Comment generation information processing method, system and device based on AIGC and program product

Abstract

The invention relates to the technical field of artificial intelligence, in particular to an AIGC-based comment generation information processing method, which is characterized in that through combination of multi-task learning and a generation model, a system can generate a plurality of comments according to user requirements and flexibly adjust the number, length, emotional tendency and theme direction of the comments. Meanwhile, the system can capture internet information in real time, dynamically rewrite the generated comments, and ensure the timeliness, correlation and accuracy of the comments. The sensitive word filtering module ensures that the generated comments meet the laws and regulations of the platform, and generation of improper content is avoided. Through an iterative feedback mechanism, the system can continuously optimize the comment generation quality, improve the user experience, improve the comment generation efficiency and quality of various platforms, and meet continuously changing market demands and user expectations.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and specifically to an AIGC-based comment generation information processing method, system, device and program product. Background Art

[0002] With the popularization and development of communication technology and mobile terminals, more and more users use online media or Internet platforms to browse information content in various forms such as text, audio, video, etc., such as social platforms, e-commerce platforms, news websites, and video platforms. Generally, online media or Internet platforms provide users with a comment function for commenting on the presented information content, as well as a reply function for evaluating / replying to the user's comments.

[0003] With the rapid development of Internet platforms such as social media, e-commerce platforms and news portals, how to efficiently and accurately generate comments that meet user needs has become an urgent problem to be solved. At present, with the rapid development of artificial intelligence content generation technology, the automatic comment generation system has gradually become an important part of the Internet platform. Especially in the fields of social media, e-commerce platforms and news portals, comment generation technology is widely used, such as Weibo comments, takeaway reviews, etc., which greatly improves the content production efficiency of the platform and enhances the user experience.

[0004] However, although the existing comment generation methods based on generative models can generate natural and fluent comments, they still have the following technical problems: the existing comment generation system can usually only generate a single comment, and it is still difficult to generate multiple comments that meet specific requirements in a short period of time; the length control in the comment generation process is often not accurate, which may generate comments that are too long or too short, resulting in the comment format not meeting the platform specifications; the existing generation methods lack an effective guidance mechanism, especially in terms of emotional tendency and topic guidance, the generated comments may deviate greatly from the target; most of the existing comment generation systems fail to capture relevant information on the Internet in real time for optimization, resulting in the lack of timeliness and relevance of the generated comments; and the generated comments may contain inappropriate content, such as sensitive words, illegal content or vulgar language. How to effectively filter these contents and ensure the compliance of comments has become a major challenge facing the current system.

[0005] In view of the above problems, the present invention proposes a comment generation information processing method based on AIGC, which combines the technologies of multiple comment generation, comment quantity and length control, sentiment and topic guidance, real-time information acquisition and rewriting, sensitive word filtering and compliance checking, aiming to provide a comprehensive and efficient comment generation system. Summary of the invention

[0006] In view of the deficiencies in the prior art, the present invention provides a comment generation information processing method, system, device and program product based on AIGC, which solves the problems of the existing generation method such as the single number of comments, inaccurate control of comment length, large deviation between the generated comments and the target, and lack of timeliness and relevance.

[0007] To achieve the above objectives, the present invention is implemented through the following technical solutions: a comment generation information processing method based on AIGC, comprising the following steps:

[0008] S1. User input

[0009] When users input specific keywords or event topics, the comments will be generated based on the information to ensure that the comments are highly relevant to the keywords or events.

[0010] S2. Comment content guidance and emotional tendency control

[0011] Conduct real-time sentiment analysis on generated comments and adjust the sentiment of comments according to the sentiment orientation target set by the user;

[0012] S3. Obtain network information in real time to rewrite comments

[0013] Capture information related to input parameters on the Internet in real time and optimize the generated comments;

[0014] S4. Sensitive word filtering

[0015] After the comment is generated, a sensitive word check is performed. If there is any non-compliant content, the comment is discarded and regenerated until the comment is output;

[0016] S5. Iterable memory feedback mechanism

[0017] Each user interaction with the generated comments will be fed back to the generation model as new feedback information, thereby continuously optimizing the model's generation capabilities.

