Text abstract generation method, device, equipment, storage medium and product

Abstract

The application discloses a text abstract generation method, device, equipment, storage medium and product, relates to the computer technical field, and discloses a text abstract generation method, comprising: obtaining an initial text; performing a prompt word conversion operation on the initial text according to a preset structured template to obtain a structured prompt word, wherein the structured prompt word reflects the semantics and structure of the initial text; sending the structured prompt word to the large language model and receiving the text abstract information output by the large language model. The application performs a prompt word conversion operation on the initial text to be converted into a text abstract to obtain a structured prompt word reflecting the semantics and structure of the initial text. Since the structured prompt word reflects the semantics and structure of the initial text, the large language model can comprehensively consider the full-text structure and semantics of the initial text according to the structured prompt word, thereby improving the accuracy of text abstract generation.

Description

Technical Field

[0001] This application relates to the field of computer technology, and in particular to methods, apparatus, devices, storage media and products for text summarization. Background Technology

[0002] Currently, with the development of computer technology, the internet and digital information are experiencing explosive growth, posing a challenge to users in reading and processing large amounts of multilingual information. This has created a demand for the rapid and efficient acquisition and understanding of large volumes of textual information. Text summarization, which presents text content in a concise, accurate, and structured manner, can significantly improve information retrieval efficiency. Therefore, how to generate text summaries is a pressing technical problem that needs to be solved.

[0003] In related technologies, key sentences or phrases are typically extracted from the original text to generate summaries. However, this extraction-based summarization method does not comprehensively consider the semantics and structure of the text, and the extracted sentences often fail to accurately represent the full text content, resulting in low accuracy of the generated text summaries.

[0004] The above content is only used to help understand the technical solution of this application and does not represent an admission that the above content is prior art. Summary of the Invention

[0005] The main objective of this application is to provide a text summarization method that aims to solve the technical problem of low accuracy in the generated text summaries.

[0006] To achieve the above objectives, this application proposes a text summarization method applied to a summarization generation platform, wherein a large language model is deployed in the summarization generation platform, and the text summarization method includes:

[0007] Get the initial text;

[0008] According to a preset structured template, the initial text is transformed into prompt words to obtain structured prompt words, wherein the structured prompt words reflect the semantics and structure of the initial text;

[0009] The structured prompt words are sent to the large language model, and the text summary information output by the large language model is received. The text summary information is generated by the large language model by performing a summary text conversion on the structured prompt words based on the summary output instruction.

[0010] Optionally, the step of performing prompt word conversion on the initial text according to a preset structured template to obtain structured prompt words includes:

[0011] Attribute word information of each level is extracted from a preset structured template, wherein the attribute word information is a keyword that identifies different categories of the structured prompt words;

[0012] The initial text and the attribute word information are sent to the large language model, and the structured prompt words output by the large language model are received. The structured prompt words are generated by the large language model extracting the descriptive text of the initial text according to the attribute word information based on the prompt word output instructions, obtaining the target semantic description text, and converting the target semantic description text into prompt words.

[0013] Optionally, the step of sending the structured prompt words to the large language model and receiving the text summary information output by the large language model includes:

[0014] Receive user-inputted requirement information and convert the requirement information into requirement parameters;

[0015] The structured prompts and the limiting parameters are sent to the large language model, and the text summary information output by the large language model is received. The text summary information is generated by the large language model by performing a summary text conversion on the structured prompts based on the summary output instructions and the requirement parameters.

[0016] Optionally, after the steps of sending the structured prompt words to the large language model and receiving the text summary information output by the large language model, the method includes:

[0017] Based on a preset context detection template, the text summary information is detected to obtain the detection result;

[0018] Determine whether the detection result meets the preset detection standard. If the detection result does not meet the detection standard, then adjust the text summary information based on the context detection template to obtain the adjusted text summary information.

[0019] Optionally, after the steps of sending the structured prompt words to the large language model and receiving the text summary information output by the large language model, the method includes:

[0020] The text summary information is displayed on the user's device;

[0021] In response to a text editing instruction issued by a user on the user terminal, wherein the text editing instruction includes a start field, an end field, and editing operation information;

[0022] The fields between the start field and the end field are combined to obtain the target editing field;

[0023] The corresponding editing operation information is used to perform the editing operation on the target editing field to obtain the edited text summary information.

[0024] Optionally, the step of performing the corresponding editing operation on the target editing field to obtain the edited text summary information includes:

[0025] The target editing field and the editing operation information are converted into prompt words to obtain the target prompt words;

[0026] The target prompt word is sent to the large language model, and the edited text summary information output by the large language model is received. The edited text summary information is generated by the large language model performing the corresponding editing operation information on the target editing field based on the summary adjustment instruction.

