Document conversion system, document conversion method, document conversion program

Abstract

This document conversion system provides a system that converts source document data into converted document data according to specific stylistic characteristics. [Solution] The document conversion device 100 of the document conversion system includes a document setting unit 111 that accepts the setting of source document data to be converted, an auxiliary data setting unit 112 that accepts the setting of style auxiliary data including at least one of the author, document genre, style category, and reference information for identifying reference documents, an instruction generation unit 114 that generates generation instructions for converting the source document data based on the style features identified based on the style auxiliary data, and a generation processing unit 115 that inputs the generation instructions and the source document data into a generation model and obtains output of converted document data.

Description

TECHNICAL FIELD

[0001] The present invention relates to a document conversion system, method, and program for converting original document data into converted document data in accordance with specific stylistic features. BACKGROUND ART

[0002] Conventionally, various techniques for generating content according to the preferences and requests of individual users have been studied. For example, in the generation of original documents for books, there is a tendency that expressions and developments matching the preferences and sensibilities of readers are required, rather than merely conveying information.

[0003] In recent years, with the development of artificial intelligence (AI) technology, methods for automatically generating texts using machine learning models have also been proposed. This makes it possible to efficiently create documents of various variations based on settings of specific plots and characters.

[0004] Patent Document 1 discloses a technique that inputs information such as plots, story developments, and characters to generate novels that can be enjoyed from the reader's perspective, and combines AI-based text generation with the utilization of existing novels. This system is configured to include a web user interface for a user to input desired information, AI-based text generation means, and image generation AI means for utilizing existing novels. PRIOR ART DOCUMENTS PATENT DOCUMENTS

[0005] PATENT DOCUMENT 1 Japanese Unexamined Patent Publication No. 2025-52598 SUMMARY OF THE INVENTION PROBLEMS TO BE SOLVED BY THE INVENTION

[0006] Incidentally, not only the content of a document, but also its style and expression greatly influence the impression it gives to the reader. For example, even with the same content, the style and expression can change the reader's level of immersion, comprehension, and even the barriers to reading itself. However, there is no known technology to optimally transform the style and expression of existing documents for the reader.

[0007] In light of these circumstances, this disclosure aims to address the problem of providing a technology that converts source document data into converted document data according to specific stylistic characteristics. [Means for solving the problem]

[0008] To solve the above problems, the document conversion system according to this disclosure comprises: a document setting unit that accepts the setting of source document data to be converted; an auxiliary data setting unit that accepts the setting of style auxiliary data including at least one of the author, document genre, style category, and reference information for identifying reference documents; an instruction generation unit that generates generation instructions for converting the source document data based on style features identified based on the style auxiliary data; and a generation processing unit that inputs the generation instructions and the source document data into a generation model and obtains output of converted document data. [Effects of the Invention]

[0009] According to the present invention, it is possible to provide a technology that converts source document data into converted document data according to specific stylistic characteristics. [Brief explanation of the drawing]

[0010] [Figure 1] System configuration diagram according to this embodiment. [Figure 2] Hardware configuration diagram of the document conversion device according to this embodiment. [Figure 3] Hardware configuration diagram of the user terminal according to this embodiment. [Figure 4] Block diagram of the document conversion device according to this embodiment. [Figure 5]A diagram illustrating the structure of the style database according to this embodiment. [Figure 6] A processing flowchart of the document conversion method according to this embodiment. [Modes for carrying out the invention]

[0011] The following document conversion system, method, and program relating to embodiments of this disclosure will be described with reference to the drawings. Note that the embodiments shown below are examples of the present invention, and this disclosure is not limited to these embodiments; various configurations can be adopted.

[0012] This embodiment describes the configuration and operation of a document conversion system, but similar configurations, computer programs, and program recording media on which such programs are stored will produce similar effects. Using a program recording medium, the program can be installed on a computer. The series of processes according to this embodiment, described below, are provided as a program executable on a computer and can be provided via non-transient computer-readable recording media such as CD-ROMs and flexible disks, as well as via communication lines.

[0013] A document conversion system consists of one or more computer devices. Each computer device includes a processing unit such as a CPU (Central Processing Unit) and a memory device. The computer device can function as a document conversion device by executing a program stored in the memory device using its processing unit. The document conversion method is implemented through the processing of the document conversion system and the device.