[0018] Preferably, the S1 step is specifically as follows:

[0019] The user specifies the number of comments to be generated by entering specific keywords or event topics and setting the number of generation parameters, so that the system generates multiple non-repetitive comments based on the same event or input keywords. The number of characters in each comment is controlled by the adaptive limit range of the number of characters specified by the user.

[0020] Preferably, the S3 step is specifically as follows:

[0021] By accessing real-time information crawling tools, the system automatically crawls the latest information related to events, products, and news from the Internet to provide the latest contextual support for comments. Based on the crawled real-time data, the system can dynamically adjust the generated comment content to ensure the timeliness and relevance of the comments.

[0022] Preferably, the S4 step is specifically as follows:

[0023] In order to ensure that the generated comments comply with the platform's laws, regulations and community norms, the system's built-in sensitive word library will be used to check each comment for sensitive words after it is generated. If non-compliant content is found, the system will discard the comment and regenerate a comment that meets the requirements until the comment is output.

[0024] Preferably, the S5 step is specifically as follows:

[0025] By collecting interactive data on comments and user feedback, these data will reflect users' evaluation and needs for comments. Based on user feedback, the system will adjust the parameters of the generation model to ensure that the subsequently generated comments are more in line with user needs.

[0026] A comment generation information processing system, comprising:

[0027] A comment generation module is used to generate multiple comments that meet user requirements based on the parameter information input by the user;

[0028] Sentiment analysis and guidance module, used to analyze the sentiment tendency of the generated comments and adjust the sentiment of the comments according to user needs;

[0029] Information capture and rewriting module, used to capture the latest information on the Internet in real time, and rewrite and optimize the generated comments in real time;

[0030] Sensitive word detection and filtering module, which integrates a multi-level sensitive word filtering mechanism to detect the generated comment content in real time and conduct compliance review;

[0031] Sensitive word library, which contains sensitive words in multiple fields and is used to check the generated comment content;

[0032] The information feedback optimization module is used to perform machine learning on the comment feedback information and optimize the generated comment content.

[0033] A comment generation information processing device, comprising:

[0034] Terminal equipment, hardware facilities used to generate experience review content;

[0035] A data processor is used to execute the work of the comment generation module and generate relevant comment content according to user requirements;

[0036] The networking module is used to connect the system to the network to capture real-time information.

[0037] The present invention provides a review generation information processing method, system, device and program product based on AIGC. It has the following beneficial effects:

[0038] 1. Through the combination of multi-task learning and generative models, the present invention enables the system to generate multiple comments according to user needs and flexibly adjust the number, length, sentiment and topic of comments. At the same time, the system can capture Internet information in real time and dynamically rewrite the generated comments to ensure the timeliness, relevance and accuracy of the comments. The sensitive word filtering module ensures that the generated comments comply with the legal and regulatory requirements of the platform and avoids the generation of inappropriate content. Through the iterative feedback mechanism, the system can continuously optimize the quality of comment generation and improve user experience. It can be widely used in social platforms, e-commerce platforms, news portals, video platforms and other fields to improve the efficiency and quality of comment generation on various platforms and meet the ever-changing market demands and user expectations.

[0039] 2. The present invention can generate multiple different comments and flexibly control the number of comments generated, thereby ensuring the diversity and comprehensiveness of the generated content. Users can set the number of comments generated according to actual needs, and the system will automatically handle the diversity of the generated comments to avoid duplication or content redundancy. Compared with the limitation of existing systems that can only generate a single comment, the efficiency of comment generation is significantly improved.

[0040] 3. The present invention can generate comments with clear emotional tendencies and themes according to user needs through the sentiment analysis and guidance module. Whether it is positive, negative or neutral sentiment, the system can accurately generate comments that meet the requirements. In addition, the system can also guide the content of comments based on the keywords or event themes provided by the user to ensure the relevance and accuracy of the content of the comments to the event or product.

