Text violation processing method and device
By using the violation determination model and the violation handling model in the advertising title review, the replacement title is automatically generated, which solves the problem that users need to manually modify the violation title, and improves the review efficiency and advertising delivery efficiency.
Patent Information
- Application Number
- CN202411750012.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-29
- Publication Date
- 2025-03-21
AI Technical Summary
In the prior art, in the review of advertisement titles, when the title is determined to be a violation, the user needs to manually modify it based on the reasons for the feedback of the violation, resulting in inefficient review and affecting the efficiency of advertising delivery.
Obtain the judgment result and the reason for the violation by obtaining the initial title text and inputting it into the pre-trained violation judgment model. Then, enter the initial title text and the violation reason to the violation processing model to generate the target title text to replace the initial title text.
It realizes user-side unconscious and intelligent text review and revisions after rejection, improves text processing efficiency, saves the time spent on manual modification, and improves advertising delivery efficiency.
Smart Images
Figure CN119669473A_ABST
Abstract
Description
Technical Field
[0001] The embodiments of the present application relate to the field of computer technology, and in particular, to a method, apparatus, computer device, computer-readable storage medium, and computer program product for processing text violations. Background Art
[0002] The ad title is a key part of the ad content. A good ad title can quickly attract the attention of the target audience and make them interested in the ad content. In order to ensure the safety of ad titles, the platform will set up a corresponding review mechanism to review the ad titles submitted by users.
[0003] However, when the review mechanism determines that the ad title violates the rules, it will only provide users with feedback on the reasons for the violation, requiring users to modify the ad title based on the reasons for the violation, which affects the efficiency of ad delivery.
[0004] It should be noted that the above content is not necessarily prior art, nor is it intended to limit the scope of patent protection of this application. Summary of the invention
[0005] The embodiments of the present application provide a text violation processing method, apparatus, computer device, computer-readable storage medium, and computer program product to solve or alleviate one or more of the technical problems raised above.
[0006] One aspect of an embodiment of the present application provides a text violation processing method, the method comprising: Get the initial title text of the target object input; Inputting the initial title text into a pre-trained violation determination model to obtain a determination result; wherein, if the determination result is a violation, the determination result includes the reason for the violation of the initial title text; In the case where the determination result is a violation, the initial title text and the violation reason are input into a pre-trained violation processing model to obtain a target title text for replacing the initial title text; The target title text is used to be returned to the target object or transferred to a manual review node for review.
[0007] Optionally, the method further comprises: When the determination result is that there is no violation, the initial title text is transferred to the manual review node.
[0008] Optionally, the violation processing model is used to output a plurality of modified texts according to the violation reason and the initial title text; obtaining a target title text for replacing the initial title text comprises: Inputting the plurality of modified texts into the violation determination model to screen out a plurality of compliant modified texts from the plurality of modified texts; The target title text is determined based on the multiple compliant modification texts.
[0009] Optionally, determining the target title text according to the plurality of compliant modification texts includes: Inputting the plurality of compliant modified texts into a pre-trained text quality assessment model respectively to obtain a text quality parameter of each of the compliant modified texts; The compliant modified text with the highest text quality parameter among the multiple compliant modified texts is determined as the target title text.
[0010] Optionally, the method further comprises: Sending the determination result and the target title text to the target object; receiving a confirmation result of the target object with respect to the determination result and the target title text; If the confirmation result is to accept the modification, the target title text is transferred to the manual review node; When the confirmation result is that the modification is not accepted, the initial title text is transferred to the manual review node.
[0011] Optionally, the method further comprises: Transfer the text to be reviewed to a manual review node, and receive the manual review result; wherein the text to be reviewed is the target title text or the initial title text transferred to the manual review node, and when the manual review result is failure to review, the manual review result includes the manual review reason for the text to be reviewed; If the manual review result is that the review is passed, the text to be reviewed is returned to the target object; In the case where the manual review result is that the review is not passed, determining whether the text to be reviewed is an initial title text that does not accept modification; In the case where the text to be reviewed is an initial title text that does not accept modification, returning the manual review result to the target object; In the case where the text to be reviewed is not the initial title text that does not accept modification, the manual review reason of the text to be reviewed and the text to be reviewed are input into a pre-trained violation processing model to obtain a secondary modified text for replacing the text to be reviewed; Transferring the second modified text to the manual review node to receive the manual review result and manual review reason corresponding to the second modified text; In the case where the manual review result corresponding to the secondary revised text is failure to pass the review, the secondary revised text and the manual review result corresponding to the secondary revised text are returned to the target object.