[0027] Furthermore, to achieve the above objectives, this application also proposes a text summarization generation apparatus, which includes:

[0028] The acquisition module is used to acquire the initial text;

[0029] The conversion module is used to perform prompt word conversion on the initial text according to a preset structured template to obtain structured prompt words, wherein the structured prompt words reflect the semantics and structure of the initial text;

[0030] The generation module is used to send the structured prompt words to the large language model and receive the text summary information output by the large language model, wherein the text summary information is generated by the large language model based on the summary output instruction to perform summary text conversion on the structured prompt words.

[0031] In addition, to achieve the above objectives, this application also proposes a text summarization generation device, the device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the text summarization generation method as described above.

[0032] In addition, to achieve the above objectives, this application also proposes a storage medium, which is a computer-readable storage medium, on which a computer program is stored, and which, when executed by a processor, implements the steps of the text summarization method described above.

[0033] In addition, to achieve the above objectives, this application also provides a computer program product, which includes a computer program that, when executed by a processor, implements the steps of the text summarization method described above.

[0034] One or more technical solutions proposed in this application have at least the following technical effects:

[0035] Unlike related technologies that typically extract key sentences or phrases from the original text to generate summaries, this extraction-based summarization method does not comprehensively consider the semantics and structure of the text. The extracted sentences often fail to accurately represent the full text, resulting in low accuracy of the generated summaries. In contrast, this application uses a summarization platform to perform cue word conversion on the initial text to be converted into a summary, obtaining structured cue words that reflect the semantics and structure of the initial text. These structured cue words are then sent to a large language model, and the platform receives the text summary information output by the large language model. Understandably, on the one hand, the summarization platform uses a deployed large language model to generate text summary information from the initial text, where the large language model can more accurately summarize the entire text. On the other hand, because the structured cue words reflect the semantics and structure of the initial text, after the summarization platform sends these structured cue words to the large language model, the large language model can comprehensively consider the full text structure and semantics of the initial text based on these structured cue words, resulting in more complete text summary information that closely matches the original meaning, thereby improving the accuracy of the generated text summary. Attached Figure Description

[0036] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0037] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0038] Figure 1 A flowchart illustrating the first embodiment of the text digest generation method of this application;

[0039] Figure 2 This is a schematic diagram of the abstract generation module in the text abstract generation method of this application;

[0040] Figure 3 A flowchart illustrating the second embodiment of the text digest generation method of this application;

[0041] Figure 4This is a schematic diagram of the JSON structure of the summary document in the text summary generation method of this application;

[0042] Figure 5 This is a schematic diagram of the AI ​​creation module in the text summary generation method of this application;

[0043] Figure 6 This is a schematic diagram of the module structure of the text digest generation device according to an embodiment of this application;

[0044] Figure 7 This is a schematic diagram of the device structure of the hardware operating environment involved in the text digest generation method in the embodiments of this application.

[0045] The purpose, features, and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0046] It should be understood that the specific embodiments described herein are merely illustrative of the technical solutions of this application and are not intended to limit this application.

[0047] To better understand the technical solution of this application, a detailed description will be provided below in conjunction with the accompanying drawings and specific implementation methods.

[0048] The main solution of this application embodiment is: to obtain initial text; to perform prompt word conversion operation on the initial text according to a preset structured template to obtain structured prompt words, wherein the structured prompt words reflect the semantics and structure of the initial text; to send the structured prompt words to the large language model and receive the text summary information output by the large language model, wherein the text summary information is generated by the large language model based on the summary output instruction to perform summary text conversion on the structured prompt words.

[0049] In this embodiment, the summary generation platform is used as the execution entity. For ease of description, it will be referred to as "platform" below.

[0050] This is because related technologies typically extract key sentences or phrases from the original text to generate summaries. However, this extraction-based summarization method does not comprehensively consider the semantics and structure of the text, and the extracted sentences often fail to accurately represent the full text content, resulting in low accuracy of the generated text summaries.

[0051] This application provides a solution to automate the generation of text summaries and improve the accuracy of text summaries.

