Personalized writing auxiliary method and device, equipment and storage medium

Abstract

The embodiment of the invention discloses a personalized writing auxiliary method and device, equipment and a storage medium. The method comprises the steps of obtaining question information, fact data, text style input information and a reference document of a to-be-generated document; utilizing a pre-trained language processing model to generate outline information of the personalized style according to the question information, the text style input information and the fact data; according to the outline information, the fact data, the text style input information and the reference document, writing document content is generated in an auxiliary mode. On the basis, the three dimensions of fact data, text style input information and the reference document are added to serve as writing auxiliary data support, writing efficiency, quality and integrity are improved, meanwhile, the text style of an author can be comprehensively considered, and the content of the writing document generated in an auxiliary mode better fits the writing style of the author.

Description

Technical Field

[0001] The embodiments of the present application relate to the field of artificial intelligence technology, and in particular to a personalized writing assistance method, device, equipment and storage medium. Background Art

[0002] With the development of information technology and intelligent technology today, the traditional document writing model faces severe challenges in modern work with high efficiency requirements. For example, documents such as reports, notices, and announcements play an important role in information communication and decision-making support.

[0003] However, the traditional document writing model is inefficient, of uneven quality, and even incomplete in content. Especially when dealing with a large number of documents and writing documents in emergency situations, the above-mentioned defects of the traditional document writing model are more prominent. Summary of the invention

[0004] The embodiments of the present application provide a personalized writing assistance method, device, equipment and storage medium to improve the efficiency, quality and completeness of document writing.

[0005] In a first aspect, an embodiment of the present application provides a personalized writing assistance method, the method comprising:

[0006] Obtaining the title information, factual data, writing style input information and reference documents of the document to be generated;

[0007] Use the pre-trained language processing model to generate outline information of personalized style based on the topic information, writing style input information, and factual data;

[0008] Assist in generating writing document content based on outline information, factual data, writing style input information, and reference documents.

[0009] In a second aspect, an embodiment of the present application provides a personalized writing assistance device, the device comprising:

[0010] The acquisition module is used to obtain the title information, factual data, writing style input information and reference documents of the document to be generated;

[0011] The first generation module is used to generate outline information of personalized writing style based on the title information, writing style input information and factual data by using the pre-trained language processing model;

[0012] The second generation module is used to assist in generating writing document content based on outline information, factual data, writing style input information, and reference documents.

[0013] In a third aspect, an embodiment of the present application further provides an electronic device, the electronic device comprising:

[0014] one or more processors;

[0015] a storage device for storing one or more programs,

[0016] When one or more programs are executed by one or more processors, the one or more processors implement the personalized writing assistance method provided in any embodiment of the present application.

[0017] In a fourth aspect, an embodiment of the present application further provides a computer-readable storage medium on which a computer program is stored, characterized in that when the program is executed by a processor, a personalized writing assistance method as provided in any embodiment of the present application is implemented.

[0018] The technical solution of the embodiment of the present application obtains the title information, factual data, writing style input information and reference documents of the document to be generated; uses a pre-trained language processing model to generate outline information of a personalized writing style based on the title information, writing style input information and factual data; and assists in generating the content of the writing document based on the outline information, factual data, writing style input information and reference documents. Based on this, the present embodiment adds the three dimensions of factual data, writing style input information and reference documents as data support for writing assistance. While improving writing efficiency, quality and completeness, it can also comprehensively consider the author's writing style, so that the content of the assisted writing document is more in line with the author's own style. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 A flowchart of a personalized writing assistance method provided in Example 1 of the present application;

[0020] Figure 2 A structural schematic diagram of a personalized writing assistance device provided in Example 2 of the present application;

[0021] Figure 3 A schematic diagram of the structure of an electronic device provided in Example 3 of the present application. DETAILED DESCRIPTION

[0022] The present application will be further described in detail below in conjunction with the accompanying drawings and embodiments. It is to be understood that the specific embodiments described herein are only used to explain the present application, rather than to limit the present application. It should also be noted that, for ease of description, only the parts related to the present application, rather than all structures, are shown in the accompanying drawings.

