Method and device for generating research report based on generative large model and storage medium

Abstract

The invention provides a method and device for generating a research report based on a generative large model and a storage medium, and relates to the technical field of artificial intelligence and the like. According to the specific implementation scheme, the method comprises the steps of obtaining a title and description information of a research report based on obtained input information of a user; the description information is used for limiting the research report to be generated; obtaining knowledge information based on the title and the description information; based on the title and the description information, generating a target outline of the research report by adopting a pre-trained generative large model and adopting an at least two-step generation mode; the target outline comprises a plurality of chapter titles with a sequential relationship and target word numbers of chapter contents; and based on the target outline and the knowledge information, adopting the generative large model to generate the text content of the research report. According to the technology disclosed by the invention, the generation efficiency of the research report can be effectively improved.

Description

Technical Field

[0001] The present disclosure relates to the field of computer technology, specifically to the field of artificial intelligence and other technical fields, and in particular to a method, device and storage medium for generating a research report based on a generative large model. Background Art

[0002] The research report is mainly completed through a large amount of preliminary research on the corresponding subject area, statistical analysis of the data, and finally integrating it into a highly professional report.

[0003] In the prior art, considering the professionalism of research reports, professional practitioners in the field usually screen out highly relevant information based on the report topic through their understanding of the field and analysis of a large amount of factual information and data in the field, and conduct statistical drawing and in-depth analysis, etc., and finally write a complete report. Summary of the invention

[0004] The present disclosure provides a method, device and storage medium for generating a research report based on a generative large model.

[0005] According to one aspect of the present disclosure, a method for generating a research report based on a generative big model is provided, comprising:

[0006] Based on the input information obtained from the user, obtain the title and description information of the research report; the description information is used to define the research report to be generated;

[0007] Based on the title and the description information, acquiring knowledge information;

[0008] Based on the title and the description information, a pre-trained generative model is used to generate a target outline of the research report in at least two steps; the target outline includes a plurality of chapter titles having a sequential relationship and a target word count of each chapter content;

[0009] Based on the target outline and the knowledge information, the generative big model is adopted to generate the main content of the research report.

[0010] According to another aspect of the present disclosure, there is provided a device for generating a research report based on a generative large model, comprising:

[0011] An information acquisition module, used to acquire the title and description information of the research report based on the input information acquired from the user; the description information is used to define the research report to be generated;

[0012] A knowledge acquisition module, used for acquiring knowledge information based on the title and the description information;

[0013] An outline generation module is used to generate a target outline of the research report based on the title and the description information, using a pre-trained generative model and at least two steps of generation; the target outline includes a plurality of chapter titles in a sequential relationship and a target word count of each chapter content;

[0014] A content generation module is used to generate the main content of the research report based on the target outline and the knowledge information using the generative big model.

[0015] According to another aspect of the present disclosure, there is provided an electronic device, comprising:

[0016] at least one processor; and

[0017] a memory communicatively connected to the at least one processor; wherein,

[0018] The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method of any possible implementation manner and the aspects described above.

[0019] According to yet another aspect of the present disclosure, a non-transitory computer-readable storage medium storing computer instructions is provided, wherein the computer instructions are used to cause the computer to execute the method of the above-mentioned aspect and any possible implementation manner.

[0020] According to yet another aspect of the present disclosure, a computer program product is provided, including a computer program, wherein when the computer program is executed by a processor, the computer program implements the above-mentioned aspects and any possible implementation method.

[0021] According to the technology disclosed in the present invention, the generation efficiency of the research report can be effectively improved. It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present invention, nor is it intended to limit the scope of the present invention. Other features of the present invention will become easy to understand through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] The accompanying drawings are used to better understand the present solution and do not constitute a limitation of the present disclosure.

