News writing system and method based on artificial intelligence

Abstract

The invention provides a news writing system and method based on artificial intelligence, and relates to the technical field of information processing.The news writing system based on artificial intelligence comprises a user interaction module, a data processing module, a big language model interaction module and an artificial intelligence module, the user interaction module is in signal connection with the data processing module, and the data processing module is in signal connection with the big language model interaction module; the large language model interaction module is in signal connection with the data processing module and the artificial intelligence module; and the artificial intelligence module constructs a relational database. The news writing method adopts the news writing system based on artificial intelligence to automatically generate news manuscripts. According to the scheme, full coverage of news collecting and editing materials is finally achieved, AI multi-mode application is achieved, the whole news collecting and editing process is simulated, and the limitation of the output length of a large model is broken through; and illegal content and infringement content are screened by using a vector database and a cloud API (Application Program Interface), so that the safety and reliability of the news manuscript content generated by the AI are ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of information processing, and particularly to a news writing system based on artificial intelligence. Background Art

[0002] Artificial intelligence is a new technical science that studies and develops theories, methods, technologies, and application systems for simulating, extending, and expanding human intelligence; it is a branch of computer science that attempts to understand the essence of intelligence and produce a new intelligent machine that can respond in a way similar to human intelligence. The research in this field includes robots, speech recognition, image recognition, natural language processing, and expert systems, etc.; in the news field, artificial intelligence can replace humans in sorting and data processing of news manuscripts.

[0003] In the related prior art, the publication number: CN106776523A, discloses a method and device for generating news bulletins based on artificial intelligence. Among them, the method includes: pre-constructing a template library based on historical news bulletins, where the template library includes the basic frameworks for generating news bulletins for each news type; determining the target news type to which the news event belongs, obtaining the target basic framework of the target news type from the template library, obtaining the structured data of the news event from a preset database, filling the content into the target basic framework according to the structured data to obtain the text of the news bulletin of the news event, generating the title of the news bulletin according to the text, and splicing the title and the text to form the news bulletin.

[0004] The existing system only realizes simple storage and classification of materials and lacks the ability of cross-modal semantic association; for example, the key information in the video (such as human actions, background details) cannot be automatically converted into text or structured data, resulting in the disconnection between the content generated by AI and the original materials.

[0005] Therefore, it is necessary to provide a news writing system based on artificial intelligence to solve the above technical problems. Summary of the Invention

[0006] The present invention provides a news writing system based on artificial intelligence, which solves the problem that the key information in the video cannot be automatically converted into text or structured data in the related technology.

[0007] To solve the above technical problems, the news writing system based on artificial intelligence provided by the present invention includes:

[0008] A user interaction module, a data processing module, a large language model interaction module, and an artificial intelligence module. The user interaction module is signal-connected to the data processing module, and the large language model interaction module is respectively signal-connected to the data processing module and the artificial intelligence module;

[0009] The artificial intelligence module constructs a relational database to record and manage relational data and establish connections for the relationships of uploaded data;

[0010] The artificial intelligence module constructs a vector database to record and manage illegal and infringing data;

[0011] The user interaction module is used for users to upload news manuscript data that needs to be written;

[0012] The data processing module includes a data classification unit, a data cleaning unit, a data storage unit, a data comparison unit, and a data editing unit. The data classification unit is used to classify the styles of news manuscript data; the data cleaning unit is used to clean the news manuscript data; the data storage unit is used to store the news manuscript data after data cleaning; the data comparison unit is responsible for searching and comparing with the vector database, and the data editing unit edits and manages the data recorded therein;

[0013] The artificial intelligence module realizes retrieval-enhanced generation of manuscripts through the large language model interaction module.

[0014] Preferably, the user interaction module includes a file upload unit and a voice conversion unit. The file upload unit is used to upload the recorded manuscript, and the voice conversion unit is used to convert the uploaded voice content into written language.

[0015] Preferably, the recorded manuscript includes recognizable video content, text content, and voice content.

