Simmary generation method, device and equipment, program product and storage medium

Abstract

The embodiment of the invention discloses a summary generation method, device and equipment, a program product and a storage medium, relates to the field of big data and can also be suitable for the field of financial science and technology, and the method comprises the steps: obtaining record data of a road show; wherein the recorded data at least comprises an audio file; performing cue word generation on the recorded data based on a preset cue word template and a semantic analysis algorithm to obtain key information of a predetermined summary generation model; inputting the key information and the recorded data into a summary generation model to obtain a road presentation summary output by the summary generation model; and obtaining target display information according to the road presentation summary, and displaying the target display information to the user. According to the method, the road presentation summary content which is clear in logic and coherent and meets the financial industry standard can be automatically obtained, and the quality and the generation efficiency of the road presentation summary are remarkably improved.

Description

Technical Field

[0001] The embodiments of the present invention relate to the field of big data and may also be applicable to the field of financial technology, and in particular, to a minutes generation method, device, equipment, program product and storage medium. Background Art

[0002] In the financial field, roadshows are an important part of information dissemination and communication in the financial field. Roadshows are usually conducted in the form of meetings, and the content covers financial information analysis, financial product introductions, research conclusions, and investment viewpoint analysis. Roadshows are an important part of information dissemination and communication in the financial field. Accurate and efficient roadshow minutes can help users quickly understand the core content of the meeting, extract key information, and provide support for subsequent decision-making. High-quality roadshow minutes can not only save a lot of browsing costs, but also improve work efficiency and ensure that team members have a clear and consistent understanding of the meeting content. Therefore, generating accurate and professional roadshow minutes is crucial for investment research in the financial industry.

[0003] The current method of generating roadshow minutes mostly uses speech recognition technology to convert meeting recordings into text, which is then manually edited and organized. Although speech recognition technology can improve the efficiency of organization, the generated text is often unstructured and inaccurate, making it difficult to directly use it to generate high-quality roadshow minutes. Summary of the invention

[0004] The embodiments of the present invention provide a minutes generation method, apparatus, device, program product and storage medium, which can automatically generate high-quality and structured roadshow minutes based on the recorded data of roadshows in the financial field, improve the efficiency of roadshow minutes generation, and improve user experience.

[0005] In a first aspect, an embodiment of the present invention provides a method for generating a minutes, comprising:

[0006] Acquire recorded data of the roadshow; wherein the recorded data at least includes an audio file;

[0007] Generate prompt words for the record data based on a preset prompt word template and a semantic analysis algorithm to obtain key information of a predetermined minutes generation model;

[0008] Input the key information and the recorded data into the minutes generation model to obtain the roadshow minutes output by the minutes generation model; wherein the minutes generation model is a vertical large model for processing financial field data;

[0009] Target display information is obtained according to the roadshow minutes, and the target display information is displayed to the user.

[0010] In a second aspect, an embodiment of the present invention provides a minutes generation device, the device comprising:

[0011] A data acquisition module, used to acquire recorded data of the road show; wherein the recorded data at least includes an audio file;

[0012] A first generation module, configured to generate prompt words for the record data based on a preset prompt word template and a semantic analysis algorithm, and obtain key information of a predetermined minutes generation model;

[0013] A second generation module is used to input the key information and the recorded data into the minutes generation model to obtain the roadshow minutes output by the minutes generation model; wherein the minutes generation model is a vertical large model for processing financial field data;

[0014] The data display module is used to obtain target display information according to the roadshow minutes and display the target display information to the user.

[0015] In a third aspect, an embodiment of the present invention further provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, a minutes generation method as described in any one of the embodiments of the present invention is implemented.

[0016] In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements a minutes generation method as described in any one of the embodiments of the present invention.

[0017] In a fifth aspect, an embodiment of the present invention provides a computer program product, including a computer program, which, when executed by a processor, implements a minutes generation method as described in any one of the embodiments of the present invention.