[0052] As can be seen from the above embodiments, this application uses a summary generation platform to perform cue word conversion on the initial text to be converted into a text summary, obtaining structured cue words that reflect the semantics and structure of the initial text. These structured cue words are then sent to the large language model, and the platform receives the text summary information output by the large language model. It is understood that, on the one hand, the summary generation platform uses the deployed large language model to generate text summary information from the initial text, where the large language model can more accurately summarize the entire text. On the other hand, since the structured cue words reflect the semantics and structure of the initial text, after the summary generation platform sends the structured cue words to the large language model, the large language model can comprehensively consider the full-text structure and semantics of the initial text based on these structured cue words, making the generated text summary information more complete and closer to the original meaning, thereby improving the accuracy of text summary generation.

[0053] It should be noted that the executing entity in this embodiment can be a computing service device with data processing, network communication, and program execution functions, such as a tablet computer, personal computer, or mobile phone, or an electronic device or terminal system capable of performing the above functions. The following description uses a summary generation platform as an example to illustrate this embodiment and the subsequent embodiments.

[0054] Based on this, embodiments of this application provide a text summarization method, referring to... Figure 1 , Figure 1 This is a flowchart illustrating the first embodiment of the text digest generation method of this application.

[0055] In this embodiment, the text summarization generation method includes steps S100 to S300:

[0056] Step S100: Obtain the initial text;

[0057] It should be noted that the initial text refers to the text to be converted into a text summary. The text summarization method of this application is applied to a summary generation platform, in which a large language model is deployed. It is understood that the above-mentioned text summarization platform is a tool or service that utilizes artificial intelligence technology, especially Natural Language Processing (NLP) technology, to automatically compress long texts into short summaries. These platforms can quickly extract key information from text, generate summaries, and help users save time and improve reading and comprehension efficiency. The working principle of text summarization platforms is usually based on language models. These models learn patterns, grammar, and vocabulary by studying large amounts of text data, and then generate human-like text based on given prompts or input. There are two main methods of summary generation: extractive and generative. Extractive summarization forms a summary by extracting key sentences from the original text, while generative summarization reorganizes the language based on an understanding of the original text to generate new summary content. The application scenarios of text summarization platforms are very wide, including but not limited to content digestion, research efficiency improvement, enhanced understanding, resource optimization, increased reader participation, content creation assistance, news updates, conference preparation, and language learning. In summary, text summarization platforms, through advanced AI technology, provide users with an effective means to quickly obtain the core content of text, and are suitable for various scenarios and needs.

[0058] In practice, the platform can obtain the initial text in several ways: users can input the initial text or upload the initial file through the platform's interactive page; or the initial text can be obtained by selecting a knowledge base file as the original text for the summary generation. No specific limitation is made here.

[0059] Step S200: According to the preset structured template, the initial text is converted into prompt words to obtain structured prompt words, wherein the structured prompt words reflect the semantics and structure of the initial text;

[0060] It's important to note that the structured template is a framework for structured prompts. It helps users create high-quality prompts through standardization and modularization to guide large language models (LLMs) in generating accurate responses. A prompt is a word or phrase following a query when a user interacts with a large language model. It guides the model to output the desired response. The accuracy of the prompt is crucial in the interaction between the user and the model; high-quality prompts lead to high-quality output. The structured prompt (StructuredPrompt) is a methodology for writing high-quality prompts, enabling deeper interaction between users and large language models through modularization and standardization.

[0061] In a specific implementation, the structured prompt words reflect the semantics and structure of the initial text. After the summary generation platform sends the structured prompt words to the large language model, the large language model can comprehensively consider the full-text structure and semantics of the initial text based on the structured prompt words, making the generated text summary information more complete and consistent with the original text meaning, thereby improving the accuracy of text summary generation.

[0062] In its specific implementation, the platform performs a prompt word conversion operation on the initial text according to a preset structured template to obtain structured prompt words. This process includes:

[0063] Attribute word information at each level is extracted from a preset structured template, wherein the attribute word information is a keyword that identifies different categories in the structured prompt words; the initial text and the attribute word information are sent to the large language model, and the structured prompt words output by the large language model are received, wherein the structured prompt words are generated by the large language model extracting the descriptive text of the attribute word information from the initial text based on the prompt word output instructions, obtaining the target semantic description text, and converting the target semantic description text into prompt words.

[0064] It should be noted that the attribute words are keywords or phrases used to identify and describe different parts of the prompt words. They are typically used as headings in Markdown format to clearly organize and display the structure of the prompt words.

[0065] Among them, attribute words have the following characteristics and functions:

[0066] Identification: Attribute words use specific formats (such as # headings in Markdown) to identify different parts of the prompt words, such as #Role, ##Skills, etc.

