the content of the review log. Specific implementation methods are as follows: After determining the review result corresponding to the SMS message to be processed, the process also includes: The pending SMS messages, the evaluation report, and the review results are stored in the review log. The SMS evaluation model is iteratively trained based on the pending SMS messages, evaluation reports, and evaluation results in the review log.

[0075] The review log is used to persistently store the relevant data for each review task. The relevant data for each review task includes at least the SMS messages to be processed, the generated evaluation report, and the final output review results. Iterative training can be understood as using the data in the continuously accumulated review log as new training samples to periodically or trigger updates to the existing SMS evaluation model, so that the trained SMS evaluation model can adapt to new language patterns and improve evaluation performance.

[0076] Specifically, for each review task (i.e., each SMS review request), a complete record of pending data, evaluation reports, and review results is generated during the review process. These records are stored in the review log, forming a continuously accumulating labeled dataset. In fact, the detailed review log also provides a complete data foundation for compliance audits, case analysis, and rule optimization.

[0077] We obtain real-world processing data (including pending SMS messages, evaluation reports, and review results) from the review logs and use this data as training data to retrain the existing SMS evaluation model. This allows the SMS evaluation model to learn from real-world processing data (especially data that has been corrected or confirmed through manual review), thereby enabling continuous calibration and improvement of the model's judgment capabilities.

[0078] For example, a health product advertisement SMS message that was judged as "high-risk" by the SMS evaluation model and recommended "not approved" was given an "approved" result after manual review. The SMS message to be processed, the evaluation report generated by the SMS evaluation model containing the erroneous risk judgment, and the final "approved" result after manual correction were all stored in the review log.

[0079] In subsequent model iterations and training cycles, this record will serve as a correction sample. During the training process of the SMS evaluation model, it will guide the model to tend to classify SMS messages containing similar semantics as "low risk" or "medium risk," rather than the original "high risk." Based on this feedback data obtained from real-world scenarios, the evaluation conclusions of the trained SMS evaluation model can increasingly approach the review results of human experts.

[0080] The data processing method provided in the embodiments of this specification transforms data in real-world scenarios into training data, enabling the trained SMS evaluation model to quickly adapt to changes in the scenario. This effectively alleviates the performance degradation problem caused by the lag in training data and steadily improves the accuracy and reliability of automated review through continuous optimization.

[0081] See Figure 3 , Figure 3 A flowchart illustrating the processing procedure of a data processing method according to an embodiment of this specification is shown, specifically including the following steps.

[0082] Step 302: Data preprocessing.

[0083] Specifically, the data preprocessing module is responsible for receiving the original SMS text (i.e., the SMS to be processed in the above embodiment) and performing word segmentation and entity recognition on the SMS text. For example, the recognized entities can be drug names, medical terms, etc.

[0084] The extracted entities are categorized to determine their corresponding classes. Based on the classification results, the system dynamically matches and generates the corresponding prompt text for the pending SMS messages. It should be noted that different classes correspond to different levels of review rigor and focus. For example, the prompt text for pending SMS messages related to pharmaceuticals will emphasize the review of efficacy claims and absolute terms, while the prompt text for general product promotions will focus more on the authenticity of prices and prohibited words.

[0085] In practical applications, the SMS messages to be processed carried in the SMS review request are usually SMS templates. That is, the merchant or enterprise requests the review of the SMS template, so that if the SMS template is approved, the merchant or enterprise can instantiate and generate a large number of SMS messages based on the SMS template.

[0086] Step 304: Model Analysis.

[0087] The SMS text and notification text provided by the data preprocessing module are input into the SMS evaluation model. The SMS evaluation model then performs semantic compliance checks, clause matching, and risk level assessments on the SMS text.

[0088] Specifically, semantic compliance detection is used to identify whether there are implicit violations (such as variations of "cure all diseases"), emotional biases (such as exaggerated promises), and prohibited / sensitive words (such as "absolutely safe") in SMS text. Terms matching is used to map SMS text to specific terms.

[0089] Based on the severity of the violation, the text message is assessed for risk level and scored (e.g., a maximum score of 5 points). High risk (3-5 points) usually corresponds to explicit illegal or seriously misleading content; medium risk (1-2 points) corresponds to potential violations or inappropriate expression; low risk (0 points) is basically compliant content.