[0023] Figure 1 is a schematic diagram according to a first embodiment of the present disclosure;

[0024] Figure 2 is a schematic diagram according to a second embodiment of the present disclosure;

[0025] Figure 3 is a schematic diagram according to a third embodiment of the present disclosure;

[0026] Figure 4 is a schematic diagram according to a fourth embodiment of the present disclosure;

[0027] Figure 5 The block diagram is a block diagram of an electronic device for implementing the method of the embodiment of the present disclosure. DETAILED DESCRIPTION

[0028] The following is a description of exemplary embodiments of the present disclosure in conjunction with the accompanying drawings, including various details of the embodiments of the present disclosure to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be recognized by those of ordinary skill in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.

[0029] Obviously, the described embodiments are only part of the embodiments of the present disclosure, but not all of them. Based on the embodiments of the present disclosure, all other embodiments obtained by ordinary technicians in the field without creative work are within the scope of protection of the present disclosure.

[0030] It should be noted that the terminal devices involved in the embodiments of the present disclosure may include but are not limited to mobile phones, personal digital assistants (PDAs), wireless handheld devices, tablet computers and other smart devices; display devices may include but are not limited to personal computers, televisions and other devices with display functions.

[0031] In addition, the term "and / or" in this article is only a description of the association relationship between the associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship.

[0032] Figure 1 is a schematic diagram according to the first embodiment of the present disclosure; Figure 1 As shown, this embodiment provides a method for generating a research report based on a generative large model, which may specifically include the following steps:

[0033] S101, based on the input information obtained from the user, obtain the title and description information of the research report; the description information is used to limit the research report to be generated;

[0034] In this embodiment, the user's input information may be text information, or may also be voice information, and the text information may be obtained by using voice recognition technology.

[0035] The description information of this embodiment can be understood as various limiting information of the user on the research report to be generated. For example, the description information may include the content that is desired to be reflected in the research report. For example, the description information may be in the form of a string and / or a sentence to limit the content that needs to be reflected in the research report. Specifically, the description information may include one, two or more strings, and may also include one, two or more sentences. Alternatively, it can also be considered that the description information includes conditional information that needs to be referenced when generating the research report.

[0036] In an application scenario of this embodiment, the user's input information may be: help generate a research report on B in field A, which may include analysis of knowledge points C and D. The title of the research report obtained at this time may be: research on B in field A, and the description information may include knowledge points C and D. For another example, the user's input information may also include the expected number of words in the research report, the expected structure of the outline, etc., which can be used as description information to limit the research report, so that the description information can be referred to in the future to generate a research report that meets the user's needs.

[0037] In this embodiment, the process of obtaining the title and description information of the research report from the user's input information can be implemented by using information mining technology.

[0038] In actual applications, the user's input information may also carry other descriptive information of the research report, which will not be listed here one by one.

[0039] S102, acquiring knowledge information based on the title and description information;

[0040] This step S102 can be understood as a knowledge information retrieval process. When generating a research report, it is far from enough to rely solely on the title and description information. It is necessary to obtain knowledge information related to the title and description information to provide rich data support for the subsequent generation of the research report.

[0041] S103, based on the title and description information, using a pre-trained generative large model, using at least two steps of generation method, to generate a target outline of the research report; the target outline includes multiple chapter titles with a sequential relationship and a target number of words for each chapter content;

[0042] The generation of the target outline of the research report of this embodiment is not a one-time generation, but a generation method of at least two steps, which can enrich the structure of the target outline of the research report, making the target outline structure of the research report more accurate and reasonable. Moreover, the target outline of the research report generated by this embodiment can also include the target number of words for each chapter content. For less important chapters, the target number of words can be less, and for more important chapters, the target number of words can be more, which can provide accurate guidance for the generation of the content of subsequent chapters.

[0043] S104. Based on the target outline and knowledge information, a generative big model is used to generate the main content of the research report.