[0067] Semantic description: Each attribute word contains specific semantics, describing the nature and content of the following text paragraph or sentence. For example, the "Skills" attribute word is usually followed by a description of the skills or abilities the model should possess.

[0068] Organizational Structure: Attribute words help organize the structure of the prompt words, making the various parts of the prompt words clearer and more orderly, and easier to read and understand.

[0069] Model guidance: Attribute words play a guiding role in the model, helping it understand the intent and importance of each part, thereby better processing and responding to user input.

[0070] Easy to expand and modify: Due to the explicitness and identifiability of attribute words, users can more easily expand and modify prompt words to adapt to different needs and scenarios.

[0071] Improve readability: In longer or more complex prompts, the use of attribute words improves overall readability, allowing other users or developers to quickly grasp the structure and key points of the prompts.

[0072] Furthermore, attribute words include, but are not limited to:

[0073] Role: Defines the role or identity a model plays in an interaction.

[0074] Profile: Provides background information or an overview of the character.

[0075] Skills: These describe the professional skills or abilities a character possesses.

[0076] Rules: Rules are the rules or constraints that characters should follow during interactions.

[0077] Workflow: A process that describes the steps or procedures by which a role completes a task.

[0078] Goals: Define the goals that a character should achieve during the interaction.

[0079] Initialization: Initialization refers to setting the behavior or steps a character takes when starting an interaction.

[0080] Understandably, attribute words play a crucial role in structured cue words, helping to build a clear, organized, easy-to-understand, and easy-to-use cue word framework.

[0081] In its implementation, the platform utilizes a deployed large language model. Through PromptEngineering system instructions, the large language model generates structured prompts based on initial text and attribute word information, without requiring specific fine-tuning. These prompts guide the large model's input, processing, and output of the summary generation results. PromptEngineering aims to help the large model achieve high accuracy and relevance in summary generation tasks; the quality of the prompt engineering significantly impacts the generation outcome.

[0082] Furthermore, the structured prompt words in this application are generated by the large language model based on the prompt word output instructions, extracting descriptive text from the initial text according to the corresponding attribute word information, obtaining target semantic descriptive text, and then converting the target semantic descriptive text into prompt words. Specifically, the large language model extracts the descriptive text that conforms to the attribute word information from the initial text to obtain the target semantic descriptive text, and finally summarizes the target semantic descriptive text at each level to convert it into prompt words. The structured prompt words are characterized by using specific attribute words, using Markdown markup syntax identifiers to control content hierarchy, identify structure and variables, and supplementing with the CoT (Coding on the Knowledge) chain method. Attribute words provide semantic cues and summaries of the prompt word content, mitigating interference from inappropriate content. Identifiers group together texts with similar semantics, reducing the difficulty of model understanding.

[0083] Step S300: Send the structured prompt words to the large language model and receive the text summary information output by the large language model, wherein the text summary information is generated by the large language model based on the summary output instruction to perform summary text conversion on the structured prompt words.

[0084] In its specific implementation, the platform sends the structured prompt words to the large language model and receives the text summary information output by the large language model. Specifically, this application uses the deployed large language model to generate text summary information for the initial text. The large language model can more accurately summarize the entire text, thereby improving the accuracy of text summary generation. That is, by using the prompt word project to generate customized summaries for multilingual texts using the large language model, no additional model training is required, which reduces costs while being more efficient and accurate.

[0085] In its specific implementation, the platform sends the structured prompt words to the large language model and receives the text summary information output by the large language model, including the following steps:

[0086] The system receives user-inputted requirement information and converts it into requirement parameters; it sends the structured prompt words and the constraint parameters to the large language model and receives text summary information output by the large language model, wherein the text summary information is generated by the large language model through summary text conversion of the structured prompt words based on the summary output instruction and the requirement parameters.

[0087] It should be noted that the demand information refers to the user's requirements for the generated summaries. After receiving the demand information input by the user, the platform converts the demand information into demand parameters, which include, but are not limited to, setting the number of summaries to be generated and the number of words in each summary.

[0088] In the specific implementation, refer to Figure 2 First, users select a summary generation template through the platform's interactive page, and then input text, upload a file, or select a knowledge base file as the initial text for summary generation. They also set the number of summaries to be generated and the word count for each summary. The summary content generated by the large model is displayed as streaming input on the front-end document editing page for subsequent user editing. This module uses Prompt Engineering system commands to transform the large language model's summary generation parameters, output language, and output format into structured prompts without requiring specific fine-tuning. This guides the large model's input, processing, and output of the summary generation results. Prompt Engineering aims to help the large model achieve high accuracy and relevance in summary generation tasks; the quality of the Prompt Engineering significantly impacts the generation effect. In short, this application utilizes Prompt Engineering to leverage the large language model for customized summary generation of multilingual text, eliminating the need for additional model training, thus reducing costs while achieving greater efficiency and accuracy.