[0090] The SMS assessment model will output a structured assessment report that includes information on location violations, related terms, review recommendations, and risk levels.

[0091] Step 306: Human-machine collaboration.

[0092] Based on the risk level output by the SMS assessment model, a tiered handling strategy is implemented.

[0093] Specifically, high-risk SMS texts are automatically blocked, and an assessment report is sent to a human review panel. Reviewers then confirm the violation and issue a "fail" result. Medium-risk SMS texts are also sent to a human review panel, where reviewers combine the assessment report's review recommendations to determine whether the review is passed or failed. Low-risk SMS texts are automatically marked as passed.

[0094] In practical applications, high-risk SMS text messages are prioritized and pushed to users for manual review through intelligent sorting.

[0095] The subsequent processing records (including SMS texts, evaluation reports, and review results) will be stored in the review log and used as training data to feed back into the SMS evaluation model to achieve closed-loop optimization.

[0096] The data processing method provided in the embodiments of this specification, combined with natural language processing and advertising-related regulations, automates the review of the compliance of pending SMS messages. It can also push pending SMS messages marked as high or medium risk to human review, significantly reducing the rate of missed reviews by humans and improving the accuracy of SMS review, thereby effectively preventing false advertising in medical fields and protecting consumer rights. When reviewing a large number of pending SMS messages, it effectively reduces review costs and is suitable for content review scenarios in highly sensitive industries such as healthcare and e-commerce.

[0097] Corresponding to the above method embodiments, this specification also provides data processing apparatus embodiments. Figure 4 A schematic diagram of the structure of a data processing apparatus according to one embodiment of this specification is shown. Figure 4 As shown, the device includes: SMS acquisition module 402 is configured to acquire SMS messages to be processed and the corresponding prompt text of the SMS messages to be processed. The SMS evaluation module 404 is configured to input the SMS to be processed and the prompt text into the SMS evaluation model to obtain an evaluation report output by the SMS evaluation model. The evaluation report includes a risk level classification of the SMS to be processed based on the review rules in the prompt text. The result acquisition module 406 is configured to determine the review result corresponding to the SMS message to be processed based on the risk level in the assessment report.

[0098] Optionally, the SMS acquisition module 402 is further configured to: Entity extraction is performed on the SMS message to be processed to obtain the target entity; The target entity is classified, and the prompt text corresponding to the SMS to be processed is determined based on the classification result. The prompt text contains the review rules corresponding to the classification result, and the strictness of the review rules varies for different classification results.

[0099] Optionally, the SMS evaluation module 404 is further configured to: The SMS message to be processed and the prompt text are input into the SMS evaluation model. Based on the SMS evaluation model, the SMS message to be processed is detected in multiple dimensions to obtain multi-dimensional detection results. Based on the detection results of the multiple dimensions, the evaluation report is obtained and output.

[0100] Optionally, the SMS evaluation module 404 is further configured to: Based on the SMS evaluation model and the review rules in the prompt text, the SMS to be processed is subjected to semantic detection, clause matching and level classification to obtain semantic detection results, compliance detection results and risk levels, wherein the risk levels include first level, second level and third level.

[0101] Optionally, the SMS evaluation module 404 is further configured to: Based on the semantic detection results and the compliance detection results, an audit suggestion is generated; Obtain and output an assessment report that includes the semantic detection results, the compliance detection results, the audit recommendations, and the risk level.

[0102] Optionally, the result obtaining module 406 is further configured to: If the risk level in the assessment report is Level 1, the assessment report will be sent to the review end. If the review end determines that the semantic detection result and the compliance detection result are correct, the review result corresponding to the pending SMS will be determined as "not approved". If the risk level in the assessment report is Level 2, the assessment report is sent to the review end so that the review end can review the pending SMS message according to the review recommendations in the assessment report and determine the review result corresponding to the pending SMS message, wherein the review result is pass or fail. If the risk level in the assessment report is Level 3, the review result for the pending SMS message is determined to be passed.

[0103] Optionally, the SMS acquisition module 402 is further configured to: In response to an SMS review request sent by the client, the pending SMS message carried in the SMS review request is obtained.

[0104] Optionally, the result obtaining module 406 is further configured to: The audit results are returned to the client.