[0012] Another aspect of an embodiment of the present application provides a text violation processing device, the device comprising: The acquisition module obtains the initial title text input by the target object; A first input module is used to input the initial title text into a pre-trained violation determination model to obtain a determination result; wherein, if the determination result is a violation, the determination result includes the violation reason of the initial title text; A second input module is used for inputting the initial title text and the violation reason into a pre-trained violation processing model to obtain a target title text for replacing the initial title text when the determination result is a violation; The target title text is used to be returned to the target object or transferred to a manual review node for review.
[0013] Another aspect of an embodiment of the present application provides a computer device, including: at least one processor; and a memory communicatively coupled to the at least one processor; Wherein: the memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the method as described above.
[0014] Another aspect of an embodiment of the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer instructions, and when the computer instructions are executed by a processor, the method described above is implemented.
[0015] Another aspect of an embodiment of the present application provides a computer program product, including a computer program, which implements the method described above when executed by a processor.
[0016] The above technical solution adopted in the embodiment of the present application may have the following advantages: After the initial review determines that the initial title text violates the rules, the model is used to process the initial title text directly to generate the modified target title text. This allows the user to be unaware and intelligently modify the text after the text review is rejected, thereby greatly improving the text processing efficiency and saving the time spent on manual modification. BRIEF DESCRIPTION OF THE DRAWINGS The accompanying drawings exemplarily illustrate the embodiments and constitute a part of the specification, and together with the text description of the specification, are used to explain the exemplary implementation of the embodiments. The embodiments shown are for illustrative purposes only and do not limit the scope of the claims. In all drawings, the same reference numerals refer to similar but not necessarily identical elements.
[0017] Figure 1 The operating environment diagram of the text violation processing method according to the first embodiment of the present application is schematically shown; Figure 2 The flowchart of the text violation processing method according to the first embodiment of the present application is schematically shown; Figure 3 Schematically shows Figure 2 Flow chart of sub-steps of step S204; Figure 4 Schematically shows Figure 3 Flow chart of sub-steps of step S302; Figure 5 The newly added flow chart of the text violation processing method according to the first embodiment of the present application is schematically shown; Figure 6 Another newly added flow chart of the text violation processing method according to the first embodiment of the present application is schematically shown; Figure 7 The following schematically shows a workflow diagram of a text violation processing method according to Embodiment 1 of the present application; Figure 8 A block diagram of a text violation processing device according to Embodiment 2 of the present application is schematically shown; and Fig. 9 The hardware architecture diagram of the computer device according to the third embodiment of the present application is schematically shown. DETAILED DESCRIPTION
[0018] In order to make the purpose, technical solutions and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not intended to limit the present application. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of the present application.
[0019] It should be noted that the descriptions involving "first", "second", etc. in the embodiments of the present application are only for descriptive purposes and cannot be understood as indicating or implying their relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In addition, the technical solutions between the various embodiments can be combined with each other, but they must be based on the ability of ordinary technicians in the field to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be deemed that such combination of technical solutions does not exist and is not within the scope of protection required by this application.
[0020] In the description of the present application, it should be understood that the numerical labels before the steps do not indicate the order in which the steps are executed, but are only used to facilitate the description of the present application and to distinguish each step, and therefore should not be understood as a limitation on the present application.
[0021] First, the following terms are explained: LLM: Large Language Model, is an artificial intelligence model used to understand and generate human language. It is trained on a large amount of text data to learn complex language patterns and perform text summarization, translation, and sentiment analysis.
[0022] NLP: Natural Language Processing, is a branch of artificial intelligence and linguistics that is used to enable computers to understand, interpret and generate human language to achieve human-computer interaction.
[0023] OCR: Optical Character Recognition, used to convert text in various types of documents (such as scanned paper documents, PDF files, or pictures taken by digital cameras) into editable and searchable data.
[0024] CTR: Click-Through Rate, an important indicator for measuring the effectiveness of Internet advertising. CTR refers to the ratio of the number of times users click on an ad to the number of times the ad is displayed, expressed as a percentage.