[0044] Specifically, the generative model can generate the content of each chapter in the target outline based on the target outline and refer to the knowledge information, that is, obtain the main content of the research report. Among them, when generating the content of each chapter, it is necessary to refer to the target number of words of the corresponding chapter in the target outline. In other words, the actual number of words in the content of each generated chapter should theoretically be greater than or equal to the target number of words of the corresponding chapter in the target outline.

[0045] In this embodiment, the generative large model used can also be called a generative language model (General Language Model; GLM), or a generative large language model.

[0046] The method of generating a research report based on a generative large model in this embodiment can automatically and intelligently generate a research report by adopting the above-mentioned scheme. Compared with the manual generation in the prior art, it can not only effectively shorten the time consumption of the research report and effectively improve the generation efficiency of the research report; but also effectively ensure the accuracy of the generated research report.

[0047] Figure 2 is a schematic diagram according to the second embodiment of the present disclosure; the method of generating a research report based on a generative large model in this embodiment, in the above Figure 1 Based on the technical solutions of the embodiments shown in the figure, the technical solutions of the present disclosure are further described in more detail. Figure 2 As shown, the method for generating a research report based on a generative large model in this embodiment may specifically include the following steps:

[0048] S201, based on the input information obtained from the user, using a pre-trained information acquisition model to obtain the title and description information of the research report; the description information is used to limit the research report to be generated;

[0049] In this embodiment, the pre-trained information acquisition model may be a base class large model, which can acquire effective information from the user's input information and obtain the title and description information of the research report.

[0050] S202, based on the title and description information, using a professional information acquisition module to acquire professional knowledge from a pre-created professional knowledge base;

[0051] S203, based on the title and description information, using a real-time information acquisition module to acquire real-time information knowledge through an external search engine and a third-party data source;

[0052] In this embodiment, for each field, a professional knowledge base of the field may be created in advance, which may include all the professional knowledge of the field.

[0053] That is to say, when this embodiment is applied, the field to which the research report belongs may be known in advance, and the professional knowledge base corresponding to the field may also be created in advance.

[0054] The professional information acquisition module can search the pre-created professional knowledge base in the field to which the research report belongs based on the title and description information to obtain professional knowledge. During the specific search, the title and description information can be first segmented at multiple granularities, and then each word obtained by the segmentation can be searched in the professional knowledge base at the granularity to obtain all the professional knowledge related to the words. For example, the professional knowledge related to each word can include the explanation of the word in the field to which the research report belongs, and can also include the application scenarios and application examples of the word in the field, etc. The multi-granularity segmentation of this embodiment can refer to the segmentation of all granularities. For example, for the string XYZ, the multi-granularity segmentation of this embodiment can include X, Y, and Z segmentations, and can also include segmentations of XY and YZ granularities, and can also include segmentations of XYZ granularity.

[0055] The real-time information acquisition module can obtain real-time information knowledge from the public network and third-party data sources through an external search engine. The real-time information knowledge can be obtained in the same way as the professional knowledge, or by retrieving the knowledge related to each word after multi-granular word segmentation. For example, the real-time information knowledge can include at least one aspect of information such as the explanation, application scenario, and application example of each word in the general field of the public network, as well as all relevant data information of each word in the third-party data source. Compared with professional knowledge, real-time information knowledge is not limited to the field to which the research report belongs, and can effectively supplement professional knowledge, making the knowledge referenced when generating the research report richer and more comprehensive.

[0056] In this embodiment, steps S202-S203 are as described above. Figure 1An implementation of step S102 of the illustrated embodiment. The professional knowledge acquired in step S202 can effectively ensure the professionalism of the research report, and the real-time information knowledge acquired in step S203 can effectively ensure the real-time nature of the research report, and can also enable the research report to be integrated with the knowledge in the general field, making it more comprehensive. Optionally, step S203 can also be deleted, and only step S202 is included. Of course, when step S203 is included, the acquired knowledge and information content is more comprehensive and rich, which can effectively ensure that the content of the generated research report is more accurate and reasonable.