[0016] Preferably, the user interaction module further includes an interaction information display unit for feeding back AI-generated content to the generation window.

[0017] Preferably, the large language interaction module includes a prompt word unit, a retrieval unit, and a loop call unit. The prompt word unit is used to use the pre-recorded prompt words, and the retrieval unit makes multiple loop calls to the cloud large language model through the loop call unit.

[0018] Preferably, the artificial intelligence module includes an AI / API unit and a manuscript vulnerability unit. The AI / API unit is used for intelligent processing of uploaded data and automatic generation of news manuscript content, and the manuscript vulnerability unit is used to detect vulnerabilities and defects in the automatically generated news manuscript content, and automatically gives an alarm prompt after the vulnerability detection fails.

[0019] Preferably, the artificial intelligence module further includes manuscript knowledge optimization, which facilitates automatic optimization of the news manuscript content with unqualified vulnerability detection to generate a finalized manuscript.

[0020] Preferably, it further includes a news release module and a news tracking module. The AI / API unit automatically retrieves the platforms for sending through the retrieval unit. The news release module is used for the display of the finalized manuscript on the allowed release platforms and provides support for one-key multi-platform push. The AI / API unit tracks and feeds back the feedback data published on each platform through the news tracking module.

[0021] Preferably, the news tracking module includes a release platform record, a view count record, a comment content record, and a comment hot discussion record. The release platform record is used to track and locate the platforms where the finalized manuscript is released. The view count record is used to count the number of views of the finalized manuscript from the time of release to the current time. The comment content record is used to screen and record the comment and suggestion information related to the finalized manuscript. The comment hot discussion record is used to screen and record the hot discussion comment content.

[0022] The present invention also provides a news writing method, which uses the above-mentioned news writing system based on artificial intelligence to automatically generate news manuscripts, and specifically includes the following steps:

[0023] S1, uploading news materials and converting file contents;

[0024] S2, calling a large language model to generate news content from news materials to generate a finalized news;

[0025] S3, detecting and optimizing the defects of the finalized news;

[0026] S4, screening for violations and infringements of the optimized finalized news;

[0027] S5, performing secondary optimization on the screened finalized news to determine the finalized manuscript.

[0028] Compared with the related technologies, the news writing system based on artificial intelligence provided by the present invention has the following

[0029] Beneficial effects:

[0030] Finally, it realizes full coverage of news gathering and editing materials, realizes the multi-modal application of AI, simulates the whole process of news gathering and editing, and breaks through the output length limit of the large model; uses a vector database + cloud API to screen for illegal and infringing content to ensure the safety and reliability of the content of the news manuscripts generated by AI. Description of the Drawings

[0031] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the accompanying drawings required for the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on the structures shown in these drawings.

[0032] Figure 1 It is a system block diagram of a relatively optimal embodiment of the news writing system based on artificial intelligence provided by the present invention;

[0033] Figure 2 For Figure 1 The block diagram of the user interaction module shown;

[0034] Figure 3 For Figure 1 The block diagram of the data processing module shown;

[0035] Figure 4 For Figure 1 The block diagram of the large language interaction module shown;

[0036] Figure 5 For Figure 1 The block diagram of the artificial intelligence module shown;

[0037] Figure 6 It is the working process of the news writing system based on artificial intelligence provided by the present invention;

[0038] Figure 7 It is a system block diagram of another relatively optimal embodiment of the news writing system based on artificial intelligence provided by the present invention;

[0039] Figure 8 For Figure 7 The block diagram of the news tracking module shown;

[0040] Figure 9 For Figure 7 The block diagram of the artificial intelligence module shown;

[0041] Figure 10 It is the working principle of another relatively optimal embodiment of the news writing system based on artificial intelligence provided by the present invention.

[0042] The realization of the objectives of the present invention, functional features, and advantages will be further described in conjunction with the embodiments and with reference to the accompanying drawings. Detailed implementation manners

[0043] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts belong to the scope of protection of the present invention.

[0044] The present invention provides a news writing system based on artificial intelligence.

