Intelligent financial research report generation method and device based on multiple models

Abstract

The present application discloses a method and device for generating intelligent financial research reports based on multiple models. The method includes: obtaining basic information of the target financial research report, the basic information at least including: the theme and multiple data types of the target financial research report; using multiple preset models to identify the theme of the target financial research report respectively, generating multiple financial research report outlines, inputting multiple test data sets of different data types into each preset model for testing, and comparing the test results corresponding to the training data set of each data type with the test results corresponding to the test data set of the same type of other preset models, determining the type of each preset model; and determining the target data type corresponding to the type of each preset model; using the preset models corresponding to the target data type in the multiple preset models to identify and analyze the theme and target data type data, and generating the target financial research report.

Description

Technical Field

[0001] The present application relates to the field of financial technology, and more specifically, to a method and device for generating intelligent financial research reports based on multiple models. Background Art

[0002] In the financial field, researchers can provide investment advice to clients by writing research reports. At present, domestic financial institutions need to write a large number of financial research reports, which requires a huge amount of work. According to the field, research reports can be divided into macro research reports, industry research reports, company research reports, futures research reports, fund research reports and bond research reports, etc.; according to the reporting cycle, they can be divided into daily reports, weekly reports, monthly reports, quarterly reports and annual reports, etc. Recently, with the rapid development of large-scale generative language models, they have been used in the intermediate process of research report writing, such as copywriting, data query, text expansion, etc. At present, the financial field generally directly generates financial research reports through large language models, but they often lack in-depth understanding and customized support for specific industries or institutions and specific corpus data, and only reflect the versatility of the large model itself. Although it reduces the development cost and time to a certain extent, it also sacrifices the accuracy and efficiency of the model in specific application scenarios, such as generating financial research reports. Summary of the invention

[0003] The embodiments of the present application provide a method and device for generating intelligent financial research reports based on multiple models, so as to at least solve the technical problem in related technologies that the generated financial research reports have low accuracy due to the lack of adaptation to financial field data.

[0004] According to one aspect of an embodiment of the present application, a method for generating intelligent financial research reports based on multiple models is provided, including: obtaining basic information of a target financial research report, the basic information including at least: the subject of the target financial research report and multiple data types, wherein the multiple data types include: news reports, market data, and policy documents; using multiple preset models to respectively identify the subject of the target financial research report, generating multiple financial research report outlines, for each of the multiple preset models, inputting multiple test data sets of different data types into each preset model for testing, and comparing the test results corresponding to the training data set of each data type with those of other preset models for the same type. The method comprises the following steps: comparing the test results corresponding to the test data sets of the type, and determining the type of each preset model according to the comparison results; determining the target data type corresponding to the type of each preset model from the multiple data types; respectively using the preset models corresponding to the target data type in the multiple preset models to identify and analyze the subject and target data type data to generate an initial financial research report outline; counting the initial financial research report outlines generated by the multiple preset models, and selecting a target outline from the multiple initial financial research report outlines; using the multiple preset models to generate multiple initial financial research reports according to the target outline; and fusing the multiple initial financial research reports to obtain a target financial research report.

[0005] Optionally, the preset models corresponding to the target data type from the multiple preset models are respectively used to identify and analyze the subject and target data type data to generate an initial financial research report outline, including: obtaining a preset corpus, the preset corpus at least including: historical financial research reports with the same format as the target financial research report and reference documents related to the subject of the target financial research report; using each of the multiple preset models to select data of the corresponding target data type from the preset corpus for analysis to obtain the initial financial research report outline.

[0006] Optionally, the method further includes: obtaining a preset corpus, the preset corpus including at least: historical financial research reports in the same format as the target financial research report and reference documents related to the subject of the target financial research report; using each preset model to receive selection instructions respectively, selecting data from the preset corpus for analysis according to the instructions of the selection instructions, and obtaining the initial financial research report outline.

[0007] Optionally, selecting a target outline from multiple initial financial research report outlines includes: comparing the formats of the multiple financial research report outlines with the target financial research report format one by one, and selecting the financial research report outline with the highest similarity as the initial financial research report outline; randomly selecting a preset model from the multiple preset models to be determined as the first target model, and receiving modification instructions in a question-and-answer manner through the first target model; extracting modification suggestions from the modification instructions using the first target model, and modifying the initial financial research report outline according to the modification suggestions to obtain the target outline.