[0057] S204. Based on the title and description information, a generative macro model is used to generate a preliminary outline of the research report;

[0058] The primary outline of the research report of this embodiment can be considered as an outline with a relatively simple structure. Compared with the target outline, the primary outline does not include enough architectural levels.

[0059] S205, based on the primary outline, title and description information of the report, using a generative macro model and at least one step of generation, generating a structure of a target outline of the research report; the structure of the target outline of the research report includes a plurality of chapter titles having a sequential relationship;

[0060] Specifically, the generative model can be based on the primary outline, further refer to the title and description information, enrich the structure of the primary outline, and generate the target outline.

[0061] Optionally, in this embodiment, a two-step generation method can be used to generate the structure of the target outline. A multi-step cycle generation method can also be used to continuously utilize the outline generated in the previous step, and refer to the title and description information, and use a generative model to continue to enrich the generated outline until the target outline of the research report is generated. The specific number of steps required for the outline generation process can be pre-set. In short, in the process of generating the target outline, the more generation steps are used, the richer the structure of the generated target outline.

[0062] S206, using a generative large model to predict the target number of words included in the content of each chapter in the target outline;

[0063] In this embodiment, the generative large model can directly predict the target number of words included in the content of each chapter in the target outline.

[0064] Alternatively, in an embodiment of the present disclosure, if the description information includes the expected number of words in the research report, then step S206 may further include the following steps:

[0065] (1) Based on the descriptive information, obtain the expected number of words in the research report;

[0066] This step can also be implemented using a pre-trained information acquisition model.

[0067] (2) Based on the expected word count of the research report, a generative large model is used to predict the target word count of each chapter in the target outline.

[0068] When predicting the target number of words in each chapter of the target outline, the generative model can refer to the expected number of words in the research report. For example, for more important chapters in the target outline, the predicted target number of words is higher, and for some less important chapters, the predicted target number of words can be appropriately lower. In short, the sum of the target number of words in all chapters should theoretically be greater than or equal to the expected number of words in the research report. In this way, it can effectively ensure that the predicted target number of words in each chapter is more accurate.

[0069] In this embodiment, steps S204-S206 are as described above. Figure 1 An implementation of step S103 of the illustrated embodiment.

[0070] S207, based on the target outline, professional knowledge and real-time information knowledge, use the generative big model to generate the main content of the research report;

[0071] Specifically, the generative big model can generate the content of each chapter in the target outline based on professional knowledge and real-time information knowledge, that is, obtain the main content of the research report. Based on this, the title and main content of the research report have been generated. If no other content is needed, the research report can be integrated and generated based on the title, target outline and main content of the research report.

[0072] Optionally, in an embodiment of the present disclosure, when implementing step S207, the following steps may be specifically adopted:

[0073] (a) Based on professional knowledge, real-time information knowledge, and the chapter titles in the target outline, a generative large model is used to generate the content of each chapter;

[0074] That is to say, for each chapter, the generative big model can generate the content of the chapter based on professional knowledge, real-time information knowledge and the title of the chapter.

[0075] (b) For any chapter, if the word count of the chapter content is less than the corresponding target word count, based on professional knowledge, real-time information knowledge, and the generated chapter content and the chapter title corresponding to the chapter, a generative large model is used to expand the chapter content until the word count of the chapter content is equal to or greater than the corresponding target word count.

[0076] In this embodiment, the step of expanding the chapter content may include only one round of expansion, or two rounds of expansion or multiple rounds of expansion. After each round of expansion, it is necessary to detect whether the number of words in the content of the expanded chapter is equal to or greater than the corresponding target number of words. If so, the process ends; otherwise, the expansion continues.

[0077] It should be noted that the generation of the content of each chapter can be independent or can be mutually influenced, so that the context of the research report is more coherent. For example, the content of each chapter can be generated in sequence according to the order of the position of each chapter from front to back in the outline. When the content of other chapters other than the first chapter is generated, the content of the first preset number of chapters before the current chapter can be used as context information and input into the generative large model together. The generative large model refers to the context information at the same time to generate the content of the current chapter, so that the context content of the generated research report is more coherent. The first preset number can be set according to actual needs or experience, for example, it can be 1, 2 or more, which is not limited here.