[0045] Embodiment 1:

[0046] Please refer to Figures 1 to 3 , in the first embodiment of the present invention, the news writing system based on artificial intelligence includes:

[0047] A user interaction module, a data processing module, a large language model interaction module, and an artificial intelligence module. The user interaction module is signal-connected to the data processing module, and the large language model interaction module is respectively signal-connected to the data processing module and the artificial intelligence module;

[0048] The artificial intelligence module constructs a relational database to record and manage relational data and establish connections for the relationships of the uploaded data;

[0049] The artificial intelligence module constructs a vector database to record and manage violation and infringement data;

[0050] The user interaction module is used for users to upload news manuscript data that needs to be written;

[0051] The data processing module includes a data classification unit, a data cleaning unit, a data storage unit, a data comparison unit, and a data editing unit. The data classification unit is used to classify the style of news manuscript data; the data cleaning unit is used to clean the news manuscript data; the data storage unit is used to store the news manuscript data after data cleaning; the data comparison unit is responsible for searching and comparing with the vector database, and the data editing unit edits and manages the data recorded therein;

[0052] The artificial intelligence module realizes retrieval-enhanced generation of manuscripts through the large language model interaction module.

[0053] The workflow of the large language model interaction module fully simulates the entire process of real content generation:

[0054] Structure selection - Title writing - Lead writing - Outline writing - Grammar polishing - Ending writing.

[0055] By circularly calling the large language model and storing and combining step by step, a complete content is finally formed.

[0056] The large language model interaction layer of this system overcomes the bottleneck that the current large language model can only generate limited content at a time through a loop-forked workflow. In the generation of communication and in-depth report manuscript content, it can generate more than 10,000 words of episode text at a time.

[0057] Ultimately, we will achieve full coverage of news editing and gathering materials, realize AI multimodal application, simulate the entire news editing and gathering process, and break through the output length limit of large models; use vector database + cloud API to screen for illegal and infringing content, and ensure the security and reliability of AI-generated news content.

[0058] Further, such as Figure 2 As shown, the user interaction module includes a file uploading unit and a voice conversion unit. The file uploading unit is used to upload the recording manuscript, and the voice conversion unit is used to convert the uploaded voice content into language and text.

[0059] At the user interaction layer, the drafts, notes, contemporaneous materials, ideas and other materials obtained during the user interview are collected in the form of digital assets and sent to the data processing layer for classification, cleaning and storage.

[0060] For the first time, the boundaries between text, audio, and video have been broken down, allowing AI to generate content that users need more accurately.

[0061] Furthermore, the recorded manuscript includes recognizable video content, text content and voice content.

[0062] The text content includes drafts, notes, concurrent works, ideas and other relevant materials; video content, text content and voice content provide support for data collection.

[0063] It is convenient to collect users' drafts, notes, synchronization, ideas and other related texts and audio and video materials into the system through a one-stop uploading process.

[0064] like Figure 2 As shown, the user interaction module also includes an interactive information display unit for feeding back AI generated content to the generation window.

[0065] It is convenient to feedback AI-generated content in the content generation window, interact with users, and provide technical support for further improving the manuscript.

[0066] Please refer again Figure 4 The large language interaction module includes a prompt word unit, a retrieval unit and a loop calling unit. The prompt word unit is used to use the pre-made prompt words recorded, and the retrieval unit performs multiple loop calls on the cloud large language model through the loop calling unit.

[0067] By building complex workflows (such as Figure 6 ) into the large language model interaction layer, and making multiple loop calls to cloud-based large language models (such as Deepseek, Tongyi Qianwen, etc.) using pre-set prompts and retrieval-augmented generation (RAG), a complete news manuscript is finally obtained.

[0068] It facilitates the automatic extraction, retrieval, and loop calls of prompts and keywords.