[0008] Optionally, the multiple initial financial research reports are fused to obtain a target financial research report, including: randomly selecting an initial financial research report from the multiple initial financial research reports to be determined as the financial research report to be modified, and determining a target paragraph to be modified from the financial research report to be modified; selecting multiple paragraph texts at the same position as the target paragraph from the remaining multiple initial financial research reports; generating text to be input from the multiple paragraph texts according to a preset format; generating fused content based on the text to be input; replacing the target paragraph with the fused content to obtain a modified financial research report; and asking questions to the modified financial research report to further modify the modified financial research report to obtain the target financial research report.

[0009] Optionally, questions are asked to the modified financial research report to further modify the modified financial research report to obtain the target financial research report, including: receiving a selection instruction, selecting text to be modified from the modified financial research report; generating input text based on the text to be modified, the input text is at least used to indicate a corpus to be searched; selecting a document associated with the text to be modified from the corpus to be searched, and generating an answer to the input text based on the associated document; and replacing the text to be modified with the answer to the input text to obtain the target financial research report.

[0010] Optionally, the multiple preset models are used to generate multiple initial financial research reports according to the target outline, including: simulating multiple different roles through preset models, wherein the multiple different roles have different focuses; generating different questions based on the target outline through the multiple different roles, and retrieving different documents based on the different questions; generating answers to the different questions based on the different documents, and generating the initial financial research report based on the answers to the different questions.

[0011] According to another aspect of the embodiment of the present application, there is also provided an intelligent financial research report generation device based on multiple models, including: an acquisition module, used to obtain basic information of a target financial research report, wherein the basic information at least includes: the subject of the target financial research report and multiple data types, wherein the multiple data types include: news reports, market data and policy documents; a first generation module, used to use multiple preset models to respectively identify the subject of the target financial research report, generate multiple initial financial research report outlines, and for each of the multiple preset models, input multiple test data sets of different data types into each preset model for testing, and compare the test results corresponding to the training data set of each data type with those of other preset models for the same type. The test results corresponding to the test data set are compared, and the type of each preset model is determined according to the comparison results; the target data type corresponding to the type of each preset model is determined from the multiple data types; the preset models corresponding to the target data type in the multiple preset models are respectively used to identify and analyze the subject and target data type data to generate an initial financial research report outline; a selection module is used to count the initial financial research report outlines generated by the multiple preset models respectively, and select the target outline from the multiple initial financial research report outlines; a second generation module is used to use the multiple preset models to generate multiple initial financial research reports according to the target outline; a fusion module is used to fuse the multiple initial financial research reports to obtain a target financial research report.

[0012] According to another aspect of the embodiment of the present application, a computer device is also provided, including: a memory and a processor, wherein the memory is used to store program instructions; the processor is connected to the memory and is used to execute the above-mentioned multi-model-based intelligent financial research report generation method.

[0013] According to another aspect of the embodiment of the present application, a non-volatile storage medium is also provided, which includes a stored computer program, wherein the device where the non-volatile storage medium is located executes the above-mentioned multi-model-based intelligent financial research report generation method by running the computer program.

[0014] According to another aspect of the embodiment of the present application, a computer program product is also provided, including computer instructions, which, when executed by a processor, implement the above-mentioned multi-model-based intelligent financial research report generation method.

[0015] In an embodiment of the present application, basic information of a target financial research report is obtained, the basic information at least including: the subject of the target financial research report and multiple data types, wherein the multiple data types include: news reports, market data, and policy documents; multiple preset models are used to respectively identify the subject of the target financial research report, generate multiple financial research report outlines, and for each of the multiple preset models, multiple test data sets of different data types are input into each preset model for testing, and the test results corresponding to the training data set of each data type are compared with the test results corresponding to the test data sets of the same type by other preset models, and the type of each preset model is determined according to the comparison results; the target data corresponding to the type of each preset model is determined from the multiple data types. type; respectively use the preset models corresponding to the target data type in the multiple preset models to identify and analyze the subject and target data type data to generate an initial financial research report outline; count the initial financial research report outlines generated by the multiple preset models, and select a target outline from the multiple initial financial research report outlines; use the multiple preset models to generate multiple initial financial research reports according to the target outline; fuse the multiple initial financial research reports to obtain a target financial research report, thereby achieving the purpose of determining the data type adapted by the multiple preset models, and then selecting different preset models according to the processed data type, thereby achieving the technical effect of improving the adaptability of the model, and then solving the technical problem in the related technology that the generated financial research report has low accuracy due to the lack of adaptation to the financial field data. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0017] Figure 1 It is a hardware structure block diagram of a computer terminal for implementing a multi-model-based intelligent financial research report generation method according to an embodiment of the present application;

[0018] Figure 2 It is a flowchart of a method for generating intelligent financial research reports based on multiple models according to an embodiment of the present application;