[0025] eCPM: Effective Cost Per Mille, is an indicator to measure the efficiency of advertising revenue. It indicates the cost paid by advertisers for every 1,000 ad impressions.
[0026] Landing Page: Landing Page, also known as landing page, refers to the first page that users enter by clicking on ads, search results, social media links, etc.
[0027] Secondly, to facilitate technical personnel in this field to understand the technical solutions provided by the embodiments of the present application, the relevant technologies are described below: After the text titles (such as advertising creative titles) of the promotional media in the promotion platform (such as the advertising delivery platform) are reviewed and rejected, only the reasons for rejection are fed back to the user. The user needs to understand the reasons for rejection and then make targeted modifications to the original advertising creativity. The overall process operation is complicated and time-consuming, with a low degree of intelligence, which also affects the efficiency of advertising delivery.
[0028] To this end, the embodiment of the present application provides a technical solution for handling text violations. In this technical solution, AI technology is combined with the advertising creative review and rejection scenario. After the user chooses to enable this function, if the advertising creative is rejected in the review stage due to non-compliant content, the "Advertising Creative Title Review Rejection AI Modification" will automatically interpret the review rejection reasons, and modify the advertising creative title according to the review rejection reasons and submit it for review again. It can achieve user-unaware modification of advertising creatives after review and rejection, which can greatly improve the overall advertising delivery efficiency of advertisers and save the time spent on manually modifying the review and rejection of advertising creative titles. See below for details.
[0029] Finally, for ease of understanding, an exemplary operating environment is provided below.
[0030] like Figure 1 As shown, the operating environment diagram includes: computer equipment 2, network 4 and user terminal 6.
[0031] Computer device 2 can be composed of single or multiple computing devices, such as rack server, blade server, tower server or cabinet server (including independent server, or server cluster composed of multiple servers), etc. The one or more computer devices can include virtualized computing instances. Computer device can load virtual machine based on virtual image and / or other data of specific software (e.g., operating system, special application, server) defined for simulation. As the demand for different types of processing services changes, different virtual machines can be loaded and / or terminated on the one or more computer devices. Multiple models can be run on computer device 2, such as violation determination model, violation processing model and text quality assessment model, etc.
[0032] The computer device 2 can receive text data (such as initial title text) uploaded by the user terminal 6 through one or more networks 4, and review and modify the text data through various models running on the computer device 2.
[0033] The network 4 may include various network devices, such as routers, switches, multiplexers, hubs, modems, bridges, repeaters, firewalls, proxy devices, and / or the like. The network 4 may include physical links, such as coaxial cable links, twisted pair cable links, fiber optic links, combinations thereof, and / or the like. The network 4 may include wireless links, such as cellular links, satellite links, Wi-Fi links, and / or the like.
[0034] The user terminal 6 can be configured to access the computer device 2. The user terminal 6 may include any type of computer device that can run a game engine, such as: a smart phone, a tablet device, a laptop computer, a smart device (e.g., a smart watch, smart glasses), a virtual reality, a gaming device, a set-top box, a digital streaming device, a car terminal, a smart TV, a TV box, an MP4 (Moving Picture Experts Group Audio Layer IV) player, etc.
[0035] The user terminal 6 can run an operating system such as Windows, Android (AndroidTM) or iOS. In addition, the user can also install various applications and program components as needed. Based on the above programs or program components, various functions can be implemented, such as inputting text, and then the computer device 2 returns a corresponding reply (i.e., the target title text or the audit reason, etc.).
[0036] The following uses the computer device 2 as the execution subject to introduce the technical solution of the present application through multiple embodiments. It should be noted that these embodiments can be implemented in a variety of different forms and should not be interpreted as being limited to the embodiments described here.
[0037] Embodiment 1 Figure 2 The flowchart of the text violation processing method according to the first embodiment of the present application is schematically shown.
[0038] like Figure 2 As shown, the text violation processing method may include steps S200 to S204, wherein: Step S200, obtaining the initial title text input by the target object.
[0039] Step S202, inputting the initial title text into a pre-trained violation determination model to obtain a determination result; wherein, when the determination result is a violation, the determination result includes the reason for the violation of the initial title text.
[0040] Step S204, when the judgment result is a violation, the initial title text and the violation reason are input into a pre-trained violation processing model to obtain a target title text for replacing the initial title text; wherein the target title text is used to return to the target object or circulate to a manual review node for review.