[0023] Embodiment 1

[0024] Figure 1 A flowchart of a personalized writing assistance method provided in Example 1 of the present application is shown in FIG. Figure 1As shown, the personalized writing assistance method provided in this embodiment can be applied to a personalized writing assistance platform mounted on a device with data processing capabilities such as a computer, and can be used in conjunction with some application software to achieve a better experience. Specifically, the method may include the following steps:

[0025] Step 101: Obtain the title information, factual data, writing style input information and reference documents of the document to be generated.

[0026] In this step, the document to be generated refers to the document that the user needs to write, wherein the title information is used to indicate the theme and writing direction of the document to be generated. In this embodiment, an input interface can be set for the personalized writing assistance platform, and the user can enter the title information in the input interface. By performing semantic analysis and keyword extraction on the title information, the prompt word corresponding to the title information is obtained, which is used to instruct the subsequent pre-trained language processing model to generate the outline information and content of the writing document.

[0027] In addition, factual data is data information related to the topic that users choose to input or obtain from the Internet, which can provide certain factual content for the writing of the document, making the generated writing content more in line with the current actual situation and industry standards, and ensuring the rigor and authority of the article. By citing real data and cases (i.e. factual data), the article's discussion is more convincing and the argument is more three-dimensional and credible.

[0028] The writing style input information is used to guide the writing style of the user (i.e. the author), helping the user to maintain a consistent style and high professionalism during the writing process, so as to maintain a unique way of expression in the new document. This method not only retains the writer's personal writing style characteristics, but also improves writing efficiency and ensures the consistency and coherence of the article as a whole.

[0029] Specifically, this step can first obtain the title information, factual data, writing style document and reference document of the document to be generated; then extract the article structure and language characteristics in the writing style document, and determine the article structure and language characteristics as writing style input information.

[0030] It should be noted that the user's writing style mainly includes the structure of the article and the language characteristics. The article structure can include personalized narrative structure, personalized argumentation structure and other structures of different types of texts. For example, in the personalized argumentation structure, the user's writing style is usually "general-specific-general", that is, first summarize an argument, then argue separately, and finally summarize and emphasize the argument.

[0031] As for language characteristics, they may include the user's habitual sentence patterns and word styles. For example, the user's habitual sentence patterns are to use short sentences to express the content more clearly, and the word style tends to be humorous, and the user may use more metaphors and other rhetorical devices to make the expression more vivid and humorous.

[0032] In this embodiment, by understanding the semantics in the writing style document, the user's habitual sentence patterns and word usage style can be derived and used as language characteristics. The content of each paragraph is then extracted, and the user's article structure is determined based on the correlation between the contents to obtain the writing style input information.

[0033] In addition, the reference document can be a typical historical creation document of the user that is related to the document to be generated. The relatedness here can be related to the type of document, the theme, the writing direction, etc. The reference document can make the content of the final generated writing document have a certain novelty and uniqueness. While maintaining personal writing characteristics, it can also draw on excellent writing reference content in the past, helping writers to achieve content innovation in writing while maintaining a certain style and structural coherence.

[0034] It should be noted that in order to improve the efficiency of subsequent calls to the model, the acquired title information, factual data, writing style input information and reference documents can be stored in the database according to a certain strategy to facilitate search and call.

[0035] Specifically, the title information, factual data, and writing style input information are structured to obtain corresponding keyword information, and the keyword information is stored in a structured database according to information category; the reference document is text vectorized to obtain prompt word information, and the prompt word information is stored in a vector database.

[0036] Structured databases and vector databases enable the model to efficiently find the required content from them, thereby speeding up the efficiency of writing assistance.

[0037] It should be noted that in addition to the above content, more information can be obtained to improve the accuracy and personalization of writing assistance, such as scene requirements, writing requirements, word limits and other requirements.

[0038] Step 102: Using a pre-trained language processing model, generate outline information of a personalized writing style according to the title information, writing style input information, and factual data.

[0039] In this step, the pre-trained language processing model can be an artificial intelligence large model. The pre-trained language processing model is used to process and fuse the article structure in the writing style input information with the title information and factual data to obtain a fused article structure; the outline chapter information in the fused article structure is extracted to obtain personalized writing style outline information that matches the title information.