[0019] Figure 3 is a schematic diagram of an interface for selecting a reference document according to an embodiment of the present application;

[0020] Figure 4 This is a schematic diagram of a modification interface of an initial financial research report outline according to an embodiment of the present application;

[0021] Figure 5This is a schematic diagram of a modification interface of a financial research report outline to be modified according to an embodiment of the present application;

[0022] Figure 6 is a flowchart of another method for generating intelligent financial research reports based on multiple models according to an embodiment of the present application;

[0023] Figure 7 It is a structural diagram of an intelligent financial research report generation device based on multiple models according to an embodiment of the present application. DETAILED DESCRIPTION

[0024] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present application.

[0025] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0026] The information collected in the embodiments of the present application is information and data authorized by the user or fully authorized by all parties, and the collection, storage, use, processing, transmission, provision, disclosure and application of the relevant data comply with the relevant laws, regulations and standards of the relevant regions, take necessary confidentiality measures, do not violate public order and good morals, and provide corresponding operation entrances for users to choose to authorize or reject automated decision-making results; if the user chooses to reject, the expert decision-making process will be entered.

[0027] When processing financial data and information, existing technologies often ignore the deep mining and application of specific fields, such as the rich internal corpus in the financial field. These internal corpora contain deep insights into financial business and direct feedback from market reactions, which limits their application effect in specific financial environments.

[0028] In order to solve the problems existing in the related technologies, the most suitable model is selected for different data types to identify and analyze, and finally the identification results of different models are integrated to obtain financial research reports. The embodiment of the present application provides an intelligent financial research report generation method based on multiple models, which can be run on Figure 1 In the computer terminal shown, the computer terminal is explained below.

[0029] The embodiment of the intelligent financial research report generation method based on multiple models provided in the embodiment of the present application can be executed in a mobile terminal, a computer terminal or a similar computing device. Figure 1 The hardware structure block diagram of a computer terminal for realizing a method for generating intelligent financial research reports based on multiple models is shown. Figure 1 As shown, the computer terminal 10 may include one or more (102a, 102b, ..., 102n are used to illustrate) processors (the processor may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA), a memory 104 for storing data, and a transmission module 106 for communication functions connected via a wired and / or wireless network. In addition, it may also include: a display, a keyboard, a cursor control device, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports of the I / O interface), a network interface, and a BUS bus. Those skilled in the art will understand that Figure 1 The structure shown is only for illustration and does not limit the structure of the above electronic device. Figure 1 More or fewer components as shown, or with Figure 1 Different configurations are shown.

[0030] It should be noted that the one or more processors and / or other data processing circuits described above may generally be referred to herein as "data processing circuits". The data processing circuits may be embodied in whole or in part as software, hardware, firmware, or any other combination thereof. In addition, the data processing circuit may be a single independent processing module, or may be incorporated in whole or in part into any of the other components in the computer terminal 10. As described in the embodiments of the present application, the data processing circuit acts as a processor control (e.g., selection of a variable resistor terminal path connected to an interface).

[0031] The memory 104 can be used to store software programs and modules of application software, such as the program instructions / data storage device corresponding to the intelligent financial research report generation method based on multiple models in the embodiment of the present application. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory 104, that is, realizing the above-mentioned intelligent financial research report generation method based on multiple models. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 104 may further include a memory remotely arranged relative to the processor, and these remote memories may be connected to the computer terminal 10 via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0032] The transmission module 106 is used to receive or send data via a network. The specific example of the above network may include a wireless network provided by a communication provider of the computer terminal 10. In one example, the transmission module 106 includes a network adapter (Network Interface Controller, NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission module 106 can be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0033] The display may be, for example, a touch screen liquid crystal display (LCD) that enables a user to interact with a user interface of the computer terminal 10 .

[0034] It should be noted that, in some optional embodiments, the above Figure 1 The computer terminal shown may include hardware elements (including circuits), software elements (including computer code stored on a computer-readable medium), or a combination of hardware elements and software elements. It should be noted that Figure 1 This is merely one example of a particular embodiment and is intended to illustrate the types of components that may be present in the computer terminal described above.

[0035] Under the above-mentioned operating environment, an embodiment of the present application provides an embodiment of an intelligent financial research report generation method based on multiple models. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0036] Figure 2is a flowchart of a method for generating intelligent financial research reports based on multiple models according to an embodiment of the present application, such as Figure 2 As shown, the method comprises the following steps:

[0037] Step S202, obtaining basic information of the target financial research report, the basic information at least including: the subject of the target financial research report and multiple data types, wherein the multiple data types include: news reports, market data and policy documents.