[0014] In this embodiment, "document" refers to data containing natural language text, or a medium on which such data is recorded. Specifically, "document" includes books in various formats such as hardcovers, paperbacks, magazines, and ebooks. Books include literary works such as novels, essays, columns, reportages, and critiques, as well as non-literary works such as specialized books on specific academic, technical, or business matters. Furthermore, "document" may also include non-literary text data such as articles on websites, blogs, news articles, and internal company documents. This disclosure describes a specific embodiment of the document conversion system using the case where the source document data is a novel as an example, but is not limited to this.

[0015] The source document data includes the full or partial text extracted from the aforementioned document. The converted document data includes data obtained by converting the full or partial text of the source document data. In this embodiment, the target of conversion includes the style and expression in the text contained in the source document data. However, the target of conversion is not limited to these, and may also include conversion of vocabulary, proper nouns, settings, etc., to the extent that the meaning and main points of the text are maintained. For example, this includes conversion from technical terms to general terms, conversion to synonyms suitable for the reader, conversion to other proper nouns suitable for the reader's attributes (nationality, age, etc.) such as names of people and places, and conversion to settings suitable for the reader's attributes such as setting, time period, and cultural background.

[0016] 1. System Configuration Figure 1 shows a diagram of the document conversion system. As shown in Figure 1, the document conversion system 1 comprises a document conversion device 100, a user terminal 200, and a generation device 300. The document conversion device 100 is connected to the user terminal 200 and the generation device 300 via a communication network NW. The document conversion system 1 performs a document conversion method as a whole through the cooperation of the methods performed by each component. In the illustrated example, only one user terminal 200 is shown, but there may be multiple user terminals. Also, the user terminal 200 may be connected to the generation device 300 via a communication network NW.

[0017] 2. Hardware Configuration Figure 2 shows a hardware configuration diagram of the document conversion apparatus 100. The document conversion apparatus 100 is configured by a server apparatus, a personal computer, or the like, and includes a processor 110, a memory 120, and a communication interface 130. The document conversion apparatus 100 may further include an input / output interface (such as a keyboard and a monitor). In addition, the document conversion apparatus 100 may be configured by one or a plurality of computers.

[0018] The processor 110 is configured by a CPU, a GPU (Graphics Processing Unit), or the like that can execute an instruction set. The processor 110 executes a computer program for executing the document conversion method, or at least a part of a plurality of computer programs that are executed in cooperation, as well as an OS (Operating System) and other applications, thereby controlling the entire operation processing of the document conversion apparatus 100. The document conversion apparatus 100 may include one or a plurality of processors 110.

[0019] The memory 120 includes a volatile memory such as a RAM (Random Access Memory) capable of storing an instruction set, and a non-volatile recording medium such as an HDD (Hard Disk Drive) or an SSD (Solid State Drive) capable of recording the OS and the above-described computer programs. The document conversion apparatus 100 includes one or a plurality of memories 120, and at least one of the one or more memories stores a database management system for functioning as a style database 121.

[0020] The communication interface 130 has an interface for connecting to a communication network, executes communication control, and performs input and output of information. The document conversion apparatus 100 includes one or a plurality of communication interfaces 130. When the document conversion apparatus 100 is operated as a server and does not have a physical input / output interface, operations are performed from a client via a communication network such as a LAN (Local Area Network) or a WAN (Wide Area Network). Similarly, when a plurality of document conversion apparatuses 100 cooperate to execute the document conversion method, communication is performed between the plurality of document conversion apparatuses 100 via the communication network.

[0021] Figure 3 shows a hardware configuration diagram of the user terminal 200. The user terminal 200 is configured by a personal computer, a tablet terminal, a smartphone, or the like, and includes a processor 210 such as a CPU, a memory 220 such as a RAM, an HDD, an SSD, or a flash storage, a communication interface 230 connected to a communication network via wireless communication, a wired LAN, or the like, an input interface 240 such as a mouse, a keyboard, a touch panel, or a microphone, and an output interface 250 such as a display or a speaker. The user terminal 200 is operated by a user who generates converted document data based on original document data, and a plurality of user terminals 200 may be present. The memory 220 stores browser applications and various types of data. The processor 210 executes, for example, a browser application, and accesses a web service provided by the document conversion apparatus 100 via a web browser, thereby realizing data communication with the document conversion apparatus 100.