[0078] In the present embodiment, in the process of specifically generating the contents of each chapter in the research report, in the first round of generation, the contents of each chapter can be generated in sequence according to the above-mentioned method. Afterwards, it is detected whether the number of words in the contents of each chapter reaches the target number of words corresponding to the chapter specified in the target outline. If not reached, then based on professional knowledge, real-time information knowledge, and the contents of the generated chapters and the chapter titles corresponding to the chapters, the generative large model is used to expand the contents of the chapters. It should be noted that when expanding the contents of chapters, the contents of the second preset number of chapters before and after the current chapter can be used as context information and input into the generative large model together. The generative large model simultaneously refers to the context information to expand the contents of the current chapter, so that the context content of the generated research report is more coherent.

[0079] Through the above method, it can be ensured that the content of each chapter in the generated research report is rich enough.

[0080] S208. Based on the content of the report body, using a generative macro model, generate at least one of a summary, a statement, an appendix, and a reference of the research report;

[0081] It should be noted that, in general, a research report needs to include each of the abstract, statement, appendix, and reference, and in specific implementation, the information content of each can also be obtained according to step S208. However, in actual applications, at least one of the above information can also be obtained according to specific needs.

[0082] S209. Generate a research report based on the research report's abstract, statement, appendix, and at least one of the references, as well as the research report's title, objective outline, and main body of the research report.

[0083] Specifically, the order of each part of information when integrating and generating a research report can refer to the order of each part of information in a manually written research report, or the order of each part can be limited according to needs, which is not limited here.

[0084] The method for generating a research report based on a generative large model in this embodiment can generate a target outline of the research report through at least two steps, which can effectively ensure that the structure of the generated target outline is richer and more accurate.

[0085] Moreover, in this embodiment, the expected number of words in the research report can also be obtained, and based on the expected number of words in the research report, a generative large model can be used to accurately predict the target number of words included in the content of each chapter in the target outline, thereby further effectively enriching the content of the target outline.

[0086] Furthermore, in this embodiment, the knowledge information acquired based on the title and description information includes not only professional knowledge but also real-time information knowledge, thus providing effective data support for the generation of research report content.

[0087] Furthermore, in this embodiment, the content of each chapter in the research report may be generated through multiple expansions, which can effectively ensure the richness of the content of the generated research report.

[0088] In summary, by adopting the above solution, this embodiment can generate professional research reports in related fields quickly, efficiently and with high quality.

[0089] Figure 3 is a schematic diagram according to the third embodiment of the present disclosure; Figure 3 As shown, this embodiment provides a device 300 for generating a research report based on a generative large model, including:

[0090] The information acquisition module 301 is used to acquire the title and description information of the research report based on the input information acquired from the user; the description information is used to define the research report to be generated;

[0091] A knowledge acquisition module 302, configured to acquire knowledge information based on the title and the description information;

[0092] An outline generation module 303 is used to generate a target outline of the research report based on the title and the description information by using a pre-trained generative model and at least two steps of generation; the target outline includes a plurality of chapter titles in a sequential relationship and a target word count of each chapter content;

[0093] The content generation module 304 is used to generate the main content of the research report based on the target outline and the knowledge information using the generative big model.

[0094] The device 300 for generating a research report based on a generative big model in this embodiment, by adopting the above-mentioned modules, realizes the realization principle and technical effect of generating a research report based on a generative big model, which is the same as the above-mentioned related method embodiments. For details, please refer to the records of the above-mentioned related method embodiments, which will not be repeated here.