[0040] During processing and fusion, the language processing model can extract corresponding keyword information and prompt word information from the aforementioned structured database or vector database as required, and generate outline chapter information according to the article structure.

[0041] It should be noted that the language processing model can also obtain relevant keywords and prompt words from other databases to enrich the outline chapter information. The obtained outline chapter information is the outline information required in this step.

[0042] In a specific example, the title of the article is "The Future of Renewable Energy", and the fact data is "The latest statistics show that the proportion of global renewable energy in total energy consumption will rise to 30% in 2023." At the same time, a 2M writing style document is uploaded. The outline of the document is (I) Introduction; (II) The current status of renewable energy; (III) Main types and their advantages and disadvantages; (IV) Policy and economic impact; (V) Future outlook; The language style is: formal, objective, and moderate use of technical terms. The writing reference document uploaded 5 excellent articles in the past, each article is less than 5M, covering similar topics such as renewable energy and environmental policy.

[0043] The title information, factual data and writing style input information are stored in a structured database; all 5 uploaded writing reference documents are vectorized and stored in a vector database. The outline chapters and language style of the writing style are refined and visualized using the big model to form prompt words for use by the big model; subsequently, the prompt word content is supplemented based on the provided title, outline and factual data to form prompt word examples and generate an introduction first; then the input 5 writing reference documents with a size of less than 5M are processed through the tool / algorithm module and the large language model (LLM).

[0044] Step 103: Assist in generating writing document content based on outline information, factual data, writing style input information, and reference documents.

[0045] In this step, the reference document can be processed using a preset text processing algorithm combination to obtain reference text features; the outline information is filled in according to factual data, writing style input information and reference text features to obtain the writing document content.

[0046] Among them, the pre-set text processing algorithm combination includes multiple algorithms from the following algorithms: word embedding, text vectorization, sentence tagging, part-of-speech tagging, subword tagging, paragraph slicing, semantic analysis, dependency parsing, individual storage, hierarchical clustering, and logistic regression.

[0047] Specifically, word embedding: convert words into vectors, but the text is composed of multiple words, and these vectors need to be combined.

[0048] Text vectorization: Convert text sequences into fixed-length vector sequences to facilitate model processing of text vectorization, ensure data root consistency, further reduce dimensions, and reduce computational complexity.

[0049] Sentence tagging: Segment the text into meaningful units, such as words, phrases, or punctuation marks. These units are called tokens. To ensure that each language unit is correctly identified, tags can handle abbreviations, numbers, special symbols, etc., to ensure that they are correctly encoded.

[0050] Part-of-speech tagging: reveals the grammatical role of words, strengthens the understanding of sentence structure, distinguishes the meanings of synonyms by part-of-speech, supports keyword extraction and event extraction in information retrieval, improves the accuracy of text classification, is crucial for named entity recognition, helps build accurate semantic models, maintains grammatical consistency in cross-language translation, and guides the generation of grammatically correct text.

[0051] Subword tagging: By splitting words into roots and affixes, it solves the problem of rare words and variants, reduces the sparsity of the vocabulary, and is highly adaptable. For multiple languages, especially Chinese, it ensures the flexibility of word order, improves accuracy and consistency in translation, and subword tagging optimizes model performance, such as providing more detailed input in language models and sequence labeling, which helps promote cross-language migration of models in multilingual tasks, especially when data is sparse, to prevent overfitting.

[0052] Paragraph slicing: Divide large documents into independent segments of related topics, maintain the context of the text, and enable the model to understand the logical coherence between paragraphs. This division helps reduce the computational burden of processing complex texts and adapt to different task requirements. In multiple scenarios, it supports the representation of document collections and cross-document analysis. Paragraph slicing provides information flexibility, allowing the model to analyze at different levels while enhancing the model's generalization ability. It is particularly suitable for processing large model text data sets, facilitating data management and efficient loading.