[0022] The generation apparatus 300 has the same hardware configuration as the document conversion apparatus 100. The generation apparatus 300 is a computer apparatus installed by an operator of the document conversion system 1 or an external operator. The generation apparatus 300 receives an input of a processing request from the document conversion apparatus 100, and generates a response thereto. In the present embodiment, the processing request includes a generation instruction and original document data. The generation apparatus 300 returns the generated response to the document conversion apparatus 100.

[0023] The generation device 300 can generate a response by inputting the data included in the processing request into the generation model. A known model that has been trained using machine learning can be used as the generation model. In this embodiment, a Large Language Model (LLM) is adopted as the generation model. An LLM is a type of natural language processing model that has been trained using large amounts of natural language data. For example, an LLM receives instructional text (prompts) and various data written in natural language as input and outputs data that is a response to them. Note that the LLM may have a multimodal configuration and can handle data input and output in various formats such as images, videos, and audio. Furthermore, in this embodiment, the generation model is not limited to an LLM, and other machine learning models or algorithms capable of generating responses to inputs may be adopted.

[0024] In this embodiment, an example is described in which the generated model is stored in the generation device 300, but a configuration in which the document conversion device 100 is equipped with the generated model can also be adopted. Furthermore, the generation device 300 may be installed inside the document conversion device 100, or installed outside the document conversion device 100 and connected to the device by wire or wireless.

[0025] 3. Functional Components Figure 4 shows a block diagram of the document conversion device 100. The processor 110 of the document conversion device 100 functions as a functional component including a document setting unit 111, an auxiliary data setting unit 112, an analysis unit 113, an instruction generation unit 114, and a generation processing unit 115 by executing a computer program stored in the memory 120. The memory 120 functions as a database including a style database 121.

[0026] The document conversion device 100 has one or more processors, at least one of which functions as a document setting unit 111, at least one as an auxiliary data setting unit 112, at least one as an analysis unit 113, at least one as an instruction generation unit 114, and at least one as a generation processing unit 115. Some of these functional components may be executed by the processor of the user terminal 200.

[0027] The style database 121 is a database in which the relationship between style support data and style features is predefined. Style support data is data set via the user terminal 200 by the support data setting unit 112 (described later), and includes at least one of the following: author, document genre, style category, and reference information. Style features are indicators that quantitatively or qualitatively represent the linguistic tendencies inherent in the author, document genre, or style category. Specifically, style features include at least one of the following: lexical features, syntactic features, rhythm, descriptive features, and rhetorical features.

[0028] A stylistic category refers to a classification of the writing style within a document. Stylistic categories may be defined according to the type of document. In the case of literary works, a stylistic category is defined as a category representing a literary school. Literary schools can be classified, for example, into the modern sensibility / psychological depiction school, the modern intellect / aesthetic school, the monologue / self-projection school, etc. Literary schools may also be classified differently depending on the country, region, or era. In the case of non-literary works, a stylistic category is defined as a type of descriptive format based on the degree of specialization, logic, etc. (for example, academic / specialized writing, simple / popular explanation, business / practical writing, etc.).

[0029] The style database 121 predefines style categories corresponding to authors or referenced documents, and first style features corresponding to authors, referenced documents, or style categories. The style database 121 also predefines second style features corresponding to document genres.

[0030] The style database 121 consists of style categories that classify multiple documents or authors based on the similarity of the stylistic features they share. Statistical methods such as cluster analysis using an analytical model can be used to define style categories. Specifically, the analytical model quantifies multiple stylistic elements extracted from documents (such as sentence length, frequency of metaphors, and part-of-speech usage ratios) and generates multiple style categories by classifying documents based on their similarity. These style categories are further classified by stylistic features unique to each category. When a large-scale language model is used as the analytical model, style categories are generated based on high-dimensional style vectors internally held by the model, or on the results of style summaries using natural language.

[0031] Figure 5 shows an example of the data structure diagram of the style database 121. The style database 121 may be constructed for each language area or country. Language areas include, for example, Japanese-speaking areas, English-speaking areas, Chinese-speaking areas, French-speaking areas, Spanish-speaking areas, etc., and style categories representing literary schools can be defined for each language area. In this disclosure, the style database 121 in the Japanese-speaking area will be used as an example.