[0041] 4. The present invention can obtain the latest information on the Internet in real time through the information capture and real-time rewriting module, and rewrite and optimize the generated comments based on this information. The system can flexibly respond to the rapidly changing network environment, ensure the timeliness of the comments, keep up with social hot spots, news events or product changes, and enhance the relevance and influence of the comments.

[0042] 5. The sensitive word detection and filtering module of the present invention ensures that each generated comment complies with the laws and regulations and the platform's provisions. Through the built-in multi-layer sensitive word library, the system can identify and automatically delete comments that do not meet the platform's requirements, avoiding inappropriate content such as sensitive words, violence, pornography, vulgar language, etc., thereby ensuring the compliance of the comments.

[0043] 6. The iterative memory feedback mechanism of the present invention can continuously optimize the quality of comment generation based on the interactive feedback of users on comments. Users provide feedback to the system through behaviors such as likes, comment quality ratings, and sharing. The system adjusts the parameters of the generation model based on these feedbacks to continuously adapt to the needs of users and improve the quality and accuracy of subsequent comments. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 It is a schematic diagram of the method flow of the present invention. DETAILED DESCRIPTION

[0045] The technical solutions in the embodiments of the present invention are described clearly and completely below. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0046] Embodiment 1:

[0047] The embodiment of the present invention provides a review generation information processing system based on AIGC, including:

[0048] The comment generation module is used to generate multiple comments that meet user requirements based on the parameter information input by the user. This module generates comment content based on a large-scale pre-trained generation model to ensure the diversity and accuracy of the comment content in terms of sentiment, length, quantity, topic, etc.

[0049] Sentiment analysis and guidance module, which uses the sentiment analysis model to analyze the sentiment tendency of the generated comments and adjusts the sentiment of the comments according to user needs. For example, when generating negative comments, the model will pay special attention to the negative sentiment words that may be contained in the comments.

[0050] The information crawling and rewriting module is used to crawl the latest information on the Internet in real time, and rewrite and optimize the generated comments in real time to ensure the timeliness and relevance of the comments. The information crawling and rewriting module obtains the latest data support through search engine API, news crawling, social media analysis and other means;

[0051] Sensitive word detection and filtering module, which integrates a multi-level sensitive word filtering mechanism to detect the generated comment content in real time and conduct compliance review. The sensitive word detection and filtering module supports sensitive word libraries in multiple fields and can be customized according to the compliance requirements of different platforms;

[0052] Sensitive word library, which contains sensitive words in multiple fields, including religion, violence, pornography, and vulgar language, is used to check the generated comment content; the sensitive word library can be updated regularly as needed to ensure its wide coverage and timeliness.

[0053] The information feedback optimization module is used to perform machine learning on the comment feedback information and optimize the generated comment content.

[0054] Embodiment 2:

[0055] An embodiment of the present invention provides a comment generation information processing device, including:

[0056] Terminal devices are hardware facilities used to generate review content, usually mobile devices or PC devices;

[0057] A data processor is used to execute the work of the comment generation module and generate relevant comment content according to user requirements;

[0058] The networking module is used to connect the system to the network to capture real-time information.

[0059] Embodiment three:

[0060] like Figure 1 As shown, the embodiment of the present invention provides a comment generation information processing method based on AIGC, comprising the following steps:

[0061] S1. User input

[0062] When users input specific keywords or event topics, the system will generate comments based on the information to ensure that the comments are highly relevant to the keywords or events. For example, when users request to generate comments about a product, the system generates comments that meet the requirements based on the characteristics of the product. Users can also specify the number of comments to be generated by setting the number of generated parameters, so that the system can generate multiple non-repetitive comments based on the same event or input, increasing the richness of the content. The system can also automatically adjust the number of generated comments according to the platform's requirements for the number of comments. The number of characters in each comment is controlled by the user-specified adaptive limit range of the number of characters in the comment. By adjusting these parameters, the system can generate comments that meet the platform specifications and avoid generating content that is too long or too short. In addition, the comment generation module combines the generation model by introducing multi-task learning, which can not only generate comment text, but also add additional information such as emotional tendency and subject direction to each comment. The system generates multiple comments based on the input text, and the content of each comment has a certain diversity to avoid duplication.