[0038] In the above step S202, the various data types include but are not limited to the above-mentioned news reports, market data and policy documents, and also include: online articles, professional books, research reports, and different application platforms may also include data within the application platform. The theme of the financial research report can be determined by the content of the financial research report to be written.

[0039] Step S204, using multiple preset models to respectively identify the subject of the target financial research report, generate multiple financial research report outlines, for each of the multiple preset models, input multiple test data sets of different data types into each preset model for testing, and compare the test results corresponding to the training data set of each data type with the test results corresponding to the test data sets of the same type of other preset models, and determine the type of each preset model according to the comparison result; determine the target data type corresponding to the type of each preset model from the multiple data types; and use the preset models corresponding to the target data type in the multiple preset models to identify and analyze the subject and target data type data to generate a financial research report outline.

[0040] In the above step S204, the above-mentioned multiple preset models can be obtained by adjusting the model parameters of the general model, for example, adjusting the training data set and hyperparameters of the model. The process of testing multiple preset models through test data sets of different data types can be to apply each model to its corresponding test data set and collect test results. The test results may include indicators such as accuracy, recall, F1 score, loss value, etc. The process of comparing the results includes: comparing the test results of each model with the test results of other models for the same type of test data set. This can be done by visualization means (such as bar charts, scatter plots) or statistical methods. Select the model that performs best on a specific data type according to the training time or prediction speed. In another optional way, performance analysis can also be performed, for example, analyzing the performance differences of different models, determining which models perform better on specific types of data, specifically, determining the criteria for evaluating model performance, such as accuracy, recall, F1 score, precision, mean square error (MSE), etc., and selecting evaluation indicators that match the task type, such as accuracy, the data type applied to the model to perform the task The data type is the target data type.

[0041] It should also be noted that multiple preset models can be selected from the initial model library.

[0042] Step S206, counting the initial financial research report outlines generated by multiple preset models respectively, and selecting a target outline from the multiple initial financial research report outlines.

[0043] In the above step S206, there are multiple ways to select the target outline. One optional way is to directly receive the selection instruction and specify the selection instruction; it can also be determined by evaluation, for example: the richness of the outline, such as: the number of words in the outline and the number of directories included in the outline, the format of the outline, etc.

[0044] After selecting the target outline, you can also adjust the order of the directories in the target outline, and you can also adjust the contents of the directories in the target outline.

[0045] Step S208, using multiple preset models to generate multiple initial financial research reports according to the target outline.

[0046] Step S210, merging multiple initial financial research reports to obtain a target financial research report.

[0047] In step S210, there are multiple ways to perform the fusion, for example, fusion of different paragraphs of multiple initial financial research reports, or fusion of the entirety of multiple initial financial research reports.

[0048] In some embodiments of the present application, the large model is carefully customized and trained in combination with the complexity and particularity of financial services. It covers all aspects of financial services, including credit, risk management, market analysis, etc., ensuring the model's deep understanding and accurate analysis of bank data. Through customized training, the model can better capture industry trends, understand customer needs, and predict market changes, thereby generating more accurate and valuable research reports. In addition, customized training also includes continuous optimization and updating of the model to adapt to the rapid changes in the financial market and the development needs of Beijing Bank's business.

[0049] Through the above steps S202 to S210, the purpose of determining the data types adapted by multiple preset models is achieved, and then different preset models are selected according to the processed data types, thereby achieving the technical effect of improving the adaptability of the model, and thus solving the technical problem of low accuracy of the generated financial research reports due to the lack of adaptation to financial field data in the related technology. The following is a detailed description.

[0050] In step S204 of the above-mentioned multi-model-based intelligent financial research report generation method, the preset models corresponding to the target data type in the multiple preset models are respectively used to identify and analyze the subject and target data type data to generate an initial financial research report outline, including: obtaining a preset corpus, the preset corpus at least including: historical financial research reports with the same format as the target financial research report and reference documents related to the subject of the target financial research report; using each of the multiple preset models to select data of the corresponding target data type from the preset corpus for analysis to obtain the initial financial research report outline.

[0051] It should be noted that the preset model can not only select data of the target data type corresponding to the preset model from the preset corpus, but also search for data of the same target data type from the Internet. Figure 3 A schematic diagram of an interface for selecting a reference document is shown, Figure 3 As shown, you can select one reference file or multiple reference files. It should also be noted that a new local folder can be created in the preset corpus, and the local folder can store local reference documents uploaded by the local device. The uploaded local reference documents can be determined according to the target financial research report theme. If the local reference document is in PDF format, it can be parsed to obtain the content in the reference document.