[0053] Semantic analysis: Through matrix analysis and topic modeling technology, text data is compressed, potential topics are revealed, and dimensions are reduced, which helps to understand the abstract concepts and topic distribution of the text. They play a key role in information retrieval, document clustering, and content analysis (hot topics or preference analysis). By analyzing the correlation between words and topics, measuring document similarity, reducing noise, and improving the sensitivity of deep models to core information, and as input to deep models, it enhances the understanding of complex text semantics.

[0054] Dependency parsing: It helps understand sentence structure by revealing grammatical dependencies between words, supports accurate information extraction, improves the depth of semantic analysis, ensures the grammatical correctness of generated text, enhances the accuracy of named entity recognition, maintains the semantic consistency of machine translation, and serves as an important feature in text classification tasks, thereby optimizing the overall understanding and processing capabilities of text complexity.

[0055] Individual storage: By recording user preferences and historical data, personalized services can be achieved, efficiency can be improved, and adaptation to user needs can be achieved.

[0056] Hierarchical clustering: In the later stage of text processing, by organizing documents into a hierarchical structure based on similarity, automatic text classification and topic discovery can be achieved, which helps to organize content and understand user interests, provide text relationships, and serve as a preprocessing method to simplify subsequent analysis. It is also used to evaluate text similarity, discover abnormal text, and verify model effects. It is an indispensable exploration and verification link in deep text analysis.

[0057] Logistic regression: It is used for subsequent prediction or classification tasks. As a simple but powerful linear model, it converts the previous complex and stable features (such as word embedding, part of speech, and semantic analysis features) into predictable outputs, such as positive and negative classification in sentiment analysis, or prediction of document topics. Logistic regression is used to verify and optimize the effects of the previous processing steps due to its strong interpretability, providing a basic prediction model and providing a benchmark or preliminary decision-making basis for subsequent possible complex models.

[0058] Still taking the above example, the above algorithm can be processed as follows:

[0059] Word embedding: Convert keywords such as “renewable energy” and “future” into vectors.

[0060] Text vectorization: Convert each part of the article into a vector sequence of uniform length to ensure consistency.

[0061] Sentence tagging: Sentences in a document are segmented and tagged to ensure that each unit is accurately identified.

[0062] Part-of-speech tagging: tag nouns, verbs, adjectives, etc. to enhance the understanding of the article structure.

[0063] Subword marking: Breaking down complex professional vocabulary such as "photovoltaic power generation" into more basic roots helps with understanding.

[0064] Paragraph slicing: Split a document into segments of related topics for easier subsequent processing.

[0065] Semantic analysis: Use topic models to analyze the potential topics of articles, such as the relationship between "clean energy" and "policy support".

[0066] Dependency parsing: understanding the logical relationships within a sentence, such as "The growth of renewable energy depends on technological innovation."

[0067] Individual storage: record user preferences so that content that meets their needs can be generated next time.

[0068] Hierarchical clustering: Categorize related information into topics to better understand different types of renewable energy.

[0069] Logistic regression: used for post-sentiment analysis to assess whether the article has a positive or negative impact on renewable energy policies.

[0070] In addition, after completing the writing of the document, you can also rewrite, continue, expand or polish the content selected in the document by operating the mouse.

[0071] That is, rewriting: for the selected text, according to the rewriting requirements, the model will modify the text content as required. Continuation: after selecting a certain text, according to the specified continuation requirements and the logical relationship of the context, write the content to ensure logical consistency. Expansion: after selecting a certain text, according to the requirements, enrich the language description of this text to make the content fuller. Polishing: when a paragraph of text is selected, the language of the paragraph content can be modified according to specific conditions or directly let the model to improve the language fluency, accuracy and expressiveness of the article, ensure clear logic and correct grammar, and enhance the readability and attractiveness of the article.

[0072] The above functions can still be processed with the help of the above models, and the contents of structured databases and vector databases can also be used as the data basis.

[0073] In this embodiment, the title information, factual data, writing style input information, and reference documents of the document to be generated are obtained; the outline information of the personalized writing style is generated based on the title information, writing style input information, and factual data using the pre-trained language processing model; and the writing document content is assisted in being generated based on the outline information, factual data, writing style input information, and reference documents. Based on this, this embodiment adds the three dimensions of factual data, writing style input information, and reference documents as data support for writing assistance. While improving writing efficiency, quality, and completeness, it can also comprehensively consider the author's writing style, so that the assisted generated writing document content is more in line with the author's own writing style.