[0069] Please refer to Figure 5 again. The artificial intelligence module includes an AI / API unit and a manuscript vulnerability unit. The AI / API unit is used for intelligent processing of uploaded data and automatic generation of news manuscript content. The manuscript vulnerability unit is used for detecting vulnerabilities and defects in the automatically generated news manuscript content, and automatically giving an alarm prompt after the vulnerability detection fails.

[0070] In this embodiment, the vector database records text logic rules and permission logic rules, providing support for self-checking of news manuscript vulnerabilities and infringement detection.

[0071] The manuscript vulnerability analysis includes vulnerability detection of file content and infringement detection of the file. The vulnerability detection facilitates self-checking of text and logical defects in the news manuscript, and gives an alarm prompt after the self-check fails;

[0072] The infringement detection facilitates self-checking of infringement issues in the news manuscript, and gives an alarm prompt after the self-check fails.

[0073] Please refer to Figure 5 again. The artificial intelligence module also includes manuscript knowledge optimization, which facilitates automatic optimization of the news manuscript content with failed vulnerability detection to generate a finalized manuscript.

[0074] After the generated news manuscript undergoes compliance inspection, the compliant news manuscript enters the copyright inspection;

[0075] The non-compliant news manuscript is reorganized by the artificial intelligence module (writing the lead according to the structure - writing the outline according to the structure and lead - polishing and perfecting according to the outline), and then undergoes compliance inspection again;

[0076] After the news manuscript passes the copyright inspection, the passed news manuscript normally outputs the finalized manuscript to the platform for display to news management personnel, and is sent to the corresponding news platform after the inspection and authorization of the news management personnel;

[0077] The news manuscript with potential infringement risks is reorganized by the artificial intelligence module (writing the lead according to the structure - writing the outline according to the structure and lead - polishing and perfecting according to the outline - compliance inspection), and then undergoes copyright inspection again.

[0078] It is convenient to conduct content self-inspection and infringement inspection on the generated news releases. After the inspection, the news releases can be automatically updated until the news releases pass the self-inspection and have no infringement issues, and then the final draft is generated and sent to the news staff.

[0079] Furthermore, through multimodal retrieval technology, key information in the video can be automatically extracted and associated with the text material, improving the accuracy and richness of AI-generated content;

[0080] The loop-forking workflow technology is used to decompose long texts into structured task chains. By calling multiple times and dynamically combining the results, the logical coherence and information integrity problems in long text generation are solved; this is a direct technical response to the "limitations of generative AI in complex content generation";

[0081] This makes up for the shortcomings of relying solely on keyword filtering and achieves accurate identification at the semantic level. This system builds a double protection barrier through technical means, which meets the industry's regulatory requirements for "algorithm transparency" and "content security".

[0082] Ultimately, after each news release is generated, Alibaba Cloud Intelligent Security Audit will be called to review any illegal content. Only content that passes the review can be output. At the same time, vector search and comparison will be performed with the daily updated mainstream media vector database. For AI-generated content that is too close to already published works, the system will require the user to rewrite it to avoid suspected infringement.

[0083] The working principle of the news writing system based on artificial intelligence provided in this embodiment is as follows:

[0084] Step A1, data collection;

[0085] Step A2, data analysis, to determine whether the collected data is sufficient;

[0086] A21, if sufficient, proceed to step A3;

[0087] A22, if it is insufficient, supplement the material to the data collection and re-analyze the data;

[0088] Step A3, choose the writing structure;

[0089] Step A4, write the introduction according to the structure;

[0090] Step A5, write an outline based on the structure and introduction;

[0091] Step A6: polish and improve the outline to generate a news release;

[0092] Step A7, compliance check;

[0093] A71. If it is compliant, proceed to step A8;

[0094] A72. If it is non-compliant, re-enter step A4;

[0095] Step A8. Copyright inspection;

[0096] A81. If it passes, output the finalized manuscript;

[0097] A82. If it fails, re-enter step A4.