[0052] In another optional manner, each preset model can also randomly select data from a preset corpus. The specific implementation process is as follows: obtaining a preset corpus, the preset corpus at least including: historical financial research reports with the same format as the target financial research report and reference documents related to the subject of the target financial research report; using each preset model to receive a selection instruction, selecting data from the preset corpus for analysis according to the instructions of the selection instruction, and obtaining the initial financial research report outline.

[0053] In some embodiments of the present application, a target outline is selected from multiple financial research report outlines, including: comparing the formats of the multiple financial research report outlines with the target financial research report format one by one, and selecting the financial research report outline with the highest similarity as the initial financial research report outline; randomly selecting a preset model from the multiple preset models to be determined as the first target model, and receiving modification instructions in a question-and-answer manner through the first target model; extracting modification suggestions from the modification instructions using the first target model, and modifying the initial financial research report outline according to the modification suggestions to obtain the target outline.

[0054] Figure 4 A schematic diagram of the modification interface of an initial financial research report outline is shown, Figure 4As shown, a modification instruction is proposed: "Modify the second title" instructs the first target model to modify the second title in the initial financial outline.

[0055] It is understandable that the target financial research report format includes the writing format of financial research reports.

[0056] In an optional manner, the selection may also be made based on the content of the financial research report outline, for example: determined based on the subject relevance between the financial research report outline and the target financial research report. Specifically, the subject relevance between the financial research report outline and the target financial research report may be analyzed through multiple preset models, and the financial research report outline with the highest average relevance may be selected as the initial financial research report outline.

[0057] In step S210, the multiple initial financial research reports are fused to obtain a target financial research report, including: randomly selecting an initial financial research report from the multiple initial financial research reports to be determined as the financial research report to be modified, and determining the target paragraph to be modified from the financial research report to be modified; selecting multiple paragraph texts at the same position as the target paragraph from the remaining multiple initial financial research reports; generating text to be input from the multiple paragraph texts according to a preset format; generating fused content based on the text to be input; replacing the target paragraph with the fused content to obtain a modified financial research report; and asking questions to the modified financial research report to further modify the modified financial research report to obtain the target financial research report.

[0058] like Figure 5 As shown, the right side of the interface is the financial research report to be modified, and the right side of the interface is one of the other multiple initial financial research reports, which can be switched. The marked paragraph on the left side of the interface is the selected paragraph text.

[0059] The specific process of generating the text to be input from the plurality of paragraph texts according to the preset format is as follows:

[0060] Taking the paragraph text generated by preset model 1: text 1, the paragraph text generated by preset model 2: text 2, and the paragraph text generated by preset model 3: text 3 as examples, the text to be input is: "Based on text 1, text 2, and text 3, generate a new paragraph of content", and the model output is the fused content.

[0061] In some embodiments of the present application, questions are asked of the modified financial research report to further modify the modified financial research report to obtain the target financial research report, including: receiving a selection instruction, selecting text to be modified from the modified financial research report; generating input text based on the text to be modified, the input text being used at least to indicate a corpus to be searched; selecting a document associated with the text to be modified from the corpus to be searched, and generating an answer to the input text based on the associated document; and replacing the text to be modified with the answer to the input text to obtain the target financial research report.

[0062] In some embodiments of the present application, the specific steps of using the multiple preset models to generate multiple initial financial research reports according to the target outline are as follows: simulating multiple different roles through preset models, wherein the multiple different roles have different focuses; generating different questions based on the target outline through the multiple different roles, and retrieving different documents based on the different questions; generating answers to the different questions based on the different documents, and generating the initial financial research report based on the answers to the different questions.

[0063] It should be noted that, in actual application scenarios, each preset model simulates multiple different roles, and different preset models can also simulate different roles respectively.

[0064] Specifically, taking the theme of the target financial research report as the development of financial research report writing software in the financial industry as an example, the different roles include: Role 1: Basic Facts Writer: The basic facts writer focuses on broadly covering the basic facts about the topic. Role 2: Policy Maker: Focuses on policies and regulations related to the development of the financial industry, and their impact on the financial industry. Role 3: Industry Expert: Provides an overview of the application and impact of financial research report writing software in the financial industry, including successful case studies and industry trends. Role 4: Technical Expert: In-depth research on financial research report writing software technologies, including: machine learning, natural language processing, and computer vision, as well as their specific applications in the financial industry.

[0065] In actual application scenarios, through multiple questions asked by several roles, for example, Question 1 and Question 2, relevant document 1 and Document 2 are retrieved, and Document 1 and Document 2 are used as inputs of the preset model to answer, forming a complete dialogue.