[0074] Embodiment 2

[0075] Figure 2 This is a schematic diagram of the structure of a personalized writing assistance device provided in Example 2 of the present application. The personalized writing assistance device provided in the embodiment of the present application can execute the personalized writing assistance method provided in any embodiment of the present application, and has the corresponding functional modules and beneficial effects of the execution method. The device can be implemented in software and / or hardware, such as Figure 2 As shown, the personalized writing assistance device specifically includes: an acquisition module 201 , a first generation module 202 , and a second generation module 203 .

[0076] The acquisition module is used to obtain the title information, factual data, writing style input information and reference documents of the document to be generated;

[0077] The first generation module is used to generate outline information of personalized writing style based on the title information, writing style input information and factual data by using the pre-trained language processing model;

[0078] The second generation module is used to assist in generating writing document content based on outline information, factual data, writing style input information, and reference documents.

[0079] Furthermore, the device also includes:

[0080] The structured processing module is used to perform structured processing on the title information, factual data, and writing style input information to obtain corresponding keyword information, and store the keyword information in a structured database according to information categories;

[0081] The vectorization processing module is used to perform text vectorization processing on the reference document, obtain prompt word information, and store the prompt word information in a vector database.

[0082] Furthermore, the acquisition module is specifically used for:

[0083] Obtain the title information, factual data, writing style documents and reference documents of the document to be generated;

[0084] The article structure and language features in the writing style document are extracted, and the article structure and language features are determined as writing style input information.

[0085] Furthermore, the first generating module is specifically used for:

[0086] Using the pre-trained language processing model, the article structure in the writing style input information is processed and integrated with the title information and factual data to obtain the integrated article structure;

[0087] Extract the outline and chapter information from the integrated article structure to obtain personalized writing style outline information that matches the title information.

[0088] Furthermore, the second generation module is specifically used for:

[0089] Processing the reference document using a preset text processing algorithm combination to obtain reference text features;

[0090] The outline information is filled in according to factual data, writing style input information and reference text features to obtain the content of the writing document.

[0091] Furthermore, the preset text processing algorithm combination includes multiple algorithms from the following algorithms:

[0092] Word embedding, text vectorization, sentence tagging, part-of-speech tagging, subword tagging, paragraph slicing, semantic analysis, dependency parsing, individual storage, hierarchical clustering, logistic regression.

[0093] Embodiment 3

[0094] Figure 3 A schematic diagram of the structure of an electronic device provided in Embodiment 3 of the present application is shown in FIG. Figure 3 As shown, the electronic device includes a processor 310, a memory 320, an input device 330 and an output device 340; the number of the processor 310 in the electronic device can be one or more. Figure 3 A processor 310 is taken as an example; the processor 310, the memory 320, the input device 330 and the output device 340 in the electronic device can be connected via a bus or other means. Figure 3 The example of connecting through bus is taken in the following.

[0095] The memory 320 is a computer-readable storage medium that can be used to store software programs, computer executable programs and modules, such as program instructions / modules corresponding to the personalized writing assistance method in the embodiment of the present invention. The processor 310 executes various functional applications and data processing of the electronic device by running the software programs, instructions and modules stored in the memory 320, that is, to implement the above-mentioned personalized writing assistance method:

[0096] Obtain the title information, factual data, writing style input information and reference documents of the document to be generated; use the pre-trained language processing model to generate outline information of personalized style based on the title information, writing style input information and factual data;

[0097] Assist in generating writing document content based on outline information, factual data, writing style input information, and reference documents.

[0098] Furthermore, after obtaining the title information, factual data, writing style input information and reference documents of the document to be generated, the method further includes:

[0099] The title information, factual data, and writing style input information are structured to obtain corresponding keyword information, and the keyword information is stored in a structured database according to information categories;

[0100] The reference document is subjected to text vectorization processing to obtain prompt word information, and the prompt word information is stored in a vector database.