[0032] Figure 5(a) shows an example of the data structure of the style database 121, which includes style categories, corresponding authors, and reference documents. Style category CA001 represents, for example, the style category "Contemporary Sensibility / Psychological Description," and is associated with one or more authors AU001 and AU002 that belong to that style category. Author AU001 is associated with one or more of the author's books (reference documents) DO001 and DO002. Note that a specific style category may be structured to be associated with one or more reference documents that belong to that style category.

[0033] Figure 5(b) shows an example of the data structure of the style database 121, which includes a first stylistic feature, its corresponding style category, author, and reference document. The first stylistic feature FE001-1 is a stylistic feature corresponding to style category CA001 "Modern Sensibility / Psychological Description," and for example, it describes stylistic features such as "masterful use of metaphor, detachment (sense of distance), urban solitude, and serene psychology." In this embodiment, the first stylistic feature and the style category have a corresponding relationship. The style category is linked to one or more authors classified under that style category, and the author is linked to one or more reference documents which are works by that author. Note that a particular first stylistic feature may be linked to one or more authors and one or more reference documents that possess that first stylistic feature.

[0034] Figure 5(c) shows an example of the data structure of the style database 121 for a second stylistic feature and its corresponding document genre. The second stylistic feature FE001-2 is a stylistic feature corresponding to the document genre "thriller," and for example, it describes stylistic features such as "intense sense of speed, emphasis on the five senses, discontinuous information disclosure, and existential threat." Note that the document genre includes, for example, thriller, gothic, science fiction, fable, private detective novel, etc., but is not limited to these genres or the number of genres. Furthermore, the second stylistic feature or document genre may be linked to reference documents classified under that document genre.

[0035] The document setting unit 111 accepts the setting of source document data to be converted. The document setting unit 111 accepts the setting of source document data by receiving the source document data transmitted via the user terminal 200. In addition, the memory 120 stores multiple source document data, and the document setting unit 111 accepts the setting of source document data by receiving the specification of the source document data to be converted from among the multiple source document data stored in the memory 120 or an external database via the user terminal 200.

[0036] The auxiliary data setting unit 112 accepts the setting of style auxiliary data, which includes the author, document genre, style category, and at least one piece of reference information, via the user terminal 200. Style auxiliary data is data for defining the style of the converted document data. Specifically, the style of the source document data is converted based on the style characteristics identified by the style auxiliary data. That is, the style of the converted document data will reflect the style characteristics identified by the style auxiliary data. Reference information refers to information such as links, identifiers, and document titles for identifying reference documents. Reference documents indicate documents used for referencing style characteristics.

[0037] The auxiliary data setting unit 112 accepts attribute data, including at least one of the user's nationality, native language, and age, via the user terminal 200. Attribute data is data that defines the expression of the converted document data. Specifically, the expression of the source document data is converted based on expression features identified by the expression features identified based on the attribute data. In other words, the expression of the converted document data reflects the expression features identified by the attribute data. Here, expression features refer to tendencies in linguistic expression that are suitable for specific reader attributes, such as vocabulary difficulty, choice of metaphors, and adjustments to descriptions based on cultural background.

[0038] The auxiliary data setting unit 112 can accept the setting of style auxiliary data by referring to the style database 121 and receiving the specification of registered authors, document genres, style categories, and reference documents via the user terminal 200. The auxiliary data setting unit 112 may also accept the setting of style auxiliary data via the user terminal 200 that includes authors, document genres, or reference documents not registered in the style database 121.

[0039] The analysis unit 113 receives analysis requests for the target document, including authors, document genres, or reference documents not registered in the style database 121, via the user terminal 200. The analysis unit 113 has the analysis model analyze the style features of the target document. The analysis model is a machine learning model and is stored in the memory 220 or the memory of the generation device 300. The machine learning model can be a large-scale language model, and the generation model provided by the generation device 300 may be used as the analysis model.