[0066] Figure 6 Another flow chart of intelligent financial research report generation method based on multiple models is shown. Figure 6As shown, first input the title of the target financial research report, select the preset model, and obtain the content requirements of the target financial research report; select a corpus, generate multiple outlines based on the selected corpus, select a target outline from the multiple outlines, generate multiple financial research reports based on the target outline, merge the multiple financial research reports, modify and insert new text paragraphs, and finally obtain the target financial research report for download.

[0067] Figure 7 is a structural diagram of a multi-model-based intelligent financial research report generation device according to an embodiment of the present application, such as Figure 7 As shown, the device comprises:

[0068] An acquisition module 60 is used to acquire basic information of a target financial research report, wherein the basic information at least includes: a subject of the target financial research report and multiple data types, wherein the multiple data types include: news reports, market data, and policy documents;

[0069] The first generating module 62 is used to respectively identify the subject of the target financial research report using a plurality of preset models, generate a plurality of initial financial research report outlines, input a plurality of test data sets of different data types into each preset model for testing, and compare the test results corresponding to the training data set of each data type with the test results corresponding to the test data sets of the same type by other preset models, and determine the type of each preset model according to the comparison results; determine the target data type corresponding to the type of each preset model from the plurality of data types; respectively use the preset models corresponding to the target data type in the plurality of preset models to identify and analyze the subject and target data type data, and generate an initial financial research report outline;

[0070] A selection module 64, configured to count the financial research report outlines generated by the plurality of preset models respectively, and select a target outline from the plurality of financial research report outlines;

[0071] A second generating module 66, configured to generate a plurality of initial financial research reports according to the target outline using the plurality of preset models;

[0072] The fusion module 68 is used to fuse the multiple initial financial research reports to obtain a target financial research report.

[0073] Through the acquisition module 60, the first generation module 62, the selection module 64, the second generation module 66 and the fusion module 68 in the above-mentioned multi-model-based intelligent financial research report generation device, the purpose of determining the data types adapted by multiple preset models and then selecting different preset models according to the processed data types is achieved, thereby achieving the technical effect of improving the adaptability of the model, and further solving the technical problem of low accuracy of the generated financial research reports due to the lack of adaptation to the financial field data in the related technology.

[0074] In the first generation module 62 of the above-mentioned multi-model-based intelligent financial research report generation device, it includes: a first generation sub-module and a second generation sub-module, wherein the first generation sub-module is used to respectively use the preset models corresponding to the target data type in the multiple preset models to identify and analyze the subject and target data type data, and generate an initial financial research report outline, including: obtaining a preset corpus, the preset corpus at least including: historical financial research reports with the same format as the target financial research report and reference documents related to the target financial research report subject; respectively using each of the multiple preset models to select the corresponding target data type data from the preset corpus for analysis to obtain the financial research report outline.

[0075] The second generating submodule is used to obtain a preset corpus, which at least includes: historical financial research reports with the same format as the target financial research report and reference documents related to the subject of the target financial research report; each preset model is used to receive a selection instruction, and data is selected from the preset corpus for analysis according to the instructions of the selection instruction to obtain the initial financial research report outline.

[0076] The first generation module 62 includes: a selection unit for selecting a target outline from the multiple initial financial research report outlines, including: comparing the formats of the multiple financial research report outlines with the target financial research report format one by one, and selecting the financial research report outline with the highest similarity as the initial financial research report outline; randomly selecting a preset model from the multiple preset models to be determined as the first target model, and receiving modification instructions in a question-and-answer manner through the first target model; using the first target model to extract modification suggestions from the modification instructions, and modifying the initial financial research report outline according to the modification suggestions to obtain the target outline.

[0077] The fusion module 68 includes: a fusion submodule, which is used to fuse the multiple initial financial research reports to obtain a target financial research report, including: randomly selecting an initial financial research report from the multiple initial financial research reports to be determined as the financial research report to be modified, and determining the target paragraph to be modified from the financial research report to be modified; selecting multiple paragraph texts at the same position as the target paragraph from the remaining multiple initial financial research reports; generating text to be input from the multiple paragraph texts according to a preset format; generating fused content based on the text to be input; replacing the target paragraph with the fused content to obtain a modified financial research report; and asking questions to the modified financial research report to further modify the modified financial research report to obtain the target financial research report.