[0101] Furthermore, the title information, factual data, writing style input information and reference documents of the document to be generated are obtained, including:

[0102] Obtain the title information, factual data, writing style documents and reference documents of the document to be generated;

[0103] The article structure and language features in the writing style document are extracted, and the article structure and language features are determined as writing style input information.

[0104] Furthermore, the pre-trained language processing model is used to generate outline information of personalized writing style based on the title information, writing style input information, and factual data, including:

[0105] Using the pre-trained language processing model, the article structure in the writing style input information is processed and integrated with the title information and factual data to obtain the integrated article structure;

[0106] Extract the outline and chapter information from the integrated article structure to obtain personalized writing style outline information that matches the title information.

[0107] Furthermore, the outline information, factual data, writing style input information and reference documents are used to assist in generating writing document content, including:

[0108] Processing the reference document using a preset text processing algorithm combination to obtain reference text features;

[0109] The outline information is filled in according to factual data, writing style input information and reference text features to obtain the content of the writing document.

[0110] Furthermore, the preset text processing algorithm combination includes multiple algorithms from the following algorithms:

[0111] Word embedding, text vectorization, sentence tagging, part-of-speech tagging, subword tagging, paragraph slicing, semantic analysis, dependency parsing, individual storage, hierarchical clustering, logistic regression.

[0112] The memory 320 may mainly include a program storage area and a data storage area, wherein the program storage area may store an operating system and at least one application required for a function; the data storage area may store data created according to the use of the terminal, etc. In addition, the memory 320 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other non-volatile solid-state storage device. In some instances, the memory 320 may further include a memory remotely arranged relative to the processor 310, and these remote memories may be connected to the electronic device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0113] Embodiment 4

[0114] Embodiment 4 of the present application further provides a storage medium containing computer executable instructions, and when the computer executable instructions are executed by a computer processor, they are used to execute a personalized writing assistance method, the method comprising:

[0115] Obtaining the title information, factual data, writing style input information and reference documents of the document to be generated;

[0116] Use the pre-trained language processing model to generate outline information of personalized style based on the topic information, writing style input information, and factual data;

[0117] Assist in generating writing document content based on outline information, factual data, writing style input information, and reference documents.

[0118] Furthermore, after obtaining the title information, factual data, writing style input information and reference documents of the document to be generated, the method further includes:

[0119] The title information, factual data, and writing style input information are structured to obtain corresponding keyword information, and the keyword information is stored in a structured database according to information categories;

[0120] The reference document is subjected to text vectorization processing to obtain prompt word information, and the prompt word information is stored in a vector database.

[0121] Furthermore, the title information, factual data, writing style input information and reference documents of the document to be generated are obtained, including:

[0122] Obtain the title information, factual data, writing style documents and reference documents of the document to be generated;

[0123] The article structure and language features in the writing style document are extracted, and the article structure and language features are determined as writing style input information.

[0124] Furthermore, the pre-trained language processing model is used to generate outline information of personalized writing style based on the title information, writing style input information, and factual data, including:

[0125] Using the pre-trained language processing model, the article structure in the writing style input information is processed and integrated with the title information and factual data to obtain the integrated article structure;

[0126] Extract the outline and chapter information from the integrated article structure to obtain personalized writing style outline information that matches the title information.

[0127] Furthermore, the outline information, factual data, writing style input information and reference documents are used to assist in generating writing document content, including:

[0128] Processing the reference document using a preset text processing algorithm combination to obtain reference text features;

[0129] The outline information is filled in according to factual data, writing style input information and reference text features to obtain the content of the writing document.

[0130] Furthermore, the preset text processing algorithm combination includes multiple algorithms from the following algorithms:

[0131] Word embedding, text vectorization, sentence tagging, part-of-speech tagging, subword tagging, paragraph slicing, semantic analysis, dependency parsing, individual storage, hierarchical clustering, logistic regression.

[0132] Of course, the storage medium containing computer executable instructions provided in the embodiment of the present application, whose computer executable instructions are not limited to the above method operations, can also execute related operations in the personalized writing assistance method provided in any embodiment of the present application.