[0018] In an embodiment of the present invention, recorded data of a roadshow is obtained; wherein the recorded data includes at least an audio file; prompt words are generated for the recorded data based on a pre-set prompt word template and a semantic analysis algorithm to obtain key information of a predetermined minutes generation model; the key information and recorded data are input into the minutes generation model to obtain a roadshow minutes output by the minutes generation model; wherein the minutes generation model is a vertical large model for processing financial field data; target display information is obtained according to the roadshow minutes, and the target display information is displayed to the user. The method of the embodiment of the present invention automatically extracts key information from the recorded data and generates key information of the minutes generation model through a prompt word template and a semantic analysis algorithm, laying the foundation for outputting high-quality roadshow minutes through the minutes generation model. According to the key information and the minutes generation model, the content of the roadshow minutes that is logically clear, coherent and meets the standards of the financial industry can be automatically obtained, which significantly improves the quality of the roadshow minutes and the efficiency of generation. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0020] Figure 1 A first flow chart of a method for generating minutes provided by an embodiment of the present invention;

[0021] Figure 2 A second flow chart of a method for generating minutes provided by an embodiment of the present invention;

[0022] Figure 3 A schematic diagram of the structure of a minutes generating device provided by an embodiment of the present invention;

[0023] Figure 4 A schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0024] The present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It is to be understood that the specific embodiments described herein are only used to explain the present invention, rather than to limit the present invention. It should also be noted that, for ease of description, only parts related to the present invention, rather than all structures, are shown in the accompanying drawings.

[0025] Figure 1This is the first flow chart of a minutes generation method provided in an embodiment of the present invention. The method of the embodiment of the present invention can automatically generate high-quality and structured roadshow minutes based on the recorded data of roadshows in the financial field, thereby improving the efficiency and quality of roadshow minutes generation and improving user experience. The information collected in the method of the embodiment of the present invention 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 relevant data comply with the relevant laws, regulations and standards of relevant countries and 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 refuse. The method can be executed by a minutes generation device provided in an embodiment of the present invention, and the device can be implemented in software and / or hardware. The following embodiments will be described by taking the device integrated in an electronic device as an example. The electronic device can be a server or a computer device, etc., with reference to Figure 1 , the method may specifically include the following steps:

[0026] Step 101: Obtain recorded data of the roadshow.

[0027] Among them, the recorded data at least includes an audio file, and the audio file includes voice audio recording the roadshow process. The recorded data may also include auxiliary files, which are related to the roadshow and are used to help the server generate roadshow minutes, such as slides, Excel tables or other forms of files related to the roadshow. In an optional embodiment, when the user needs to use the server to automatically generate roadshow minutes, all recorded data of the roadshow can be uploaded to the server, and the server can receive the recorded data uploaded by the user. In an optional embodiment, the server can detect in real time whether there is new recorded data through a pre-set data acquisition interface. When the server finds that there is recorded data for which the roadshow minutes have not been generated, the server can automatically obtain the recorded data of the roadshow so as to generate the roadshow minutes based on the recorded data.

[0028] Step 102: Generate prompt words for the record data based on a preset prompt word template and a semantic analysis algorithm to obtain key information of a predetermined minutes generation model.

[0029] Among them, the prompt word template is a pre-designed text framework for generating prompt words input into the minutes generation model. The key information of the minutes generation model is generated based on the prompt words, which is used to guide the minutes generation model to accurately generate roadshow minutes. The prompt words can include information such as the core theme of the roadshow, key terms, and specific task requirements. The semantic analysis algorithm is used to understand the semantic structure of the roadshow content, identify core themes, professional terms, logical relationships, and important data. The minutes generation model in this solution is a vertical large model for processing financial field data after fine-tuning and training the general large model. The large model is a large-scale language model with extensive language understanding and generation capabilities. The vertical model is a language model optimized and customized for a specific field.

[0030] Specifically, after obtaining the recorded data, the server can extract the voice of the audio file in the recorded data and convert the voice into text data. Analyze the auxiliary files in the recorded data, and extract the key information in the auxiliary files through a pre-set semantic analysis algorithm, such as the title, charts and related numerical data in the slides. After analyzing the key information of the recorded data of the roadshow, fill the key information of the recorded data into the prompt word template to generate specific prompt words. After obtaining the prompt words, the roadshow recorded data can be subjected to logical structure analysis, and the prompt words can be further optimized according to the analysis results to obtain the key information of the minutes generation model. Exemplarily, the prompt words initially generated are: "Extract key words and core concepts from the meeting." Perform a logical structure analysis on the recorded data of the roadshow, including analyzing the macroeconomics, industry trends and risk assessment of the recorded data of the roadshow, and generate the key information for the minutes generation according to the analysis results: "Extract key words and core concepts from the meeting, focusing on terms such as price-earnings ratio, price-to-book ratio, macroeconomics, technology industry and new energy industry.