[0089] In practice, users can customize the number of key points and the word count per point in the generated summary based on the text and actual needs, thus achieving personalized requirements. Furthermore, this application supports the generation of summaries for multilingual texts, and can generate accurate and concise summaries while meeting the limitations on the number of points and word count, eliminating the need for manual translation by the user and making it more convenient.

[0090] In a specific implementation, after the platform sends the structured prompt words to the large language model and receives the text summary information output by the large language model, the method includes:

[0091] Based on a preset context detection template, the text summary information is detected to obtain a detection result; it is then determined whether the detection result meets a preset detection standard. If the detection result does not meet the detection standard, the text summary information is adjusted based on the context detection template to obtain an adjusted text summary information.

[0092] It should be noted that the context detection template is a preset template used to detect the readability (reading fluency) and word / sentence errors of the context information of the text summary. It is usually constructed manually based on experience. If the detection result of the text summary information does not meet the detection standard, it means that the text summary information has sentence problems, including but not limited to unfluent sentences and incorrect use of conjunctions.

[0093] In its implementation, the platform uses a preset context detection template to detect the text summary information, obtains the detection results, and determines whether the results meet preset detection standards. If the results do not meet the standards, it indicates that the text summary information has grammatical issues. The platform then adjusts the text summary information based on the context detection template to obtain a revised text summary. In other words, the platform fine-tunes the summary based on the performance of the prompt words to achieve the desired effect. The underlying pre-trained large language model possesses powerful natural language processing capabilities, thus enabling the system to generate high-quality summaries without requiring extensive corpus training by utilizing high-performance structured prompt words. The model can handle multiple languages, reducing the complexity and cost of multilingual text processing.

[0094] Unlike related technologies that typically extract key sentences or phrases from the original text to generate summaries, this extraction-based summarization method does not comprehensively consider the semantics and structure of the text. The extracted sentences often fail to accurately represent the full text, resulting in low accuracy of the generated summaries. In contrast, this application uses a summarization platform to perform cue word conversion on the initial text to be converted into a summary, obtaining structured cue words that reflect the semantics and structure of the initial text. These structured cue words are then sent to a large language model, and the platform receives the text summary information output by the large language model. Understandably, on the one hand, the summarization platform uses a deployed large language model to generate text summary information from the initial text, where the large language model can more accurately summarize the entire text. On the other hand, because the structured cue words reflect the semantics and structure of the initial text, after the summarization platform sends these structured cue words to the large language model, the large language model can comprehensively consider the full text structure and semantics of the initial text based on these structured cue words, resulting in more complete text summary information that closely matches the original meaning, thereby improving the accuracy of the generated text summary.

[0095] Based on the first embodiment described above, this application also proposes another embodiment, which is referred to below. Figure 3 The text summarization method includes:

[0096] In a specific implementation, after the platform sends the structured prompt words to the large language model and receives the text summary information output by the large language model, the method includes:

[0097] Step A100: Display the text summary information on the user's device;

[0098] In its specific implementation, after receiving the text summary information output by the large language model, the platform displays the text summary information to the user on the user's end, allowing the user to edit and modify it based on the text summary information. In other words, the platform supports users to perform various rich text editing operations.

[0099] Step A200: In response to a text editing instruction issued by the user on the user terminal, wherein the text editing instruction includes a start field, an end field, and editing operation information;

[0100] In its implementation, after the platform displays the text summary information to the user, the user can issue text editing commands on the user's end to achieve interaction between the user and the platform. The text editing commands include a start field, an end field, and editing operation information. The start field indicates the starting field of the required editing adjustment (add, delete, modify, etc.), that is, the first field after the start cursor. The end field indicates the ending field of the required editing adjustment, that is, the first field before the end cursor. The editing operation information includes, but is not limited to, adding, deleting, modifying, and other editing operations.

[0101] Step A300: Integrate the fields between the start field and the end field to obtain the target editing field;

[0102] In the specific implementation, the consecutive fields between the start field and the end field are integrated into a field set, that is, the platform integrates the fields between the start field and the end field to obtain the target editing field.

[0103] Step A400: Perform the corresponding editing operation on the target editing field to obtain the edited text summary information.