[0095] Figure 4 is a schematic diagram according to a fourth embodiment of the present disclosure; Figure 4 As shown, the device 400 for generating a research report based on a generative large model in this embodiment includes Figure 3 The apparatus 300 for generating a research report based on a generative large model has modules with the same names and functions as: an information acquisition module 401 , a knowledge acquisition module 402 , an outline generation module 403 and a content generation module 404 .

[0096] like Figure 4 As shown, the outline generation module 403 includes:

[0097] A first outline generating unit 4031 is used to generate a primary outline of the research report based on the title and the description information using the generative model;

[0098] A second outline generating unit 4032 is used to generate a structure of a target outline of the research report based on the primary outline of the research report, the title and the description information, using the generative model and at least one step of generating method; the structure of the target outline of the research report includes a plurality of chapters having a sequential relationship;

[0099] The prediction unit 4033 is used to use the generative large model to predict the target number of words included in the content of each chapter in the target outline.

[0100] Further optionally, the prediction unit 4033 is used to:

[0101] Based on the description information, obtaining the expected number of words of the research report;

[0102] Based on the expected word count of the research report, the generative macro model is used to predict the target word count of each chapter in the target outline.

[0103] Further optionally, if Figure 4 As shown, the knowledge acquisition module 402 includes:

[0104] A professional knowledge acquisition unit 4021 is used to acquire professional knowledge from a pre-created professional knowledge base using a professional information acquisition module based on the title and the description information;

[0105] The real-time knowledge acquisition unit 4022 is used to acquire real-time information knowledge from the public network and third-party data sources based on the title and the description information by using a real-time information acquisition module and an external search engine.

[0106] Further optionally, a content generation module 404 is used to generate the main content of the research report based on the target outline, the professional knowledge and the real-time information knowledge using the generative big model;

[0107] Further optionally, if Figure 4 As shown, the apparatus 400 for generating a research report based on a generative large model in this embodiment further includes:

[0108] The integration module 405 is used to integrate and generate the research report based on the title of the research report, the target outline and the main content of the research report.

[0109] Further optionally, if Figure 4 As shown, in the device 400 for generating a research report based on a generative large model in this embodiment, the content generation module 404 includes:

[0110] A content generation unit 4041 is used to generate the content of each chapter based on the professional knowledge, the real-time information knowledge and the titles of each chapter in the target outline using the generative macro model;

[0111] The content expansion unit 4042 is used to expand the content of any chapter, if the number of words in the content of the chapter is less than the corresponding target number of words, based on the professional knowledge, the real-time information knowledge, and the generated content of the chapter and the chapter title corresponding to the chapter, using the generative big model to expand the content of the chapter until the number of words in the content of the chapter is equal to or greater than the corresponding target number of words.

[0112] Further optionally, the content generating unit 4041 is further configured to:

[0113] Based on the content of the report body, using the generative macro model, generate at least one of the summary, statement, appendix and reference of the research report;

[0114] The integration module 405 is used to integrate and generate the research report based on at least one of the abstract, statement, appendix and reference of the research report, as well as the title of the research report, the objective outline and the main content of the research report.

[0115] The device 300 for generating a research report based on a generative big model in this embodiment, by adopting the above-mentioned modules, realizes the realization principle and technical effect of generating a research report based on a generative big model, which is the same as the above-mentioned related method embodiments. For details, please refer to the records of the above-mentioned related method embodiments, which will not be repeated here.

[0116] In the technical solution disclosed herein, the acquisition, storage and application of user personal information involved are in compliance with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0117] According to an embodiment of the present disclosure, the present disclosure also provides an electronic device, a readable storage medium and a computer program product.

[0118] Figure 5 A schematic block diagram of an example electronic device 500 that can be used to implement an embodiment of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present disclosure described and / or required herein.

[0119] like Figure 5 As shown, the device 500 includes a computing unit 501, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 502 or a computer program loaded from a storage unit 508 into a random access memory (RAM) 503. In the RAM 503, various programs and data required for the operation of the device 500 can also be stored. The computing unit 501, the ROM 502, and the RAM 503 are connected to each other via a bus 504. An input / output (I / O) interface 505 is also connected to the bus 504.