[0040] The analysis unit 113 inputs the object to be analyzed into an analysis model and obtains output of stylistic features from the analysis model. The analysis unit 113 determines the similarity between the analyzed stylistic features and stylistic features associated with existing stylistic categories registered in the stylistic database 121. The similarity can be determined by inputting the analyzed stylistic features and existing stylistic features into a large-scale language model. Alternatively, the similarity may be determined, for example, by vectorizing each stylistic feature and determining the similarity between the vectors. The similarity determination is performed on all or some of the existing stylistic categories registered in the stylistic database 121.

[0041] The analysis unit 113 determines whether the similarity score satisfies the similarity conditions. The analysis unit 113 can determine whether the similarity score satisfies the similarity conditions by inputting the similarity score and the similarity conditions into a large-scale language model. In this case, the similarity conditions may be text in which the conditions are defined in natural language. Alternatively, the similarity conditions may be, for example, a threshold for cosine similarity, and if the threshold is met, the stylistic features may be determined to be similar. If the similarity score satisfies the similarity conditions, the analysis unit 113 determines that the analyzed stylistic features are similar to stylistic features associated with existing stylistic categories.

[0042] The analysis unit 113 updates the style database 121 by associating the analysis target with existing style categories determined to be similar, if the similarity score meets the similarity criteria. Specifically, authors or reference documents included in the analysis target are registered in association with a first style characteristic corresponding to the existing style category. Furthermore, if the similarity score meets the similarity criteria, the analysis unit 113 updates the style database 121 by associating document genres or reference documents included in the analysis target with existing document genres determined to be similar to them. Document genres or reference documents included in the analysis target are registered in association with a second style characteristic corresponding to the existing document genre.

[0043] If the similarity score does not meet the similarity criteria, the analysis unit 113 generates a new style category corresponding to the document being analyzed. The analysis unit 113 registers the document being analyzed in the style database 121, associating it with the newly generated style category. The new style category is registered in the style database 121, associating it with the style features analyzed by the analysis unit 113. Specifically, authors or reference documents included in the document being analyzed are registered in association with the new style category. Similarly, if the similarity score does not meet the similarity criteria, the analysis unit 113 generates a new document genre corresponding to the document genre or reference documents included in the document being analyzed, and registers the document genre or reference documents in the style database 121, associating them with the new document genre.

[0044] In this embodiment, the style database 121 is shared among multiple user terminals 200. That is, the style database 121 is expanded by specific authors, document genres, and reference documents (books) newly set by the user, enabling the conversion of documents based on the style characteristics of new style categories and the style characteristics of new document genres.

[0045] The instruction generation unit 114 generates generation instructions for converting the source document data based on the stylistic features identified by the stylistic supplementary data. Specifically, the instruction generation unit 114 refers to the stylistic database 121, identifies a stylistic feature that includes a first stylistic feature and a second stylistic feature corresponding to the data included in the stylistic supplementary data, and generates generation instructions based on that stylistic feature. The generation instructions are instructions for a generation model that reproduces the identified stylistic features in the converted document data.

[0046] The generation instructions are, for example, textual descriptions of instructions to thoroughly implement the syntax, vocabulary selection, and rhythm of a specified "first stylistic feature" as a generation rule in the generation model. Similarly, the generation instructions also include textual descriptions of instructions to mix vocabulary, gadgets, and atmosphere specific to the document genre of a specified "second stylistic feature" into the main text. Here, the "first stylistic feature" and the "second stylistic feature" are set as variables, and at least one of the first and second stylistic features registered in the stylistic database 121 is referenced.

[0047] The instruction generation unit 114 generates generation instructions for converting the source document data based on the expression features identified by the attribute data. In this embodiment, the generation instructions include text describing the expression features corresponding to the attribute data. The instruction generation unit 114 may also generate generation instructions based on identified expression features by referring to an attribute database in which expression features and their corresponding attribute data are predefined. The attribute database is stored in memory 120. The attribute database defines, for example, the correspondence between a first expression feature and the corresponding user's age, a second expression feature and the corresponding user's nationality, and a third expression feature and the corresponding user's native language.

[0048] Attribute data such as age is classified by categories such as teenagers, adults, and the elderly. The corresponding expression features are defined as "dynamism, empathy, and simple yet sharp language" for age "teenagers." Attribute data such as nationality is classified by categories such as Japan, the United States, and Germany. The corresponding expression features are defined as "a sense of the seasons, shared silence, and individuality within a group" for nationality "Japan." Attribute data and expression features are not limited to these examples and may be further subdivided depending on the attribute. Generation instructions are, for example, generation rules in a generative model, written as text instructions that transform a document to fit the specified "attribute data" or "expression features."