[0104] In its implementation, the summary content is displayed on the document editing page, and the editing module supports various rich text editing operations. The editing module provides complex editing functions including basic Word features, Markdown formatting, code blocks, and quotation blocks to meet diverse user needs. After editing, the document is automatically saved to ensure its integrity and consistency. The document is parsed by the editor into tree-structured JSON data, as shown in the diagram. Figure 4 As shown, each modification saves the entire updated document JSON. When reading a saved document, the complete JSON is presented to the editor for parsing. The standard input structure for the summary generated by the model is text with segmented formatting. Users can freely add nested document blocks of other formats on this basis to personalize the summary text and ensure that the final output summary document meets expectations.

[0105] In its specific implementation, the platform performs editing operations on the target edit field according to the corresponding editing operation information to obtain the edited text summary information, including:

[0106] The target editing field and the editing operation information are converted into prompt words to obtain target prompt words; the target prompt words are sent to the large language model, and the edited text summary information output by the large language model is received, wherein the edited text summary information is generated by the large language model performing the corresponding editing operation information on the target editing field based on the summary adjustment instruction.

[0107] In its implementation, in addition to various editing functions, this application also supports various AI-generated content creation techniques for the generated summary documents. (See reference...) Figure 5 Users can select text and choose the corresponding creative ability to process it. Each creative ability has corresponding prompts. The engineer converts the user's selected text and instructions into a standard calling format, then requires a standardized output format, and then calls the large model to complete the corresponding creative task. The generated content is then displayed in a streaming manner on the front end for the user to choose from. Specific creative functions include continuation writing, polishing, changing expectations, translation, grammar and spell checking, abbreviation, and expansion. Through these functions, users can make more detailed adjustments and optimizations to the generated summary documents as needed to meet more precise information processing requirements.

[0108] In specific implementation, the editing operation information includes, but is not limited to, continuation, polishing, changing expectations, translation, grammar and spell checking, abbreviation and expansion, etc. Users only need to click the corresponding button with editing operation information on the platform interaction page. The platform converts the target editing field selected by the user and the editing operation information into prompt words to obtain target prompt words. Then, it uses a large language model to perform the corresponding editing operation on the target editing field and outputs the edited text summary information, that is, the text summary information created by AI.

[0109] In its specific implementation, this application's secondary AI creation based on the Big Prophecy model allows users to freely rewrite the generated content, and various creation tasks can be completed with one click, truly providing a convenient and complete text summary writing process. That is, this application provides rich AI creation functions while supporting basic rich text editing capabilities, meeting the needs for rewriting the generated content.

[0110] It should be noted that the above examples are only for understanding this application and do not constitute a limitation on the text digest generation method of this application. Any simple transformations based on this technical concept are within the protection scope of this application.

[0111] This application also provides a text summarization device; please refer to... Figure 6The text summarization generation device includes:

[0112] Module 10 is used to obtain the initial text;

[0113] The conversion module 20 is used to perform prompt word conversion on the initial text according to a preset structured template to obtain structured prompt words, wherein the structured prompt words reflect the semantics and structure of the initial text;

[0114] The generation module 30 is used to send the structured prompt words to the large language model and receive the text summary information output by the large language model, wherein the text summary information is generated by the large language model based on the summary output instruction to perform summary text conversion on the structured prompt words.

[0115] Optionally, the conversion module 20 includes:

[0116] The extraction module is used to extract attribute word information at each level from a preset structured template, wherein the attribute word information is a keyword that identifies different categories of the structured prompt words;

[0117] The prompt word conversion module is used to send the initial text and the attribute word information to the large language model, and receive the structured prompt words output by the large language model. The structured prompt words are generated by the large language model extracting the descriptive text of the corresponding attribute word information from the initial text based on the prompt word output instructions, obtaining the target semantic description text, and converting the target semantic description text into prompt words.

[0118] Optionally, the generation module 30 includes:

[0119] The parameter conversion module is used to receive user-input requirement information and convert the requirement information into requirement parameters;

[0120] The summary text conversion module is used to send the structured prompt words and the restriction parameters to the large language model, and receive the text summary information output by the large language model, wherein the text summary information is generated by the large language model through summary text conversion of the structured prompt words based on the summary output instruction and the requirement parameters.

[0121] Optionally, the text summarization generating apparatus further includes:

[0122] The detection module is used to detect the text summary information based on a preset context detection template and obtain the detection result;

[0123] The adjustment module is used to determine whether the detection result meets the preset detection standard. If the detection result does not meet the detection standard, the text summary information is adjusted based on the context detection template to obtain the adjusted text summary information.