[0133] Through the above description of the implementation method, the technicians in the relevant field can clearly understand that the present application can be implemented with the help of software and necessary general hardware, and of course it can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present application can be essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory (FLASH), hard disk or optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, server, or network device, etc.) to execute the methods of each embodiment of the present application.

[0134] It is worth noting that in the embodiment of the above-mentioned search device, the various units and modules included are only divided according to functional logic, but are not limited to the above-mentioned division, as long as the corresponding functions can be achieved; in addition, the specific names of the functional units are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application.

[0135] Note that the above are only preferred embodiments of the present application and the technical principles used. Those skilled in the art will understand that the present application is not limited to the specific embodiments herein, and that various obvious changes, readjustments and substitutions can be made by those skilled in the art without departing from the scope of protection of the present application. Therefore, although the present application is described in more detail through the above embodiments, the present application is not limited to the above embodiments, and may include more other equivalent embodiments without departing from the concept of the present application, and the scope of the present application is determined by the scope of the appended claims.

Claims

1. A personalized writing assistance method, characterized in that: The method comprises: Obtaining the title information, factual data, writing style input information and reference documents of the document to be generated; Using a pre-trained language processing model, generating outline information of a personalized writing style according to the topic information, the writing style input information, and the factual data; The writing document content is assisted in being generated according to the outline information, the factual data, the writing style input information and the reference document.

2. The method according to claim 1, characterized in that After obtaining the title information, factual data, writing style input information and reference documents of the document to be generated, the method further includes: Structuring the title information, the factual data, and the writing style input information to obtain corresponding keyword information, and storing the keyword information in a structured database according to information categories; The reference document is subjected to text vectorization processing to obtain prompt word information, and the prompt word information is stored in a vector database.

3. The method according to claim 1, characterized in that The step of obtaining the title information, factual data, writing style input information, and reference documents of the document to be generated includes: Obtain the title information, factual data, writing style documents and reference documents of the document to be generated; The article structure and language features in the writing style document are extracted, and the article structure and language features are determined as writing style input information.

4. The method according to claim 3, characterized in that The method of using a pre-trained language processing model to generate outline information of a personalized writing style according to the topic information, the writing style input information, and the factual data includes: Using a pre-trained language processing model, the article structure in the writing style input information is processed and integrated with the title information and the factual data to obtain an integrated article structure; The outline and chapter information in the fused article structure is extracted to obtain personalized style outline information that matches the title information.

5. The method according to claim 3, characterized in that: The step of assisting in generating writing document content according to the outline information, the factual data, the writing style input information, and the reference document includes: Processing the reference document using a preset text processing algorithm combination to obtain reference text features; The outline information is filled with content according to the factual data, the writing style input information and the reference text features to obtain the writing document content.

6. The method according to claim 5, characterized in that The preset text processing algorithm combination includes multiple algorithms from the following algorithms: Word embedding, text vectorization, sentence tagging, part-of-speech tagging, subword tagging, paragraph slicing, semantic analysis, dependency parsing, individual storage, hierarchical clustering, logistic regression.

7. A personalized writing assistance device, characterized in that: The device comprises: The acquisition module is used to obtain the title information, factual data, writing style input information and reference documents of the document to be generated; A first generating module, configured to generate outline information of a personalized writing style according to the topic information, the writing style input information, and the factual data by using a pre-trained language processing model; The second generation module is used to assist in generating writing document content based on the outline information, the factual data, the writing style input information and the reference document.

8. The device according to claim 7, characterized in that The device also includes: A structured processing module, used to perform structured processing on the title information, the factual data, and the writing style input information to obtain corresponding keyword information, and store the keyword information in a structured database according to information categories; The vectorization processing module is used to perform text vectorization processing on the reference document to obtain prompt word information, and store the prompt word information in a vector database.

9. An electronic device, characterized in that: include: one or more processors; a storage device for storing one or more programs, When the one or more programs are executed by the one or more processors, the one or more processors implement the personalized writing assistance method as described in any one of claims 1-6.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the personalized writing assistance method as described in any one of claims 1 to 6 is implemented.

Images