[0098] Embodiment 2:

[0099] Please refer to Figures 7 to 8 In the second embodiment of the present invention, the news writing system based on artificial intelligence further includes: a news release module and a news tracking module. The AI / API unit automatically retrieves the platforms that can be sent through the retrieval unit. The news release module is used to display the platforms where the finalized manuscript is allowed to be released, providing support for one-key multi-platform push. The AI / API unit tracks and feeds back the feedback data published on each platform through the news tracking module.

[0100] Through the artificial intelligence module, it is convenient to provide one-key multi-platform push for the finalized manuscript after news writing, facilitating the targeted release and tracking of the finalized manuscript, providing convenience for the operation and running of the system, reducing the operations of management personnel, and ensuring more efficient release while guaranteeing the quality of news generation.

[0101] Please refer to Figures 7 to 9 The news tracking module includes a release platform record, a view count record, a comment content record, and a comment popularity record. The release platform record is used to track and locate the platforms where the finalized manuscript is released. The view count record is used to count the number of views of the finalized manuscript from the time of release to the current time. The comment content record is used to screen and record the comment and suggestion information related to the finalized manuscript. The comment popularity record is used to screen and record the popular comment content.

[0102] Through the release platform record and the view count record, it is convenient to identify the attention of different platforms to news of different styles, providing a data basis for subsequent news generation and push platforms, and facilitating a more intuitive understanding of the news styles that are more popular on each platform.

[0103] Through the comment content record and the comment popularity record, it is convenient to extract and collect the suggestion-type comments and popular comments after the finalized manuscript is released, and to analyze in real time the comment content that has become popular after the finalized manuscript is released, providing support for the comment feedback after news release, and facilitating the timely response and handling by news management personnel.

[0104] Please refer to again Figure 9The artificial intelligence module further includes user comment analysis and manuscript style optimization. The user comment analysis identifies, analyzes, and records based on the suggested content in the comments. The manuscript style optimization automatically optimizes the style of the final manuscript generated subsequently according to the analysis results of the comment suggestions.

[0105] The artificial intelligence module identifies the comment suggestion data of each platform through the user comment analysis;

[0106] The artificial intelligence module automatically analyzes the identified comment suggestion data and determines whether the suggested direction of the comment suggestion data has application value in the field of news writing;

[0107] The artificial intelligence module records the comment suggestion data with application value;

[0108] The artificial intelligence module feeds back the comment suggestion data through the manuscript style optimization. After feedback, the artificial intelligence module adaptively optimizes the writing style in the relational database to facilitate the cyclic iterative optimization of the news writing style.

[0109] Ultimately, while realizing real-time data tracking of the released final manuscript, automatically collecting, identifying, and analyzing the hot discussions and suggestion-based comments, and intelligently iteratively optimizing the writing style in the relational database according to the application value of the suggestion-based comments to improve the subsequent news writing style and quality and increase the reading volume.

[0110] As Figure 10 shown, the working principle of the news writing system based on artificial intelligence provided in this embodiment is as follows:

[0111] S100, the artificial intelligence module automatically generates a final manuscript according to the current news style;

[0112] S200, the artificial intelligence module displays the final manuscript to the recommended news platforms;

[0113] S300, after manual authorization, the final manuscript is sent to multiple news platforms with one click;

[0114] S400, the artificial intelligence module identifies and feeds back the browsing and comment content of the released final manuscript;

[0115] S500, the artificial intelligence module automatically optimizes the generation style of subsequent manuscripts.

[0116] The present invention also provides a news writing method.

[0117] Specifically, for the news writing method, the automatic generation of news manuscripts is performed using the above-mentioned news writing system based on artificial intelligence, which specifically includes the following steps:

[0118] S1, Uploading news materials and converting file content;

[0119] S2, Invoking a large language model to generate news content from news materials and generating finalized news;

[0120] S3, Detecting and optimizing defects in the finalized news;

[0121] S4, Screening for violations and infringements in the optimized finalized news;

[0122] S5, Conducting secondary optimization on the screened finalized news to determine the finalized manuscript.

[0123] The above are only the preferred embodiments of the present invention, and do not limit the patent scope of the present invention. Any equivalent structural transformation made under the concept of the present invention by using the content of the specification and drawings of the present invention, or directly / indirectly applied in other related technical fields, is included in the patent protection scope of the present invention.