[0105] The device further includes: The post-processing module is configured to store the pending SMS messages, the evaluation report, and the review results in the review log; and to iteratively train the SMS evaluation model based on the pending SMS messages, the evaluation report, and the review results in the review log.

[0106] This specification provides a data processing apparatus in one embodiment that acquires a text message to be processed and its corresponding prompt text. By determining the prompt text corresponding to the text message to be processed, targeted guidance is provided for subsequent review. The text message to be processed and the prompt text are input into a text message evaluation model to obtain an evaluation report output by the text message evaluation model. Natural language processing based on the text message evaluation model can perform semantic understanding of the text message to be processed, thereby avoiding deep violations that may be missed by human reviewers due to fatigue or negligence. According to the risk level in the evaluation report, the review result corresponding to the text message to be processed is determined. This significantly reduces the overall labor cost of review, improves review efficiency, and reduces the risk of missed or incorrect reviews due to human fatigue, achieving an optimized balance between cost, efficiency, and accuracy.

[0107] The above is an illustrative scheme of a data processing apparatus according to this embodiment. It should be noted that the technical solution of this data processing apparatus and the technical solution of the data processing method described above belong to the same concept. For details not described in detail in the technical solution of the data processing apparatus, please refer to the description of the technical solution of the data processing method described above.

[0108] Figure 5 A structural block diagram of a computing device 500 according to one embodiment of this specification is shown. The components of the computing device 500 include, but are not limited to, a memory 510 and a processor 520. The processor 520 is connected to the memory 510 via a bus 530, and a database 550 is used to store data.

[0109] The computing device 500 also includes an access device 540, which enables the computing device 500 to communicate via one or more networks 560. Examples of these networks include Public Switched Telephone Network (PSTN), Local Area Network (LAN), Wide Area Network (WAN), Personal Area Network (PAN), or combinations of communication networks such as the Internet. The access device 540 may include one or more of any type of wired or wireless network interface (e.g., a network interface card (NIC)), such as an IEEE 802.11 Wireless Local Area Network (WLAN) wireless interface, a Wi-MAX (Worldwide Interoperability for Microwave Access) interface, an Ethernet interface, a Universal Serial Bus (USB) interface, a cellular network interface, a Bluetooth interface, or a Near Field Communication (NFC) interface.

[0110] In one embodiment of this specification, the above-described components of the computing device 500 and Figure 5 Other components, not shown, can also be connected to each other, for example, via a bus. It should be understood that... Figure 5 The block diagram of the computing device shown is for illustrative purposes only and is not intended to limit the scope of this specification. Those skilled in the art can add or replace other components as needed.

[0111] The computing device 500 can be any type of stationary or mobile computing device, including mobile computers or mobile computing devices (e.g., tablet computers, personal digital assistants, laptop computers, notebook computers, netbooks, etc.), mobile phones (e.g., smartphones), wearable computing devices (e.g., smartwatches, smart glasses, etc.) or other types of mobile devices, or stationary computing devices such as desktop computers or personal computers (PCs). The computing device 500 can also be a mobile or stationary server.

[0112] The processor 520 is used to execute the following computer program / instructions, which, when executed by the processor, implement the steps of the above-described data processing method.

[0113] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the computing device embodiments are basically similar to the data processing method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions of the data processing method embodiments.

[0114] An embodiment of this specification also provides a computer-readable storage medium storing a computer program / instructions that, when executed by a processor, implement the steps of the above-described data processing method.

[0115] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the computer-readable storage medium embodiments are basically similar to the data processing method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions of the data processing method embodiments.

[0116] An embodiment of this specification also provides a computer program product, including a computer program / instructions that, when executed by a processor, implement the steps of the above-described data processing method.

[0117] The above is an illustrative scheme of a computer program product according to this embodiment. It should be noted that the technical solution of this computer program product and the technical solution of the data processing method described above belong to the same concept. For details not described in detail in the technical solution of the computer program product, please refer to the description of the technical solution of the data processing method described above.

[0118] The foregoing has described specific embodiments of this specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in a different order than that shown in the embodiments and may still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require the specific or sequential order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0119] The computer instructions include computer program code, which may be in the form of source code, object code, executable file, or certain intermediate forms. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording media, USB flash drive, portable hard drive, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content included in the computer-readable medium may be appropriately added or removed according to the requirements of patent practice. For example, in some regions, according to patent practice, computer-readable media may not include electrical carrier signals and telecommunication signals.