[0041] The text violation processing method provided in this embodiment directly processes the initial title text through the model after the initial review determines that the initial title text violates the rules, and generates a modified target title text. In this way, the text can be modified after the text review is rejected without user perception and in an intelligent manner, thereby greatly improving the text processing efficiency and saving the time spent on manual modification.
[0042] The following combination Figure 2 , each step in steps S200~S204 and other optional steps are described in detail.
[0043] Step S200 , get the initial title text of the target object input.
[0044] In specific implementation, the user terminal 6 may be provided with various components such as a text input box. The user may input the initial title text through the input box. The initial title text may be an advertisement title, a product title, a video title, a live broadcast room title, etc.
[0045] In some embodiments, the text input box may provide multiple suggested titles for the keyword input by the user, for the user to select. The initial title text finally obtained may be the suggested title selected by the user.
[0046] Step S202 , input the initial title text into a pre-trained violation determination model to obtain a determination result; wherein, when the determination result is a violation, the determination result includes the reason for the violation of the initial title text.
[0047] In specific implementation, the violation determination model can be based on LLM, NLP, etc. The violation determination model can be trained based on the following data: title text that has passed review (such as manual review or review by other review mechanisms), text content extracted using OCR tools on approved advertising images or covers, historical modification records of title text by AI or manual, and historical title text that has been rejected by review, etc. According to actual needs, the violation determination model can also be obtained by training based on other models or algorithms using other available raw data. In some embodiments, a third-party general or special language large model can also be called through an interface.
[0048] The trained violation determination model is used to review the input title text (such as the initial title text), determine its rejection risk, estimate the probability of review rejection, and locate the category of the rejection reason (i.e., the violation reason).
[0049] In some embodiments, the violation reasons may include sensitive public opinion, illegal and irregular behavior, misleading or common sense errors, etc. In other embodiments, other violation reasons may also include false advertising, too many special symbols, account problems, etc.
[0050] In this embodiment, the initial title text is automatically and intelligently reviewed through a violation determination model, and specific reasons for the violation are given when a violation is determined. This can effectively improve the review efficiency, ensure the compliance of the title content, and provide a clear improvement direction for subsequent modifications to the title text.
[0051] In an optional embodiment, the method further comprises: When the determination result is that there is no violation, the initial title text is transferred to the manual review node.
[0052] In some embodiments, the manual review node can analyze and review the creativity, emotion, etc. of the title text. In other embodiments, the manual review node can also review other aspects.
[0053] In this embodiment, manual review of the initial title text for which the model has not detected any illegal content can further ensure that the creativity and emotional expression of the title content are appropriate, while avoiding potential risks of illegal activities and improving the overall quality of the title text.
[0054] Step S204 , when the judgment result is a violation, the initial title text and the violation reason are input into a pre-trained violation processing model to obtain a target title text for replacing the initial title text; wherein, the target title text is used to return to the target object or circulate to a manual review node for review.
[0055] In specific implementation, the violation processing model can be based on LLM, NLP, etc. The violation processing model can be obtained through the following data training: various normal title data, violation title data, and title modification data training. According to actual needs, it can also be based on other models or algorithms, and use other available raw data for training to obtain the violation processing model. In some embodiments, a third-party general or special language large model can also be called through an interface.
[0056] In some embodiments, the violation processing model can modify and replace the initial title text by synonym replacement, word order adjustment, content reconstruction, etc. according to the violation reason. For example, when a few words in the initial title text involve sensitive public opinion, other non-sensitive words are used to replace them to obtain a target title text that meets the regulations. In other embodiments, the violation processing model can also process the initial title text in other ways.
[0057] It should be noted that the user can choose whether to use the text violation processing method provided in this embodiment. If the user chooses not to use it, the initial title text entered by the user will not be processed by the text violation processing method provided in this embodiment.
[0058] In this embodiment, after determining that the title text violates the rules, the violation processing model is directly used to process the title text according to the violation reasons. This can save the step of the user (i.e., the target object) who inputs the initial title text to modify it according to the violation reasons, thereby improving the review efficiency and reducing the burden of users repeatedly modifying the title due to unfamiliarity with the regulations.