Claims

1. An artificial intelligence-based news writing system, characterized in that, Including: A user interaction module, a data processing module, a large language model interaction module, and an artificial intelligence module. The user interaction module is signal-connected to the data processing module, and the large language model interaction module is respectively signal-connected to the data processing module and the artificial intelligence module; The artificial intelligence module constructs a relational database to record and manage relational data and establish connections for the relationships of uploaded data; The artificial intelligence module constructs a vector database to record and manage illegal and infringing data; The user interaction module is used for users to upload news manuscript data that needs to be written; The data processing module includes a data classification unit, a data cleaning unit, a data storage unit, a data comparison unit, and a data editing unit. The data classification unit is used to classify the style of news manuscript data; the data cleaning unit is used to clean the news manuscript data; the data storage unit is used to store the news manuscript data after data cleaning; the data comparison unit is responsible for searching and comparing with the vector database, and the data editing unit edits and manages the data recorded therein; The artificial intelligence module realizes retrieval-enhanced generation of manuscripts through the large language model interaction module.

2. The news writing system based on artificial intelligence according to claim 1, characterized in that, The user interaction module includes a file upload unit and a voice conversion unit. The file upload unit is used to upload the recorded manuscript, and the voice conversion unit is used to convert the uploaded voice content into written language.

3. The news writing system based on artificial intelligence according to claim 2, wherein The recorded manuscript includes recognizable video content, text content, and voice content.

4. The news writing system based on artificial intelligence according to claim 3, characterized in that, The user interaction module further includes an interaction information display unit for feedback of AI-generated content to the generation window.

5. The news writing system based on artificial intelligence according to claim 4, characterized in that, The large language interaction module includes a prompt word unit, a retrieval unit, and a loop call unit. The prompt word unit is used to use the pre-recorded prompt words, and the retrieval unit makes multiple loop calls to the cloud large language model through the loop call unit.

6. The news writing system based on artificial intelligence according to claim 5, characterized in that, The artificial intelligence module includes an AI / API unit and a manuscript vulnerability unit. The AI / API unit is used for intelligent processing of uploaded data and automatic generation of news manuscript content. The manuscript vulnerability unit is used to detect vulnerabilities and defects in the automatically generated news manuscript content, and automatically gives an alarm prompt after the vulnerability detection fails.

7. The news writing system based on artificial intelligence according to claim 6, characterized in that, The artificial intelligence module further includes manuscript knowledge optimization, which facilitates automatic optimization of the news manuscript content with unqualified vulnerability detection to generate a finalized manuscript.

8. The news writing system based on artificial intelligence according to claim 7, characterized in that, It further includes a news release module and a news tracking module. The AI / API unit automatically retrieves the platforms that can be sent through the retrieval unit. The news release module is used for the display of the finalized manuscript on the allowed release platforms and provides support for one-key multi-platform push. The AI / API unit tracks and feeds back the feedback data published on each platform through the news tracking module.

9. The news writing system based on artificial intelligence according to claim 8, wherein, The news tracking module includes publication platform records, view count records, comment content records, and comment popularity records. The publication platform records are used to track and locate the platforms where the finalized manuscripts are published. The view count records are used to record the number of views of the finalized manuscripts from the time of publication to the current time. The comment content records are used to screen and record the comment and suggestion information related to the finalized manuscripts. The comment popularity records are used to screen and record the popular comment content.

10. A news writing method, which uses the artificial intelligence-based news writing system described in any one of claims 1-9 to automatically generate news manuscripts, characterized in that, Specifically, it includes the following steps: S1. Upload news materials and convert the file content; S2. Call the large language model to generate news content from the news materials and generate the finalized news; S3. Conduct defect detection and optimization on the finalized news; S4. Screen the finalized news after optimization for violations and infringement; S5. Conduct secondary optimization on the finalized news after screening to determine the finalized manuscript.

Images