[0031] In an optional implementation, the recorded data of the roadshow also includes at least one auxiliary file. After obtaining the recorded data, the audio file is converted into text and data preprocessed to obtain a text file of the audio file; the analysis points of each auxiliary file are determined according to the type of each auxiliary file and the analysis points corresponding to each file type that are preset; for each auxiliary file, the key information of the current auxiliary file is analyzed according to the analysis points of the current auxiliary file to obtain the first information of the current auxiliary file, and the first information of the roadshow minutes is determined according to the first information of each auxiliary file. Multi-dimensional semantic analysis is performed on the text file according to the first information, the prompt word template and the semantic analysis algorithm to obtain key information of at least one dimension of the minutes generation model.

[0032] Step 103: input the key information and recorded data into the minutes generation model to obtain the roadshow minutes output by the minutes generation model.

[0033] Among them, the minutes generation model is a vertical large model for processing financial field data. In an optional implementation, before obtaining the recorded data of the roadshow, the server can obtain financial field data through various pre-set interfaces; when the pre-determined general large model does not meet the preset conditions, the financial field data is analyzed for entity relationships, and a financial knowledge base is established based on the analysis results; a financial knowledge embedding layer is established based on the financial knowledge base and the general large model, and an embedding vector of the general large model is generated according to the financial indication embedding layer; the embedding vector is input into the general large model, and the model parameters of the general large model are adjusted according to the output information of the general large model to obtain the minutes generation model.

[0034] Key information is extracted from the recorded data through a semantic analysis algorithm, and is used to guide the minutes generation model to generate prompt information of specific content. Specifically, after obtaining the key information, the text data of the audio file and the key information in the auxiliary file are sorted and integrated to obtain the integrated text corresponding to the text data and the auxiliary file. The integrated text and the key information of the minutes generation model are input into the minutes generation model in a structured form. After obtaining the integrated text and the key information, the minutes generation model can parse the semantics of the integrated text and the key information. Through the semantic matching technology, the key information is matched with the content in the integrated text to obtain the minutes content related to the key information, and the roadshow minutes are generated based on the content. Exemplarily, the key information includes "extracting key words and core concepts in the meeting", and the minutes generation model will generate a keyword summary containing terms such as price-earnings ratio and price-to-book ratio. The key information also includes "generating a full text summary of the meeting", and the minutes generation model will also generate a concise summary covering macroeconomic analysis, industry trends, individual stock recommendations and risk assessment. By inputting the optimized key information and record data into the minutes generation model, the minutes generation model can generate high-quality, professional roadshow minutes that meet user needs, meeting the financial industry's strict requirements for professionalism and accuracy.

[0035] Step 104: Obtain target display information according to the roadshow minutes, and display the target display information to the user.

[0036] Among them, the target display information is what is finally displayed to the user, including information about the content of the roadshow minutes. Specifically, after obtaining the roadshow minutes, the server can directly extract the target display information based on the specific needs of the user obtained in advance or the preset display rules. Exemplarily, if the user needs a keyword summary, the core terms and concepts in the minutes are extracted. If the user needs a full-text summary, the server extracts the main discussion points and conclusions in the minutes. If the user needs to extract the Q&A content in the meeting, the key questions and answers in the Q&A session are extracted. After obtaining the target display information, the server can display the target display information to the user according to the format requirements of the user obtained in advance.

[0037] In an optional implementation, after obtaining the roadshow minutes, obtaining target display information according to the roadshow minutes, and displaying the target display information to the user may include: determining a display template corresponding to the roadshow according to the roadshow content of the roadshow, and determining candidate display information according to the roadshow minutes and the display template; displaying the candidate display information to the user, and if modification information sent by the user is received within a preset time interval, modifying the candidate display information according to the modification information to obtain the target display information.