[0059] In an optional embodiment, the violation processing model is used to output multiple modified texts according to the violation reason and the initial title text. Figure 3 As shown, step S204 includes: S300, inputting the plurality of modified texts into the violation determination model to screen out a plurality of compliant modified texts from the plurality of modified texts; S302: Determine the target title text according to the plurality of compliant modification texts.
[0060] In this embodiment, the modified text obtained by the violation processing model will also be reviewed by the violation determination model to ensure that the processed title text is fully compliant and improve the security and reliability of the title content.
[0061] In an optional embodiment, if Figure 4 As shown, step S302 includes: S400, inputting the plurality of compliant modified texts into a pre-trained text quality assessment model respectively to obtain a text quality parameter of each compliant modified text; S402: Determine the compliant modified text with the highest text quality parameter among the multiple compliant modified texts as the target title text.
[0062] In some embodiments, the text quality parameter may be the predicted CTR (click-through rate) of the compliant modified text and the eCPM (effective cost per thousand impressions) of different cross-product combinations of the compliant modified text and other related content (such as pictures / videos / landing pages in the advertisement when the text is used as the advertisement title). In other embodiments, the text quality parameter may also be other parameters.
[0063] In specific implementation, the text quality assessment model can be obtained by training based on LLM, NLP, etc., using a large number of title texts that are put into use or simulated and their corresponding text quality parameters. According to actual needs, the text quality assessment model can also be obtained by training based on other models or algorithms and using other training data.
[0064] In this embodiment, the compliant modified text with the highest text quality parameter is selected as the target title text. Providing high-quality text titles can effectively increase the audience's interest in clicking on the title, thereby increasing the exposure of the content.
[0065] In an optional embodiment, if Figure 5 As shown, the method also includes: S500, sending the determination result and the target title text to the target object; S502, receiving a confirmation result of the target object with respect to the determination result and the target title text; S504, when the confirmation result is to accept the modification, the target title text is transferred to the manual review node; S506: When the confirmation result is that the modification is not accepted, the initial title text is transferred to the manual review node.
[0066] In some embodiments, when the initial title text is judged to be non-compliant, a prompt indicating the risk of rejection will be displayed below the input box of the user terminal 6, and the modified target title text will be displayed to guide the user (i.e., the target object) to choose whether to adopt the modified title. In other embodiments, the user may be notified in other ways to obtain the user's confirmation result.
[0067] In this embodiment, before the modified target title text is put into the next step for manual review, the user's (i.e., the target object) opinion on the target title text will be confirmed first. In this way, unnecessary waste of manual review resources can be avoided, while improving user satisfaction and participation.
[0068] In an optional embodiment, if Figure 6 As shown, the method also includes: S600, transferring the text to be reviewed to a manual review node, and receiving a manual review result; wherein the text to be reviewed is the target title text or the initial title text transferred to the manual review node, and when the manual review result is failure to review, the manual review result includes a manual review reason for the text to be reviewed; S602, when the manual review result is that the review is passed, the text to be reviewed is returned to the target object; S604, when the manual review result is failure to review, determining whether the text to be reviewed is an initial title text that does not accept modification; S606, when the text to be reviewed is an initial title text that does not accept modification, returning the manual review result to the target object; S608, when the text to be reviewed is not the initial title text that does not accept modification, inputting the manual review reason of the text to be reviewed and the text to be reviewed into a pre-trained violation processing model to obtain a secondary modified text for replacing the text to be reviewed; S610, transferring the second modified text to the manual review node to receive the manual review result and manual review reason corresponding to the second modified text; S612: When the manual review result corresponding to the second modified text is failure to pass the review, the second modified text and the manual review result corresponding to the second modified text are returned to the target object.
[0069] In this embodiment, the title text is manually reviewed, and the title text that meets the requirements is returned to the user (i.e., the target object) as the final text. The title text that fails the first manual review is modified for the second time using the violation handling model. In this way, the compliance of the advertising title can be effectively improved, the risk of violation can be reduced, and the creativity and attractiveness of the advertising content can be ensured, thereby improving user satisfaction. The title text that still fails to pass the manual review after the second revision or the initial title text that the user chooses not to accept the modification will be directly returned, which can simplify the review process and improve the efficiency of the review process.
[0070] In order to make this application easier to understand, the following Figure 7 An exemplary application is provided. S11, receiving an initial advertisement title (ie, initial title text) input by an advertiser O (ie, a target object).