[0049] The instruction generation unit 114 can generate generation instructions by combining identified stylistic features and expression features. A generation instruction is, for example, a text description of an instruction to transform a specified "source text" based on "variables." Here, "source text" refers to the source document data. The "variables" refer to at least one selected from document category (first stylistic feature), document genre (second stylistic feature), age (first expression feature), nationality (second expression feature), and native language (third expression feature). This generation instruction enables the transformation of source document data with a style corresponding to the user's preferred author or document genre, and with expressions that suit the user's age and nationality.

[0050] The generation processing unit 115 inputs the generation instruction and the source document data into the generation model and obtains the output of converted document data. Specifically, the generation processing unit 115 transmits the generation instruction and the source document data to the generation device 300 and inputs the generation instruction and the source document data into the generation model of the generation device 300. The generation device 300 generates converted document data by converting the source document data according to the generation instruction. The generation processing unit 115 can acquire the converted document data generated by the generation device 300 and output it to the user terminal 200. This allows the user to receive converted document data that has been transformed to a writing style corresponding to their preferred author and document genre, and to an expression that suits the user's age and nationality.

[0051] The generation processing unit 115 can divide the source document data according to its data capacity. The data capacity is determined by factors such as the number of tokens and characters. The generation processing unit 115 assigns an index to the divided source document data and inputs the divided source document data and generation instructions into the generation model. At this time, multiple divided source document data may be input into multiple generation models and processed in parallel. The generation processing unit 115 obtains multiple divided converted document data corresponding to each divided source document from the generation model. The generation processing unit 115 can refer to the index assigned during division, combine the multiple divided converted document data, and generate a single converted document data corresponding to the source document data. Alternatively, the generation processing unit 115 may convert the divided source document data in index order and output the obtained converted document data sequentially to the user terminal 200 in index order.

[0052] The document conversion device 100 may restrict the use of the generated converted document data on the user terminal 200. Specifically, the document conversion device 100 processes the text of the converted document based on the converted document data and outputs a converted document viewing screen (not shown) to the user terminal 200. The user terminal 200 views the converted document via the converted document viewing screen. The document conversion device 100 restricts operations on the converted document viewing screen by the user terminal 200, such as copying text, taking screenshots of the screen, and other operations related to the duplication of the converted document. In this way, the document conversion device 100 can restrict the converted document to the user's private use.

[0053] Figure 6 shows a processing flowchart of the document conversion method. The document conversion method is realized by the functional configuration of the document conversion device 100.

[0054] In the document setting process of step S101, the document conversion device 100 receives the setting of the source document data to be converted from the user terminal 200. In the auxiliary data setting step of process S102, the document conversion device 100 receives the setting of style auxiliary data from the user terminal 200, which includes at least one of the author, document genre, style category, and reference information for identifying the referenced document. The document conversion device 100 also receives the setting of attribute data from the user terminal 200, which includes at least one of the user's nationality, native language, and age. In step S103, the document conversion device 100 determines whether the style-related data received in step S102 contains data not registered in the style database 121. If it does not contain data not registered (NO in step S103), the document conversion device 100 executes the process in step S105. If the data includes unregistered information (YES in step S103), in the analysis step of step S104, the document conversion device 100 accepts an analysis request for the target of analysis that includes authors, document genres, or reference documents not registered in the style database 121. The document conversion device 100 analyzes the style features of the target of analysis using an analysis model and determines the similarity between the analyzed style features and the style features of existing style categories registered in the style database 121. If the two are similar, the document conversion device 100 updates the style database 121 by associating the target of analysis with an existing style category. If the two are not similar, the document conversion device 100 generates a new style category corresponding to the target of analysis and registers the target of analysis in the style database by associating it with the newly generated style category. In the instruction generation process of step S105, the document conversion device 100 generates generation instructions for converting the source document data based on the stylistic features identified by the stylistic supplementary data. The document conversion device 100 can also generate generation instructions by combining the stylistic features with the expression features identified by the attribute data. In the generation process of step S106, the document conversion device 100 inputs the generation instruction and the source document data into the generation model and obtains the output of converted document data. The document conversion device 100 can output the obtained converted document data to the user terminal 200.