[0078] The fusion submodule includes: a modification unit, which is used to ask questions to the modified financial research report to further modify the modified financial research report to obtain the target financial research report, including: receiving a selection instruction, selecting the text to be modified from the modified financial research report; generating an input text according to the text to be modified, and the input text is at least used to indicate a corpus to be searched; selecting a document associated with the text to be modified from the corpus to be searched, and generating an answer to the input text based on the associated document; and replacing the text to be modified with the answer to the input text to obtain the target financial research report.

[0079] The second generation module 66 includes: a third generation submodule, which is used to generate multiple initial financial research reports according to the target outline using the multiple preset models, including: simulating multiple different roles through preset models, wherein the multiple different roles have different focuses; generating different questions based on the target outline through the multiple different roles, and retrieving different documents based on the different questions; generating answers to the different questions based on the different documents, and generating the initial financial research report based on the answers to the different questions.

[0080] It should be noted that Figure 6 The intelligent financial research report generation device based on multiple models is used to execute Figure 2 The multi-model-based intelligent financial research report generation method shown, therefore the relevant explanations in the above-mentioned multi-model-based intelligent financial research report generation method are also applicable to the multi-model-based intelligent financial research report generation device, and will not be repeated here.

[0081] An embodiment of the present application also provides a computer device, including: a memory and a processor, wherein the memory is used to store program instructions; the processor is connected to the memory and is used to execute the above-mentioned multi-model-based intelligent financial research report generation method.

[0082] An embodiment of the present application also provides a non-volatile storage medium, which includes a stored computer program, wherein the device where the non-volatile storage medium is located executes the above-mentioned multi-model-based intelligent financial research report generation method by running the computer program.

[0083] An embodiment of the present application also provides a computer program product, including a non-volatile computer-readable storage medium, wherein the non-volatile computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the multi-model-based intelligent financial research report generation method in the present application are implemented.

[0084] An embodiment of the present application also provides a computer program product, including computer instructions, which, when executed by a processor, implement the steps of the multi-model-based intelligent financial research report generation method in the present application.

[0085] An embodiment of the present application also provides a computer program, which, when executed by a processor, implements the steps of the multi-model-based intelligent financial research report generation method in the present application.

[0086] The serial numbers of the above-mentioned embodiments of the present application are for description only and do not represent the advantages or disadvantages of the embodiments.

[0087] In the above embodiments of the present application, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, please refer to the relevant description of other embodiments.

[0088] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of the units can be a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0089] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0090] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0091] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, server or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, disk or optical disk, etc., which can store program code.

[0092] The above is only a preferred implementation of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A method for generating intelligent financial research reports based on multiple models, characterized in that: include: Obtaining basic information of a target financial research report, wherein the basic information at least includes: a subject of the target financial research report and multiple data types, wherein the multiple data types include: news reports, market data, and policy documents; Using multiple preset models to respectively identify the subject of the target financial research report, generate multiple financial research report outlines, for each of the multiple preset models, input multiple test data sets of different data types into each preset model for testing, and compare the test results corresponding to the training data set of each data type with the test results corresponding to the test data set of the same type of other preset models, and determine the type of each preset model according to the comparison results; determine the target data type corresponding to the type of each preset model from the multiple data types; respectively use the preset models corresponding to the target data type in the multiple preset models to identify and analyze the subject and target data type data, and generate a financial research report outline; Counting the financial research report outlines generated by the plurality of preset models respectively, and selecting a target outline from the plurality of financial research report outlines; Using the plurality of preset models to generate a plurality of initial financial research reports according to the target outline; Performing fusion processing on the multiple initial financial research reports to obtain a target financial research report; Select a target outline from a number of financial research outlines, including: Comparing the formats of the multiple financial research report outlines with the target financial research report format one by one, and selecting the financial research report outline with the highest similarity as the initial financial research report outline; Randomly selecting a preset model from the plurality of preset models to determine as a first target model, and receiving a modification instruction in a question-and-answer manner through the first target model; extracting modification suggestions from the modification instructions using the first target model, and modifying the initial financial research report outline according to the modification suggestions to obtain the target outline; The multiple initial financial research reports are fused to obtain a target financial research report, including: Randomly selecting an initial financial research report from the multiple initial financial research reports to be determined as the financial research report to be modified, and determining a target paragraph to be modified from the financial research report to be modified; Select multiple paragraphs of text at the same position as the target paragraph from the remaining multiple initial financial research reports; Generate text to be input by using the multiple paragraph texts according to a preset format; generating fused content based on the text to be input; The target paragraph is replaced by the fused content to obtain a modified financial research report; Asking questions to the modified financial research report to further modify the modified financial research report to obtain the target financial research report; Using the plurality of preset models to generate a plurality of initial financial research reports according to the target outline includes: Simulating multiple different roles through a preset model, wherein the multiple different roles have different focuses; Generate different questions based on the target outline by the multiple different roles, and retrieve different documents based on the different questions; Answers to the different questions are generated based on the different documents, and the initial financial research report is generated based on the answers to the different questions.