[0071] S12, judging whether the initial advertisement title violates the regulations through the violation determination model.
[0072] If yes, go to step S13, otherwise go to step S21.
[0073] S13, modify the initial advertisement title through the violation processing model to obtain multiple compliant modified titles.
[0074] S14, inputting multiple compliant modified titles into a text quality assessment model to obtain text quality parameters of each compliant modified title.
[0075] S15, determining the compliant modified title with the highest text quality parameter as the target advertisement title.
[0076] S16, sending the violation reason of the initial advertisement title and the target advertisement title to the advertiser O for confirmation, and determining whether the confirmation result is acceptance of modification.
[0077] If yes, go to step S17, otherwise go to step S22.
[0078] S17, transferring the target advertisement title to a manual review node for review, and determining whether the manual review result is approved.
[0079] If no, go to step S18, if yes, go to step S24.
[0080] S18, modify the target advertisement title through the violation processing model to obtain a secondary modified title (i.e., secondary modified text).
[0081] S19, transfer the second modified title flow to the manual review node for review, and determine whether the manual review result is passed.
[0082] If no, proceed to step S20, and if yes, proceed to step S25.
[0083] S20, returning the second modified title and the manual review result corresponding to the second modified title to the advertiser O.
[0084] S21, notifying advertiser O that the initial advertisement title has passed the review.
[0085] S22, transferring the initial advertisement title to the manual review node for review, and determining whether the manual review result is approved.
[0086] If no, go to step S23, if yes, go to step S21.
[0087] S23, returning the manual review result corresponding to the initial advertisement title to the advertiser O.
[0088] S24, notifying advertiser O that the target advertisement title has passed the review.
[0089] S25, notifying advertiser O that the second modified title has passed the review.
[0090] Embodiment 2 Figure 8 The block diagram of the text violation processing device according to the second embodiment of the present application is schematically shown. The device can be divided into one or more program modules, one or more program modules are stored in a storage medium, and are executed by one or more processors to complete the embodiment of the present application. The program module referred to in the embodiment of the present application refers to a series of computer program instruction segments that can complete specific functions. The following description will specifically introduce the functions of each program module in this embodiment. Figure 8 As shown, the device 1000 may include: an acquisition module 1100, a first input module 1200, and a second input module 1300, wherein: An acquisition module 1100 acquires the initial title text input by the target object; The first input module 1200 is used to input the initial title text into a pre-trained violation determination model to obtain a determination result; wherein, if the determination result is a violation, the determination result includes the reason for the violation of the initial title text; The second input module 1300 is used to input the initial title text and the violation reason into a pre-trained violation processing model to obtain a target title text for replacing the initial title text when the determination result is violation; The target title text is used to be returned to the target object or transferred to a manual review node for review.
[0091] As an optional embodiment, the apparatus 1000 further includes a first flow transfer module, configured to: When the determination result is that there is no violation, the initial title text is transferred to the manual review node.
[0092] As an optional embodiment, the violation processing model is used to output multiple modified texts according to the violation reason and the initial title text, and the second input module 1300 is further used to: Inputting the plurality of modified texts into the violation determination model to screen out a plurality of compliant modified texts from the plurality of modified texts; The target title text is determined based on the multiple compliant modification texts.
[0093] As an optional embodiment, the second input module 1300 is further used for: Inputting the plurality of compliant modified texts into a pre-trained text quality assessment model respectively to obtain a text quality parameter of each of the compliant modified texts; The compliant modified text with the highest text quality parameter among the multiple compliant modified texts is determined as the target title text.
[0094] As an optional embodiment, the apparatus 1000 further includes a sending module, configured to: Sending the determination result and the target title text to the target object; receiving a confirmation result of the target object with respect to the determination result and the target title text; If the confirmation result is to accept the modification, the target title text is transferred to the manual review node; When the confirmation result is that the modification is not accepted, the initial title text is transferred to the manual review node.