[0055] The disclosures herein include the following document conversion systems, apparatus, methods, and programs.

[0056] (Item 1) A document conversion system comprising: a document setting unit that accepts the setting of source document data to be converted; an auxiliary data setting unit that accepts the setting of style auxiliary data including at least one of the author, document genre, style category, and reference information for identifying reference documents; an instruction generation unit that generates generation instructions for converting the source document data based on the style features identified by the style auxiliary data; and a generation processing unit that inputs the generation instructions and the source document data into a generation model and obtains output of converted document data. (Item 2) The document conversion system according to item 1, wherein the auxiliary data setting unit accepts the setting of attribute data including at least one of the user's nationality, native language, and age for defining the representation of the converted document data, and the instruction generation unit generates the generation instruction by combining the stylistic features and the expression features identified by the attribute data. (Item 3) The document conversion system according to item 1 or item 2, wherein the instruction generation unit refers to a style database in which a style category corresponding to the author or the reference document and a first style feature corresponding to the author, the reference document, or the style category are defined in advance, and generates the generation instruction based on the first style feature. (Item 4) A document conversion system according to any one of item 3, wherein the style database predefines a second style characteristic corresponding to the document genre, and the instruction generation unit generates the generation instruction based on the first style characteristic and the second style characteristic. (Item 5) A document conversion system according to item 3 or item 4, comprising an analysis unit that accepts analysis requests for the subject to be analyzed, including authors, document genres, or reference documents not registered in the style database, wherein the analysis unit analyzes the style features of the subject to be analyzed using an analysis model, determines the similarity between the analyzed style features and the style features of existing style categories registered in the style database, and updates the style database by associating the subject to be analyzed with an existing style category if the similarity conditions are met. (Item 6) The document conversion system according to any of item 5, wherein the analysis unit generates a new style category corresponding to the analysis target if the similarity score does not satisfy the similarity conditions, and registers the analysis target in the style database in association with the style category. (Item 7) The aforementioned source document data is a literary work, and the aforementioned style category is a category representing a literary school obtained by analyzing the aforementioned literary work using an analytical model, and the document conversion system is as described in any of items 1 to 6. (Item 8) A document conversion method comprising: a document setting step in which a computer accepts the settings of source document data to be converted; an auxiliary data setting step in which a computer accepts the settings of style auxiliary data including at least one of the author, document genre, style category, and reference information for identifying reference documents; an instruction generation step in which an instruction is generated to convert the source document data based on the style features identified by the style auxiliary data; and a generation step in which the instruction is input into a generation model and the source document data is output to convert document data. (Item 9) A document conversion program that causes a computer to function in order to perform the document conversion method described in item 8.

[0057] This disclosure is not limited to the embodiments described above, and various modifications, improvements, and variations are possible without departing from the spirit and scope of this disclosure. [Explanation of symbols]

[0058] 1. Document conversion system 100 Document conversion devices 110 processors 111 Document Settings Section 112 Auxiliary Data Setting Unit 113 Analysis Department 114 Instruction generation section 115 Generation Processing Unit 120 memory 121 Style Database 130 Communication Interfaces 200 user terminals 300 generator

Claims

1. A document setting unit that accepts the settings for the source document data to be converted, An auxiliary data setting unit that accepts the setting of style-related auxiliary data, which includes at least one of the following: author, document genre, style category, and reference information for identifying the referenced document. An instruction generation unit generates generation instructions for converting the source document data based on the stylistic features identified by the stylistic supplementary data, A generation processing unit inputs the generation instruction and the source document data into a generation model and obtains the output of converted document data. It includes an analysis unit that accepts analysis requests for documents that include authors, document genres, or reference documents not registered in the style database, The instruction generation unit refers to a style database in which a style category corresponding to the author or the reference document and a first style feature corresponding to the author, the reference document, or the style category are defined in advance, and generates the generation instruction based on the first style feature. The aforementioned analysis unit, The stylistic features of the subject of analysis are analyzed using an analysis model. The degree of similarity between the analyzed stylistic features and the stylistic features of existing stylistic categories registered in the stylistic database is determined. A document conversion system that, if the similarity score satisfies the similarity conditions, updates the style database by associating the analysis target with an existing style category.