2. The method according to claim 1, characterized in that Respectively using the preset models corresponding to the target data type from the plurality of preset models to identify and analyze the subject and target data type data, and generating a financial research report outline, including: Acquire a preset corpus, wherein the preset corpus at least includes: historical financial research reports in the same format as the target financial research report and reference documents related to the subject of the target financial research report; Each of the multiple preset models is used to select data of the corresponding target data type from the preset corpus for analysis to obtain the outline of the financial research report.

3. The method according to claim 1, characterized in that The method further comprises: Acquire a preset corpus, wherein the preset corpus at least includes: historical financial research reports in the same format as the target financial research report and reference documents related to the subject of the target financial research report; Each preset model is used to receive a selection instruction respectively, and data is selected from the preset corpus for analysis according to the instructions of the selection instruction to obtain the financial research report outline.

4. The method according to claim 1, characterized in that Asking questions to the modified financial research report to further modify the modified financial research report to obtain the target financial research report, including: Receive a selection instruction and select the text to be modified from the modified financial research report; Generating an input text according to the text to be modified, wherein the input text is at least used to indicate a corpus to be searched; Selecting a document associated with the text to be modified from the corpus to be searched, and generating an answer to the input text based on the associated document; The answer to the input text is used to replace the text that needs to be modified to obtain the target financial research report.

5. An intelligent financial research report generation device based on multiple models, characterized in that: include: An acquisition module, used to acquire basic information of a target financial research report, wherein the basic information at least includes: a subject of the target financial research report and multiple data types, wherein the multiple data types include: news reports, market data, and policy documents; A first generation module is used to use multiple preset models to respectively identify the subject of the target financial research report, generate multiple initial financial research report outlines, and for each of the multiple preset models, input multiple test data sets of different data types into each preset model for testing, and compare the test results corresponding to the training data set of each data type with the test results corresponding to the test data set of the same type of other preset models, and determine the type of each preset model according to the comparison results; determine the target data type corresponding to the type of each preset model from the multiple data types; and use the preset models corresponding to the target data type in the multiple preset models to identify and analyze the subject and target data type data, and generate an initial financial research report outline; A selection module, for counting the financial research report outlines generated by the plurality of preset models respectively, and selecting a target outline from the plurality of financial research report outlines; A second generating module, configured to generate a plurality of initial financial research reports according to the target outline by using the plurality of preset models; A fusion module, used for fusing the multiple initial financial research reports to obtain a target financial research report; Select a target outline from a number of financial research outlines, including: Comparing the formats of the multiple financial research report outlines with the target financial research report format one by one, and selecting the financial research report outline with the highest similarity as the initial financial research report outline; Randomly selecting a preset model from the plurality of preset models to determine as a first target model, and receiving a modification instruction in a question-and-answer manner through the first target model; extracting modification suggestions from the modification instructions using the first target model, and modifying the initial financial research report outline according to the modification suggestions to obtain the target outline; The multiple initial financial research reports are fused to obtain a target financial research report, including: Randomly selecting an initial financial research report from the multiple initial financial research reports to be determined as the financial research report to be modified, and determining a target paragraph to be modified from the financial research report to be modified; Select multiple paragraphs of text at the same position as the target paragraph from the remaining multiple initial financial research reports; Generate text to be input by using the multiple paragraph texts according to a preset format; generating fused content based on the text to be input; The target paragraph is replaced by the fused content to obtain a modified financial research report; Asking questions to the modified financial research report to further modify the modified financial research report to obtain the target financial research report; Using the plurality of preset models to generate a plurality of initial financial research reports according to the target outline includes: Simulating multiple different roles through a preset model, wherein the multiple different roles have different focuses; Generate different questions based on the target outline by the multiple different roles, and retrieve different documents based on the different questions; Answers to the different questions are generated based on the different documents, and the initial financial research report is generated based on the answers to the different questions.

6. A computer device, characterized in that: include: A memory and a processor, wherein the memory is used to store program instructions; The processor is connected to the memory and is used to execute the multi-model based intelligent financial research report generation method described in any one of claims 1 to 4.

7. A computer program product comprising computer instructions, characterized in that When the computer instructions are executed by the processor, the intelligent financial research report generation method based on multiple models as described in any one of claims 1 to 4 is implemented.

Images