[0120] A number of components in the device 500 are connected to the I / O interface 505, including: an input unit 506, such as a keyboard, a mouse, etc.; an output unit 507, such as various types of displays, speakers, etc.; a storage unit 508, such as a disk, an optical disk, etc.; and a communication unit 509, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 509 allows the device 500 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0121] The computing unit 501 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the computing unit 501 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The computing unit 501 performs the various methods and processes described above, such as the above-mentioned methods of the present disclosure. For example, in some embodiments, the above-mentioned methods of the present disclosure may be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as a storage unit 508. In some embodiments, part or all of the computer program may be loaded and / or installed on the device 500 via ROM 502 and / or communication unit 509. When the computer program is loaded into RAM 503 and executed by the computing unit 501, one or more steps of the above-mentioned methods of the present disclosure described above may be performed. Alternatively, in other embodiments, the computing unit 501 may be configured to perform the above-mentioned methods of the present disclosure in any other appropriate manner (e.g., by means of firmware).

[0122] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0123] The program code for implementing the method of the present disclosure may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that the program code, when executed by the processor or controller, enables the functions / operations specified in the flow chart and / or block diagram to be implemented. The program code may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0124] In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0125] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0126] The systems and techniques described herein may be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), and the Internet.

[0127] A computer system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The relationship of client and server is generated by computer programs running on respective computers and having a client-server relationship with each other. The server may be a cloud server, a server of a distributed system, or a server combined with a blockchain.

[0128] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps recorded in this disclosure can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved, and this document does not limit this.

[0129] The above specific implementations do not constitute a limitation on the protection scope of the present disclosure. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present disclosure shall be included in the protection scope of the present disclosure.

Claims

1. A method for generating a research report based on a generative large model, comprising: Based on the input information obtained from the user, obtain the title and description information of the research report; The description information is used to define the research report to be generated; Based on the title and the description information, acquiring knowledge information; Based on the title and the description information, a pre-trained generative model is used to generate a target outline of the research report in at least two steps; the target outline includes a plurality of chapter titles having a sequential relationship and a target word count of each chapter content; Based on the target outline and the knowledge information, the generative big model is adopted to generate the main content of the research report.

2. The method according to claim 1, wherein: Based on the title and the description information, a pre-trained generative model is used to generate a target outline of the research report in at least two steps, including: Based on the title and the description information, using the generative macro model, generate a primary outline of the research report; Based on the primary outline of the research report, the title and the description information, the generative model is used to generate the structure of the target outline of the research report in at least one step of generation; the structure of the target outline of the research report includes a plurality of chapter titles having a sequential relationship; The generative large model is used to predict the target number of words included in the content of each chapter in the target outline.

3. The method according to claim 2, wherein: The generative model is used to predict the target word count of each chapter in the target outline, including: Based on the description information, obtaining the expected number of words of the research report; Based on the expected word count of the research report, the generative macro model is used to predict the target word count of each chapter in the target outline.

4. The method according to claim 1, wherein: Based on the title and the description information, knowledge information is acquired, including: Based on the title and the description information, using a professional information acquisition module to acquire professional knowledge from a pre-created professional knowledge base; Based on the title and the description information, a real-time information acquisition module is used to acquire real-time information knowledge from the public network and third-party data sources through an external search engine.

5. The method according to claim 4, wherein: Based on the target outline and the knowledge information, the generative big model is used to generate the main content of the research report, including: Based on the target outline, the professional knowledge and the real-time information knowledge, the generative big model is used to generate the main content of the research report; Furthermore, the method further comprises: The research report is generated based on the title of the research report, the objective outline and the main content of the research report.