[0038] Among them, the display template is a pre-set template for generating target display information. Different roadshows can correspond to different display templates. According to the theme and content type of the roadshow, the display template corresponding to the current roadshow can be determined from various display templates. After determining the display template of the roadshow, extract the content corresponding to the display template from the roadshow minutes. Process the extracted content according to the format of the display template to generate candidate display information. After obtaining the candidate display information, the candidate display information is displayed to the user through the display interface. If the user needs to modify the candidate display information, the modification information can be sent to the server through the display interface within a preset time interval. The server can receive the modification information sent by the user within the preset time interval, and adjust the candidate display information according to the user's modification information to generate the final target display information. If the server does not receive the user's modification information within the preset time interval, the candidate display information is directly determined as the target display information.

[0039] By determining the appropriate display template based on the roadshow content, the server can generate display information that meets user needs and habits, thereby improving user experience. In addition, users can also adjust the display content according to their own needs, further improving user satisfaction.

[0040] The technical solution of this embodiment obtains the recorded data of the roadshow; wherein the recorded data includes at least an audio file; based on a pre-set prompt word template and a semantic analysis algorithm, prompt words are generated for the recorded data to obtain key information of a predetermined minutes generation model; the key information and the recorded data are input into the minutes generation model to obtain the roadshow minutes output by the minutes generation model; wherein the minutes generation model is a vertical large model for processing financial field data; target display information is obtained according to the roadshow minutes, and the target display information is displayed to the user. The technical solution of this embodiment automatically extracts key information from the recorded data and generates key information of the minutes generation model through a prompt word template and a semantic analysis algorithm, laying the foundation for outputting high-quality roadshow minutes through the minutes generation model. According to the key information and the minutes generation model, the roadshow minutes content that is logically clear, coherent and meets the standards of the financial industry can be automatically obtained, which significantly improves the quality of the roadshow minutes and the efficiency of generation.

[0041] Figure 2 The second flow chart of a method for generating minutes provided in an embodiment of the present invention is a refinement of the above embodiment. The specific method can be as follows Figure 2 As shown, the method may include the following steps:

[0042] Step 201: obtain financial field data through various pre-set interfaces; when the pre-determined general large model does not meet the preset conditions, optimize the general large model according to the financial field data to obtain a minutes generation model that meets the preset conditions.

[0043] Among them, each interface is used to connect the server and various financial data sources, including financial news websites, stock exchanges, financial research institutions and internal corporate databases. The server can obtain various types of financial data in real time through various interfaces, such as text data (news articles and research reports), structured data (financial statements and market data) and multimedia data. The preset conditions are predetermined based on domain big data, etc., and are used to measure whether the general big model can accurately generate roadshow minutes in the financial field. When the general big model does not meet the preset conditions, the general big model can be optimized by financial field data to obtain a minutes generation model that meets the preset conditions. In this scheme, optionally, the financial field data optimizes the general big model to obtain a minutes generation model that meets the preset conditions. It includes the following steps A1-step A2:

[0044] Step A1: Conduct entity relationship analysis on financial data and establish a financial knowledge base based on the analysis results.

[0045] Among them, an entity is an identifiable object or concept with specific meaning in the text. In the financial field, entities can include company names, stock codes, financial indicators, and industry names. The financial knowledge base is used to store financial field data according to the results of entity relationship analysis. Specifically, after obtaining the financial field data, the financial field data is cleaned to remove noise and irrelevant content, and the cleaned financial field data is converted into a unified format to facilitate entity relationship analysis. Furthermore, the server can use named entity recognition technology to identify key entities from financial field data. After identifying the key entities, the relationship between the entities is identified through a pre-set relationship extraction model to obtain the analysis results of the entity relationship of the financial field data. After obtaining the analysis results, a financial knowledge base containing nodes (entities) and edges (relationships) can be constructed based on the entities and relationships.

[0046] Step A2: Establish a financial knowledge embedding layer based on the financial knowledge base and the general large model, and generate an embedding vector of the general large model based on the financial instruction embedding layer. Input the embedding vector into the general large model, and adjust the model parameters of the general large model according to the output information of the general large model to obtain a minutes generation model.