[0124] Optionally, the text summarization generating apparatus further includes:

[0125] The display module is used to display the text summary information on the user's end.

[0126] A response module is used to respond to a text editing instruction issued by a user on the user terminal, wherein the text editing instruction includes a start field, an end field, and editing operation information;

[0127] The integration module is used to integrate the fields between the start field and the end field to obtain the target editing field;

[0128] The editing module is used to perform editing operations on the target editing field according to the corresponding editing operation information to obtain the edited text summary information.

[0129] Optionally, the editing module includes:

[0130] The target prompt word determination module is used to convert the target editing field and the editing operation information into prompt words to obtain the target prompt words;

[0131] The intelligent creation module is used to send the target prompt words to the large language model and receive the edited text summary information output by the large language model. The edited text summary information is generated by the large language model performing the corresponding editing operation information on the target editing field based on the summary adjustment instruction.

[0132] The text summarization device provided in this application, employing the text summarization method in the above embodiments, can solve the technical problem of text summarization. Compared with the prior art, the beneficial effects of the text summarization device provided in this application are the same as those of the text summarization method provided in the above embodiments, and other technical features in the text summarization device are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0133] This application provides a text summarization generation device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, which are executed by the at least one processor to enable the at least one processor to perform the text summarization generation method in Embodiment 1 above.

[0134] The following is for reference. Figure 7 This document illustrates a structural diagram of a text summarization generation device suitable for implementing embodiments of this application. The text summarization generation device in these embodiments may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Description), PMPs (Portable Media Players), in-vehicle terminals (e.g., in-vehicle navigation terminals), and fixed terminals such as digital TVs and desktop computers. Figure 7 The text summarization device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments of this application.

[0135] like Figure 7 As shown, the text summarization device may include a processing unit 1001 (e.g., a central processing unit, a graphics processing unit, etc.) that can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 1002 or a program loaded from a storage device 1003 into a random access memory (RAM) 1004. The RAM 1004 also stores various programs and data required for the operation of the text summarization device. The processing unit 1001, ROM 1002, and RAM 1004 are interconnected via a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Typically, the following systems can be connected to the I / O interface 1006: input devices 1007 including, for example, a touchscreen, touchpad, keyboard, mouse, image sensor, microphone, accelerometer, gyroscope, etc.; output devices 1008 including, for example, a liquid crystal display (LCD), speaker, vibrator, etc.; storage devices 1003 including, for example, magnetic tape, hard disk, etc.; and communication devices 1009. The communication device 1009 allows the text summarizing device to communicate wirelessly or wiredly with other devices to exchange data. Although the figures show text summarizing devices with various systems, it should be understood that implementing or having all of the systems shown is not required. More or fewer systems may be implemented alternatively.

[0136] Specifically, according to the embodiments disclosed in this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments disclosed in this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from storage device 1003, or installed from ROM 1002. When the computer program is executed by processing device 1001, it performs the functions defined in the methods of the embodiments disclosed in this application.

[0137] The text summarization device provided in this application, employing the text summarization method described in the above embodiments, can solve the technical problem of text summarization. Compared with the prior art, the beneficial effects of the text summarization device provided in this application are the same as those of the text summarization method provided in the above embodiments, and other technical features of this text summarization device are the same as those disclosed in the method of the previous embodiment, and will not be repeated here.

[0138] It should be understood that the various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0139] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0140] This application provides a computer-readable storage medium having computer-readable program instructions (i.e., a computer program) stored thereon, the computer-readable program instructions being used to execute the text summarization generation method in the above embodiments.

[0141] The computer-readable storage medium provided in this application may be, for example, a USB flash drive, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, system, or device. The program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination thereof.

[0142] The aforementioned computer-readable storage medium may be included in the text summarizing device; or it may exist independently and not be assembled into the text summarizing device.

[0143] The aforementioned computer-readable storage medium carries one or more programs that, when executed by the text summarizing device, cause the text summarizing device to generate text summaries.

[0144] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a Local Area Network (LAN) or a Wide Area Network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0145] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0146] The modules described in the embodiments of this application can be implemented in software or hardware. The names of the modules do not necessarily limit the functionality of the unit itself.

[0147] The readable storage medium provided in this application is a computer-readable storage medium that stores computer-readable program instructions (i.e., a computer program) for executing the above-described text summarization generation method, thereby solving the technical problem of text summarization generation. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as those of the text summarization generation method provided in the above embodiments, and will not be repeated here.

[0148] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the text summarization method described above.