6. The method according to claim 5, wherein: Based on the target outline, the professional knowledge and the real-time information knowledge, the generative big model is used to generate the main content of the research report, including: Based on the professional knowledge, the real-time information knowledge and the titles of each chapter in the target outline, the generative big model is used to generate the content of each chapter; For any chapter, if the number of words in the content of the chapter is less than the corresponding target number of words, based on the professional knowledge, the real-time information knowledge, and the generated content of the chapter and the chapter title corresponding to the chapter, the generative big model is used to expand the content of the chapter until the number of words in the content of the chapter is equal to or greater than the corresponding target number of words.

7. The method according to any one of claims 1 to 6, wherein: The method further comprises: Based on the content of the report body, using the generative macro model, generate at least one of the summary, statement, appendix and reference of the research report; The research report is generated based on at least one of the abstract, statement, appendix and reference of the research report, as well as the title of the research report, the objective outline and the main content of the research report.

8. A device for generating a research report based on a generative large model, comprising: An information acquisition module, used to acquire the title and description information of the research report based on the input information acquired from the user; The description information is used to define the research report to be generated; A knowledge acquisition module, used for acquiring knowledge information based on the title and the description information; An outline generation module is used to generate a target outline of the research report based on the title and the description information, using a pre-trained generative model and at least two steps of generation; the target outline includes a plurality of chapter titles in a sequential relationship and a target word count of each chapter content; A content generation module is used to generate the main content of the research report based on the target outline and the knowledge information using the generative big model.

9. The device according to claim 8, wherein: The outline generation module includes: A first outline generating unit, configured to generate a primary outline of the research report based on the title and the description information and using the generative model; A second outline generating unit is used to generate a structure of a target outline of the research report based on the primary outline of the research report, the title and the description information, using the generative model and at least one step of generating method; the structure of the target outline of the research report includes a plurality of chapter titles having a sequential relationship; The prediction unit is used to use the generative large model to predict the target number of words included in the content of each chapter in the target outline.

10. The device according to claim 9, wherein: The prediction unit is used for: Based on the description information, obtaining the expected number of words of the research report; Based on the expected word count of the research report, the generative macro model is used to predict the target word count of each chapter in the target outline.

11. The device according to claim 8, wherein: The knowledge acquisition module comprises: A professional knowledge acquisition unit, configured to acquire professional knowledge from a pre-created professional knowledge base using a professional information acquisition module based on the title and the description information; The real-time knowledge acquisition unit is used to acquire real-time information knowledge from the public network and third-party data sources based on the title and the description information by using a real-time information acquisition module and an external search engine.

12. The device according to claim 11, wherein The content generation module is used to generate the main content of the research report based on the target outline, the professional knowledge and the real-time information knowledge using the generative big model; Furthermore, the device also includes: An integration module is used to integrate and generate the research report based on the title of the research report, the target outline and the main content of the research report.

13. The device according to claim 12, wherein: The content generation module comprises: A content generation unit, configured to generate the content of each chapter based on the professional knowledge, the real-time information knowledge and the titles of each chapter in the target outline, using the generative macro model; The content expansion unit is used for, for any chapter, if the number of words in the content of the chapter is less than the corresponding target number of words, based on the professional knowledge, the real-time information knowledge, and the generated content of the chapter and the chapter title corresponding to the chapter, using the generative big model to expand the content of the chapter until the number of words in the content of the chapter is equal to or greater than the corresponding target number of words.

14. The device according to any one of claims 12 to 13, wherein: The content generation unit is further used for: Based on the content of the report body, using the generative macro model, generate at least one of the summary, statement, appendix and reference of the research report; The integration module is used to integrate and generate the research report based on at least one of the abstract, statement, appendix and reference of the research report, as well as the title of the research report, the objective outline and the main content of the research report.

15. An electronic device, comprising: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 7.

16. A non-transitory computer-readable storage medium storing computer instructions, wherein: The computer instructions are used to cause the computer to execute the method according to any one of claims 1-7.

17. A computer program product, comprising a computer program, which, when executed by a processor, implements the method according to any one of claims 1 to 7.

Images