2. The auxiliary data setting unit accepts the setting of attribute data, including at least one of the user's nationality, native language, and age, for defining the representation of the converted document data. The document conversion system according to claim 1, wherein the instruction generation unit generates the generation instruction by combining the stylistic features and the expression features identified by the attribute data.

3. The document conversion system according to claim 1, wherein the style database predefines a second style characteristic corresponding to the document genre, and the instruction generation unit generates the generation instruction based on the first style characteristic and the second style characteristic.

4. The document conversion system according to claim 1, wherein the analysis unit generates a new style category corresponding to the analysis target if the similarity score does not satisfy the similarity conditions, and registers the analysis target in the style database in association with the style category.

5. The aforementioned original document data is a literary work, The document conversion system according to claim 1 or claim 2, wherein the style category is a category representing a literary school obtained by analyzing the literary work using an analytical model.

6. Computers A document setup process that accepts the settings for the source document data to be converted, A supplementary data setting process that accepts the setting of style supplementary data including at least one of the author, document genre, style category, and reference information for identifying the referenced document, An instruction generation step that generates generation instructions for converting the source document data based on the stylistic features identified by the stylistic supplementary data, A generation process in which the generation instructions and the source document data are input to a generation model to obtain output of converted document data, The analysis process accepts requests for analysis of authors, document genres, or reference documents not registered in the style database, and then executes the analysis. The instruction generation step refers to a style database in which a style category corresponding to the author or the reference document and a first style feature corresponding to the author, the reference document, or the style category are defined in advance, and generates the generation instruction based on the first style feature. The aforementioned analysis step is The stylistic features of the subject of analysis are analyzed using an analysis model. The degree of similarity between the analyzed stylistic features and the stylistic features of existing stylistic categories registered in the stylistic database is determined. A document transformation method that, if the similarity score satisfies the similarity conditions, updates the style database by associating the analysis target with an existing style category.

7. A document setting unit that accepts the setting of source document data to be converted, An auxiliary data setting unit that accepts the setting of style-related auxiliary data, which includes at least one of the following: author, document genre, style category, and reference information for identifying the referenced document. An instruction generation unit generates generation instructions for converting the source document data based on the stylistic features identified by the stylistic supplementary data, A generation processing unit inputs the generation instruction and the source document data into a generation model and obtains the output of converted document data. It includes an analysis unit that accepts analysis requests for documents that include authors, document genres, or reference documents not registered in the style database, The instruction generation unit refers to a style database in which a style category corresponding to the author or the reference document and a first style feature corresponding to the author, the reference document, or the style category are defined in advance, and generates the generation instruction based on the first style feature. The aforementioned analysis unit, The stylistic features of the subject of analysis are analyzed using an analysis model. The degree of similarity between the analyzed stylistic features and the stylistic features of existing stylistic categories registered in the stylistic database is determined. A document conversion system that, if the aforementioned similarity score does not satisfy the similarity conditions, generates a new style category corresponding to the subject to be analyzed, and registers the subject to be analyzed in the style database in association with the said style category.

8. A computer, A document setup process that accepts the settings for the source document data to be converted, A supplementary data setting process that accepts the setting of style supplementary data including at least one of the author, document genre, style category, and reference information for identifying the referenced document, An instruction generation step that generates generation instructions for converting the source document data based on the stylistic features identified by the stylistic supplementary data, A generation process in which the generation instructions and the source document data are input to a generation model to obtain output of converted document data, The analysis process accepts requests for analysis of authors, document genres, or reference documents not registered in the style database, and then executes the analysis. The instruction generation step refers to a style database in which a style category corresponding to the author or the reference document and a first style feature corresponding to the author, the reference document, or the style category are defined in advance, and generates the generation instruction based on the first style feature. The aforementioned analysis step is The stylistic features of the subject of analysis are analyzed using an analysis model. The degree of similarity between the analyzed stylistic features and the stylistic features of existing stylistic categories registered in the stylistic database is determined. A document conversion method that, if the aforementioned similarity score does not satisfy the similarity conditions, generates a new style category corresponding to the subject to be analyzed, and registers the subject to be analyzed in the style database in association with the said style category.

9. A document conversion program that causes a computer to function to perform the document conversion method described in Claim 6 or Claim 8.

Images