[0149] The computer program product provided in this application can solve the technical problem of text summarization. Compared with the prior art, the beneficial effects of the computer program product provided in this application are the same as those of the text summarization method provided in the above embodiments, and will not be repeated here.

[0150] The above description is only a part of the embodiments of this application and does not limit the patent scope of this application. All equivalent structural transformations made under the technical concept of this application and using the contents of the specification and drawings of this application, or direct / indirect applications in other related technical fields, are included in the patent protection scope of this application.

Claims

1. A text summarization method, characterized in that, Applied to a text summarization platform, in which a large language model is deployed, the text summarization method includes: Get the initial text; According to a preset structured template, the initial text is transformed into structured prompts, which reflect the semantics and structure of the initial text. The structured prompts are generated based on attribute word information, which is used to guide the large language model and help it understand the input data. The structured prompt words are sent to the large language model, and the text summary information output by the large language model is received. The text summary information is generated by the large language model based on the summary output instruction to convert the structured prompt words into summary text. The step of performing prompt word conversion on the initial text according to a preset structured template to obtain structured prompt words includes: Attribute word information of each level is extracted from a preset structured template, wherein the attribute word information is a keyword that identifies different categories of the structured prompt words; The initial text and the attribute word information are sent to the large language model, and the structured prompt words output by the large language model are received. The structured prompt words are generated by the large language model extracting the descriptive text of the initial text according to the attribute word information based on the prompt word output instructions, obtaining the target semantic description text, and converting the target semantic description text into prompt words.

2. The text summarization method as described in claim 1, characterized in that, The step of sending the structured prompt words to the large language model and receiving the text summary information output by the large language model includes: Receive user-inputted requirement information and convert the requirement information into requirement parameters; The structured prompt words and the requirement parameters are sent to the large language model, and the text summary information output by the large language model is received. The text summary information is generated by the large language model by performing a summary text conversion on the structured prompt words based on the summary output instruction and the requirement parameters.

3. The text summarization method as described in claim 1, characterized in that, After the steps of sending the structured prompt words to the large language model and receiving the text summary information output by the large language model, the method includes: Based on a preset context detection template, the text summary information is detected to obtain the detection result; Determine whether the detection result meets the preset detection standard. If the detection result does not meet the detection standard, then adjust the text summary information based on the context detection template to obtain the adjusted text summary information.

4. The text summarization method as described in claim 1, characterized in that, After the steps of sending the structured prompt words to the large language model and receiving the text summary information output by the large language model, the method includes: The text summary information is displayed on the user's device; In response to a text editing instruction issued by a user on the user terminal, wherein the text editing instruction includes a start field, an end field, and editing operation information; The fields between the start field and the end field are combined to obtain the target editing field; The corresponding editing operation information is used to perform the editing operation on the target editing field to obtain the edited text summary information.

5. The text summarization method as described in claim 4, characterized in that, The step of performing the corresponding editing operation on the target editing field to obtain the edited text summary information includes: The target editing field and the editing operation information are converted into prompt words to obtain the target prompt words; The target prompt word is sent to the large language model, and the edited text summary information output by the large language model is received. The edited text summary information is generated by the large language model performing the corresponding editing operation information on the target editing field based on the summary adjustment instruction.

6. A text summarization generation apparatus, characterized in that, The device includes: The acquisition module is used to acquire the initial text; The conversion module is used to perform prompt word conversion on the initial text according to a preset structured template to obtain structured prompt words. The structured prompt words reflect the semantics and structure of the initial text. The structured prompt words are generated based on attribute word information, which is used to guide the large language model and help the large language model understand the input data. The generation module is used to send the structured prompt words to the large language model and receive the text summary information output by the large language model, wherein the text summary information is generated by the large language model based on the summary output instruction to perform summary text conversion on the structured prompt words; The step of performing a prompt word conversion operation on the initial text according to a preset structured template to obtain structured prompt words includes: Attribute word information of each level is extracted from a preset structured template, wherein the attribute word information is a keyword that identifies different categories of the structured prompt words; The initial text and the attribute word information are sent to the large language model, and the structured prompt words output by the large language model are received. The structured prompt words are generated by the large language model extracting the descriptive text of the initial text according to the attribute word information based on the prompt word output instructions, obtaining the target semantic description text, and converting the target semantic description text into prompt words.

7. A text summarization generation device, characterized in that, The device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the text summarization method as described in any one of claims 1 to 5.

8. A storage medium, characterized in that, The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, it implements the steps of the text summarization method as described in any one of claims 1 to 5.

9. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the steps of the text summarization method as described in any one of claims 1 to 5.

Images