[0095] As an optional embodiment, the apparatus 1000 further includes a second flow transfer module, configured to: Transfer the text to be reviewed to a manual review node, and receive the manual review result; wherein the text to be reviewed is the target title text or the initial title text transferred to the manual review node, and when the manual review result is failure to review, the manual review result includes the manual review reason for the text to be reviewed; If the manual review result is that the review is passed, the text to be reviewed is returned to the target object; In the case where the manual review result is that the review is not passed, determining whether the text to be reviewed is an initial title text that does not accept modification; In the case where the text to be reviewed is an initial title text that does not accept modification, returning the manual review result to the target object; In the case where the text to be reviewed is not the initial title text that does not accept modification, the manual review reason of the text to be reviewed and the text to be reviewed are input into a pre-trained violation processing model to obtain a secondary modified text for replacing the text to be reviewed; Transferring the second modified text to the manual review node to receive the manual review result and manual review reason corresponding to the second modified text; In the case where the manual review result corresponding to the secondary revised text is failure to pass the review, the secondary revised text and the manual review result corresponding to the secondary revised text are returned to the target object.
[0096] Embodiment 3 Fig. 9The schematic diagram of the hardware architecture of a computer device 10000 suitable for implementing the text violation processing method according to the third embodiment of the present application is schematically shown. In some embodiments, the computer device 10000 may be a terminal device such as a smart phone, a wearable device, a tablet computer, a personal computer, a vehicle terminal, a game console, a virtual device, a workbench, a digital assistant, a set-top box, a robot, etc. In other embodiments, the computer device 10000 may be a rack server, a blade server, a tower server, or a cabinet server (including an independent server, or a server cluster composed of multiple servers), etc. Fig. 9 As shown, the computer device 10000 includes but is not limited to: a memory 10010, a processor 10020, and a network interface 10030 that can communicate with each other through a system bus. Among them: The memory 10010 includes at least one type of computer-readable storage medium, and the readable storage medium includes flash memory, hard disk, multimedia card, card-type memory (such as SD or DX memory), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the memory 10010 can be an internal storage module of the computer device 10000, such as a hard disk or memory of the computer device 10000. In other embodiments, the memory 10010 can also be an external storage device of the computer device 10000, such as a plug-in hard disk equipped on the computer device 10000, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. Of course, the memory 10010 can also include both the internal storage module of the computer device 10000 and its external storage device. In this embodiment, the memory 10010 is generally used to store the operating system and various application software installed in the computer device 10000, such as the program code of the text violation processing method, etc. In addition, the memory 10010 can also be used to temporarily store various data that have been output or will be output.
[0097] In some embodiments, the processor 10020 may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other chips. The processor 10020 is generally used to control the overall operation of the computer device 10000, such as performing control and processing related to data interaction or communication with the computer device 10000. In this embodiment, the processor 10020 is used to run the program code stored in the memory 10010 or process data.
[0098] The network interface 10030 may include a wireless network interface or a wired network interface, and the network interface 10030 is generally used to establish a communication link between the computer device 10000 and other computer devices. For example, the network interface 10030 is used to connect the computer device 10000 to an external terminal through a network, and to establish a data transmission channel and a communication link between the computer device 10000 and the external terminal. The network may be a wireless or wired network such as an intranet, the Internet, the Global System of Mobile communication (GSM), Wideband Code Division Multiple Access (WCDMA), 4G network, 5G network, Bluetooth, Wi-Fi, etc.
[0099] It should be pointed out that Fig. 9 Only a computer device having components 10010 - 10030 is shown, but it should be understood that implementation of all of the components shown is not a requirement, and more or fewer components may alternatively be implemented.
[0100] In this embodiment, the text violation processing method stored in the memory 10010 can also be divided into one or more program modules and executed by one or more processors (such as processor 10020) to complete the embodiment of the present application.
[0101] Embodiment 4 An embodiment of the present application also provides a computer-readable storage medium, on which a computer program is stored, wherein when the computer program is executed by a processor, the steps of the text violation processing method in the embodiment are implemented.
[0102] In this embodiment, the computer-readable storage medium includes flash memory, hard disk, multimedia card, card-type memory (for example, SD or DX memory, etc.), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, disk, optical disk, etc. In some embodiments, the computer-readable storage medium may be an internal storage unit of a computer device, such as a hard disk or memory of the computer device. In other embodiments, the computer-readable storage medium may also be an external storage device of a computer device, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. equipped on the computer device. Of course, the computer-readable storage medium may also include both the internal storage unit of the computer device and its external storage device. In this embodiment, the computer-readable storage medium is generally used to store an operating system and various application software installed on the computer device, such as the program code of the text violation processing method in the embodiment. In addition, the computer-readable storage medium may also be used to temporarily store various types of data that have been output or are to be output.