[0047] After obtaining the financial knowledge base, the graph neural network can be used to perform embedding learning on the financial knowledge base. The graph neural network generates entity embedding vectors for each node (entity) and relationship embedding vectors for the relationships between nodes by aggregating the information of neighboring nodes. After obtaining the entity embedding vector and the relationship embedding vector, a financial knowledge embedding layer is added to the encoder part of the general large model. The embedding vector of the financial knowledge embedding layer is fused with the input of the general large model, and the fused embedding vector is input into the general large model, so that the general large model can use financial knowledge to generate high-quality text. After the embedding vector is input into the general large model, the output information of the general large model is obtained, and the model parameters of the general large model are adjusted using the real financial data corresponding to the predetermined output information. At the same time, the general large model is trained to complete multiple financial-related tasks, such as term recognition, text classification, and question-answer generation, to obtain a vertical large model for generating financial roadshow minutes.

[0048] The knowledge in the financial knowledge base is embedded into the general large model, and the model's ability to understand financial text is enhanced by generating an embedding vector and fusing it with the model input. By fine-tuning the large model and performing multi-task learning, the performance of the large model in the financial field can be further optimized, and a minutes generation model that can accurately generate roadshow minutes can be obtained.

[0049] Step 202: Obtain recorded data of the roadshow.

[0050] The recorded data at least includes an audio file.

[0051] Step 203: Perform text conversion and data preprocessing on the audio file to obtain a text file of the audio file.

[0052] The recorded data also includes at least one auxiliary file. Specifically, after obtaining the audio file, a predetermined speech recognition tool can be used to convert the audio file into a text format. After obtaining the text corresponding to the audio file, the text is preprocessed, such as data cleaning, removing noise and irrelevant content in the text, such as deleting special symbols, extra spaces, repeated sentences or paragraphs, background noise and informal conversations that are irrelevant to the roadshow theme, etc. The text after data preprocessing is determined as a text file of the audio file.

[0053] Step 204: Analyze key information of each auxiliary file according to the type of each auxiliary file to obtain the first information of the roadshow minutes.

[0054] Among them, the types of auxiliary files include image types, table types, slide types, and document types, etc. Different types of auxiliary files correspond to different key information analysis methods. According to the type of each auxiliary file, the key information analysis method corresponding to each auxiliary file can be determined, and the corresponding auxiliary file can be analyzed according to the key information analysis method to obtain the first information of the roadshow minutes. The first information of the roadshow minutes is used to help the server generate key information of the minutes generation model. In this scheme, optionally, key information analysis is performed on each auxiliary file according to the type of each auxiliary file to obtain the first information of the roadshow minutes, including: determining the analysis points of each auxiliary file according to the type of each auxiliary file and the analysis points corresponding to each file type set in advance; for each auxiliary file, performing key information analysis on the current auxiliary file according to the analysis points of the current auxiliary file to obtain the first information of the current auxiliary file, and determining the first information of the roadshow minutes according to the first information of each auxiliary file.

[0055] Specifically, different types of auxiliary files correspond to different analysis points. For example, the analysis points of image types (charts in slides) can be to extract information such as data, titles, labels, and trend lines in charts. The analysis points of table types can be to extract data, titles, cell contents, and formulas in tables. The analysis points of document types can be to extract the title, text content, tables, and charts of documents. After determining the analysis points of the auxiliary files, key information analysis can be performed on the auxiliary files based on the analysis points of the auxiliary files to obtain the first information of the auxiliary files. For example, use a text recognition tool to extract text content in an image. Use image recognition technology to identify data points and trend lines in charts. After obtaining the first information of each auxiliary file, integrate the first information of all auxiliary files to obtain the first information of the roadshow minutes.

[0056] Step 205: Perform multi-dimensional semantic analysis on the text file according to the first information, the prompt word template and the semantic analysis algorithm to obtain key information of at least one dimension of the minutes generation model.