[0063] S2. Comment content guidance and emotional tendency control

[0064] The system uses the sentiment analysis and guidance module to perform real-time sentiment analysis on the generated comments based on the sentiment classifier of BERT or RoBERTa, and adjusts the sentiment of the comments according to the sentiment tendency target set by the user. Whether the user needs positive, negative or neutral comments, the system can meet the needs. This step provides a flexible sentiment tendency control mechanism that can guide the generated comments to develop in the direction of specific emotions and topics.

[0065] S3. Obtain network information in real time to rewrite comments

[0066] Crawl information related to the input parameters on the Internet in real time and optimize the generated comments. Specifically: the system automatically crawls the latest information related to events, products, and news from the Internet by accessing real-time information crawling tools (such as search engine crawlers, API interfaces, etc.). When an event occurs, the system can obtain relevant reports or user discussions in real time to provide the latest contextual support for the comments. In addition, based on the captured real-time data, the system can also dynamically adjust the generated comment content; for example: when the discussion heat of an event changes, the system will automatically adjust the comment content to reflect the new information or opinions to ensure the timeliness and relevance of the comments. Through this step, the comment generation system not only generates comments based on user input, but also crawls relevant information on the Internet in real time to optimize the generated comments.

[0067] S4. Sensitive word filtering

[0068] In order to ensure that the generated comments comply with the platform's laws, regulations and community norms, the system's built-in sensitive word library will be used to check each comment for sensitive words after it is generated. If non-compliant content is found, the system will discard the comment and regenerate a comment that meets the requirements until the comment is output. The system can adjust the sensitive word detection rules according to the specific requirements of the platform to ensure that each comment meets the platform's compliance standards. This step can effectively identify and delete non-compliant content in comments to avoid generating violent, pornographic, vulgar and other content.

[0069] S5. Iterable memory feedback mechanism

[0070] Each time a user interacts with a generated comment, it will be fed back to the generation model as new feedback information, thereby continuously optimizing the model's generation capability. The system collects user feedback information by collecting interaction data on comments (such as likes, comments, shares, etc.). These data will reflect the user's evaluation and needs for the comments. Based on user feedback, the system will adjust the parameters of the generation model to ensure that subsequently generated comments are more in line with user needs. For example, if a certain type of comment receives a higher score or likes, the system will give priority to generating similar comment styles or emotional tendencies.

[0071] This comment generation information processing method has broad application prospects and can effectively improve the efficiency and quality of comment generation on various platforms. It is suitable for the following scenarios:

[0072] Social media platforms: Users interact frequently on social media platforms, and the system needs to quickly generate a large number of comments that match emotions, topics, and user interests. The present invention can generate multiple comments in a short period of time, and can make real-time adjustments based on hot topics, sentiment analysis, and user feedback, thereby improving the platform's interactivity and user experience.

[0073] E-commerce platforms: There are a large number of product reviews on e-commerce platforms, and the quality requirements are high. The system can generate multiple reviews for each product, covering different sentiment tendencies and detailed descriptions. Through sentiment analysis and topic guidance, the reviews can accurately reflect the user's feelings about the product, and enhance the user's trust in the product and willingness to buy.

[0074] News portals: News websites need to process a large number of comments about news events in real time. Through the information capture and real-time rewriting module, the system can quickly obtain the latest news information and generate comments for users that are consistent with the news content and emotional tendencies, thereby improving user interaction and the content quality of the website.