[0103] Embodiment 5 An embodiment of the present application also provides a computer program product, including a computer program, which implements the method in the above embodiment when executed by a processor.
[0104] Obviously, those skilled in the art should understand that the modules or steps of the above-mentioned embodiments of the present application can be implemented by general-purpose computer devices, they can be concentrated on a single computer device, or distributed on a network composed of multiple computer devices, optionally, they can be implemented by executable program codes of computer devices, so that they can be stored in a storage device and executed by the computer device, and in some cases, the steps shown or described can be executed in a different order from that herein, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. In this way, the embodiments of the present application are not limited to any specific combination of hardware and software.
[0105] It should be noted that the above are only preferred embodiments of the present application, and the patent protection scope of the present application is not limited thereto. Any equivalent structure or equivalent process transformation made using the contents of the specification and drawings of the present application, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present application.
Claims
1. A method for processing text violations, characterized in that: The method comprises: Get the initial title text of the target object input; Inputting the initial title text into a pre-trained violation determination model to obtain a determination result; wherein, if the determination result is a violation, the determination result includes the reason for the violation of the initial title text; In the case where the determination result is a violation, the initial title text and the violation reason are input into a pre-trained violation processing model to obtain a target title text for replacing the initial title text; The target title text is used to be returned to the target object or transferred to a manual review node for review.
2. The method according to claim 1, characterized in that The method further comprises: When the determination result is that there is no violation, the initial title text is transferred to the manual review node.
3. The method according to claim 1, characterized in that The violation processing model is used to output a plurality of modified texts according to the violation reason and the initial title text; and obtain a target title text for replacing the initial title text, including: Inputting the plurality of modified texts into the violation determination model to screen out a plurality of compliant modified texts from the plurality of modified texts; The target title text is determined based on the multiple compliant modification texts.
4. The method according to claim 3, characterized in that According to the plurality of compliance modification texts, the target title text is determined, including: Inputting the plurality of compliant modified texts into a pre-trained text quality assessment model respectively to obtain a text quality parameter of each of the compliant modified texts; The compliant modified text with the highest text quality parameter among the multiple compliant modified texts is determined as the target title text.
5. The method according to claim 1, characterized in that The method further comprises: Sending the determination result and the target title text to the target object; receiving a confirmation result of the target object with respect to the determination result and the target title text; If the confirmation result is to accept the modification, the target title text is transferred to the manual review node; When the confirmation result is that the modification is not accepted, the initial title text is transferred to the manual review node.
6. The method according to claim 5, characterized in that The method further comprises: Transfer the text to be reviewed to a manual review node, and receive the manual review result; wherein the text to be reviewed is the target title text or the initial title text transferred to the manual review node, and when the manual review result is failure to review, the manual review result includes the manual review reason for the text to be reviewed; If the manual review result is that the review is passed, the text to be reviewed is returned to the target object; In the case where the manual review result is that the review is not passed, determining whether the text to be reviewed is an initial title text that does not accept modification; In the case where the text to be reviewed is an initial title text that does not accept modification, returning the manual review result to the target object; In the case where the text to be reviewed is not the initial title text that does not accept modification, the manual review reason of the text to be reviewed and the text to be reviewed are input into a pre-trained violation processing model to obtain a secondary modified text for replacing the text to be reviewed; Transferring the second modified text to the manual review node to receive the manual review result and manual review reason corresponding to the second modified text; In the case where the manual review result corresponding to the secondary revised text is failure to pass the review, the secondary revised text and the manual review result corresponding to the secondary revised text are returned to the target object.
7. A text violation processing device, characterized in that: The device comprises: The acquisition module obtains the initial title text input by the target object; A first input module is used to input the initial title text into a pre-trained violation determination model to obtain a determination result; wherein, if the determination result is a violation, the determination result includes the violation reason of the initial title text; A second input module is used for inputting the initial title text and the violation reason into a pre-trained violation processing model to obtain a target title text for replacing the initial title text when the determination result is a violation; The target title text is used to be returned to the target object or transferred to a manual review node for review.
8. A computer device, characterized in that: include: at least one processor; and a memory communicatively connected to the at least one processor; wherein: The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 6.
9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and when the computer instructions are executed by a processor, the method according to any one of claims 1 to 6 is implemented.
10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.