[0120] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that the embodiments in this specification are not limited to the described order of actions, because according to the embodiments in this specification, some steps can be performed in other orders or simultaneously. Furthermore, those skilled in the art should also understand that the embodiments described in this specification are all preferred embodiments, and the actions and modules involved are not necessarily essential to the embodiments in this specification.

[0121] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0122] The preferred embodiments disclosed above are merely illustrative of this specification. The optional embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. Clearly, many modifications and variations can be made based on the embodiments described herein. These embodiments are selected and specifically described in this specification to better explain the principles and practical applications of the embodiments, thereby enabling those skilled in the art to better understand and utilize this specification. This specification is limited only by the claims and their full scope and equivalents.

Claims

1. A data processing method, comprising: Obtain the pending SMS messages and the corresponding prompt text for the pending SMS messages; The pending SMS message and the prompt text are input into the SMS evaluation model to obtain the evaluation report output by the SMS evaluation model. The evaluation report includes the risk level of the pending SMS message based on the review rules in the prompt text. Based on the risk level stated in the assessment report, the review result corresponding to the pending SMS message is determined.

2. The method as described in claim 1, wherein obtaining the SMS message to be processed and the corresponding prompt text of the SMS message to be processed includes: Entity extraction is performed on the SMS message to be processed to obtain the target entity; The target entity is classified, and the prompt text corresponding to the SMS to be processed is determined based on the classification result. The prompt text contains the review rules corresponding to the classification result, and the strictness of the review rules varies for different classification results.

3. The method as described in claim 1, wherein the SMS message to be processed and the prompt text are input into an SMS evaluation model to obtain an evaluation report output by the SMS evaluation model, comprising: The SMS message to be processed and the prompt text are input into the SMS evaluation model. Based on the SMS evaluation model, the SMS message to be processed is detected in multiple dimensions to obtain multi-dimensional detection results. Based on the detection results of the multiple dimensions, the evaluation report is obtained and output.

4. The method as described in claim 3, wherein the SMS message to be processed is detected in multiple dimensions based on the SMS evaluation model to obtain multi-dimensional detection results, including: Based on the SMS evaluation model and the review rules in the prompt text, the SMS to be processed is subjected to semantic detection, clause matching and level classification to obtain semantic detection results, compliance detection results and risk levels, wherein the risk levels include first level, second level and third level.

5. The method as described in claim 4, wherein the evaluation report is obtained and output based on the detection results of the multiple dimensions, comprising: Based on the semantic detection results and the compliance detection results, an audit suggestion is generated; Obtain and output an assessment report that includes the semantic detection results, the compliance detection results, the audit recommendations, and the risk level.

6. The method as described in claim 5, wherein determining the review result corresponding to the pending SMS message based on the risk level in the assessment report includes: If the risk level in the assessment report is Level 1, the assessment report will be sent to the review end. If the review end determines that the semantic detection result and the compliance detection result are correct, the review result corresponding to the pending SMS will be determined as "not approved". If the risk level in the assessment report is Level 2, the assessment report is sent to the review end so that the review end can review the pending SMS message according to the review recommendations in the assessment report and determine the review result corresponding to the pending SMS message, wherein the review result is pass or fail. If the risk level in the assessment report is Level 3, the review result for the pending SMS message is determined to be passed.

7. The method according to any one of claims 1-6, after determining the review result corresponding to the SMS message to be processed, further comprising: The pending SMS messages, the evaluation report, and the review results are stored in the review log. The SMS evaluation model is iteratively trained based on the pending SMS messages, evaluation reports, and evaluation results in the review log.

8. A computing device, comprising: Memory and processor; The memory is used to store computer programs / instructions, and the processor is used to execute the computer programs / instructions, which, when executed by the processor, implement the steps of the data processing method according to any one of claims 1 to 7.

9. A computer-readable storage medium storing a computer program / instructions that, when executed by a processor, implement the steps of the data processing method according to any one of claims 1 to 7.

10. A computer program product comprising a computer program / instructions that, when executed by a processor, implement the steps of the data processing method according to any one of claims 1 to 7.

Images