[0075] Video platforms: The generation of comments on video platforms needs to take into account not only the video content itself, but also the emotional response of the audience. Through the systematic sentiment analysis and feedback mechanism, the platform can generate comments for each video that match the audience's emotions, and adjust the comment content in real time to ensure the authenticity and timeliness of the comments.

[0076] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A review generation information processing method based on AIGC, characterized by: The following steps are involved: S1. User input When users input specific keywords or event topics, the comments will be generated based on the information to ensure that the comments are highly relevant to the keywords or events. S2. Comment content guidance and emotional tendency control Conduct real-time sentiment analysis on generated comments and adjust the sentiment of comments according to the sentiment orientation target set by the user; S3. Obtain network information in real time to rewrite comments Capture information related to input parameters on the Internet in real time and optimize the generated comments; S4. Sensitive word filtering After the comment is generated, a sensitive word check is performed. If there is any non-compliant content, the comment is discarded and regenerated until the comment is output; S5. Iterable memory feedback mechanism Each user interaction with the generated comments will be fed back to the generation model as new feedback information, thereby continuously optimizing the model's generation capabilities.

2. The AIGC-based comment generation information processing method according to claim 1 is characterized by: The S1 step is specifically as follows: The user specifies the number of comments to be generated by entering specific keywords or event topics and setting the number of generation parameters, so that the system generates multiple non-repetitive comments based on the same event or input keywords. The number of characters in each comment is controlled by the adaptive limit range of the number of characters specified by the user.

3. The AIGC-based comment generation information processing method according to claim 1 is characterized by: The S3 step is specifically as follows: By accessing real-time information crawling tools, the system automatically crawls the latest information related to events, products, and news from the Internet to provide the latest contextual support for comments. Based on the crawled real-time data, the system can dynamically adjust the generated comment content to ensure the timeliness and relevance of the comments.

4. The AIGC-based comment generation information processing method according to claim 1 is characterized by: The S4 step is specifically as follows: In order to ensure that the generated comments comply with the platform's laws, regulations and community norms, the system's built-in sensitive word library will be used to check each comment for sensitive words after it is generated. If non-compliant content is found, the system will discard the comment and regenerate a comment that meets the requirements until the comment is output.

5. The AIGC-based comment generation information processing method according to claim 1 is characterized by: The S5 step is specifically as follows: By collecting interactive data on comments and user feedback, these data will reflect users' evaluation and needs for comments. Based on user feedback, the system will adjust the parameters of the generation model to ensure that the subsequently generated comments are more in line with user needs.

6. A comment generation information processing system, characterized in that: include: A comment generation module is used to generate multiple comments that meet user requirements based on the parameter information input by the user; Sentiment analysis and guidance module, used to analyze the sentiment tendency of the generated comments and adjust the sentiment of the comments according to user needs; Information capture and rewriting module, used to capture the latest information on the Internet in real time, and rewrite and optimize the generated comments in real time; Sensitive word detection and filtering module, which integrates a multi-level sensitive word filtering mechanism to detect the generated comment content in real time and conduct compliance review; Sensitive word library, which contains sensitive words in multiple fields and is used to check the generated comment content; The information feedback optimization module is used to perform machine learning on the comment feedback information and optimize the generated comment content.

7. A comment generation information processing device, characterized in that: include: Terminal equipment, hardware facilities used to generate experience review content; The data processor is used to execute the work of the comment generation module and generate relevant comment content according to user requirements; The networking module is used to connect the system to the network to capture real-time information.

8. An electronic device, characterized in that: include: one or more processors; A storage device for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the method according to any one of claims 1 to 5.

9. A readable storage medium, characterized in that: The readable storage medium stores a program or instruction, and when the program or instruction is executed by the processor, the steps of the comment generation information processing method as described in any one of claims 1-5 are implemented.

10. A computer program product, characterized in that: When the instructions in the computer program product are executed by a processor of an electronic device, the electronic device executes the steps of the comment generation information processing method as described in any one of claims 1 to 5.

Images