[0057] Among them, each dimension is predetermined based on domain big data, etc., to help the server generate more comprehensive key information. The dimensions in this solution may include subject dimension, terminology dimension, data dimension, sentiment dimension, and logical structure dimension, etc. The prompt word template is pre-designed and is used to generate a text framework for prompt words input into the minutes generation model. After obtaining the first information, the first information is subjected to semantic analysis of each dimension according to the semantic analysis algorithm. For example, based on the semantic analysis results of the core theme of the meeting and the slide title, the key information of the theme dimension is determined to be "extracting discussions on the topic in the meeting." After obtaining the key information of all dimensions, the key information of the minutes generation model is determined based on the key information of all dimensions, so that the minutes generation model can generate multi-dimensional roadshow minutes based on the key information.

[0058] Step 206: input the key information and recorded data into the minutes generation model to obtain the roadshow minutes output by the minutes generation model.

[0059] Among them, the minutes generation model is a large vertical model used to process data in the financial field.

[0060] Step 207: Obtain target display information according to the roadshow minutes, and display the target display information to the user.

[0061] In the technical solution of this embodiment, financial field data is obtained through various pre-set interfaces; when the pre-determined general large model does not meet the preset conditions, the general large model is optimized according to the financial field data to obtain a minutes generation model that meets the preset conditions. The recorded data of the roadshow is obtained. Among them, the recorded data at least includes an audio file. Among them, the recorded data also includes at least one auxiliary file. The audio file is converted into text and data preprocessed to obtain a text file of the audio file. According to the type of each auxiliary file, the key information of each auxiliary file is analyzed to obtain the first information of the roadshow minutes. According to the first information, the prompt word template and the semantic analysis algorithm, the text file is multi-dimensionally semantically analyzed to obtain the key information of at least one dimension of the minutes generation model. The key information and the recorded data are input into the minutes generation model to obtain the roadshow minutes output by the minutes generation model; wherein the minutes generation model is a vertical large model for processing financial field data. According to the roadshow minutes, the target display information is obtained, and the target display information is displayed to the user. The technical solution of this embodiment obtains data from multiple financial data sources, which can ensure the diversity and comprehensiveness of the data and lay the foundation for subsequent model optimization. The big model is optimized based on the data in the financial field, so that the big model can better adapt to the specific needs of the financial field and improve the performance of the big model in financial text processing tasks. The key information of each auxiliary file is analyzed according to the type of each auxiliary file, and the core content of each file can be accurately extracted. Through the semantic analysis algorithm, the text file is analyzed from multiple dimensions, and the key information in the text can be accurately extracted to provide high-quality input for the minutes generation model, ensuring that the generated roadshow minutes have the characteristics of highlighting key points and clear logic. According to the key information and minutes generation model, the roadshow minutes with clear logic, coherence and in line with the standards of the financial industry can be automatically obtained, which significantly improves the quality of the roadshow minutes and the efficiency of generation.

[0062] Figure 3 The structure diagram of a minutes generating device provided in an embodiment of the present invention is suitable for executing the minutes generating method provided in an embodiment of the present invention. Figure 3 As shown, the device may specifically include:

[0063] The data acquisition module 301 is used to acquire the recorded data of the road show; wherein the recorded data at least includes an audio file;

[0064] The first generation module 302 is used to generate prompt words for the record data based on a preset prompt word template and a semantic analysis algorithm to obtain key information of a predetermined minutes generation model;

[0065] The second generation module 303 is used to input the key information and the record data into the minutes generation model to obtain the roadshow minutes output by the minutes generation model; wherein the minutes generation model is a vertical large model for processing financial field data;

[0066] The data display module 304 is used to obtain target display information according to the roadshow minutes and display the target display information to the user.

[0067] Optionally, the recorded data further includes at least one auxiliary file, and the first generating module 302 is specifically used to: perform text conversion and data preprocessing on the audio file to obtain a text file of the audio file;

[0068] Perform key information analysis on each auxiliary file according to the type of each auxiliary file to obtain the first information of the roadshow minutes;

[0069] A multi-dimensional semantic analysis is performed on the text file according to the first information, the prompt word template and the semantic analysis algorithm to obtain key information of at least one dimension of the minutes generation model.

[0070] Optionally, the first generating module 302 is further used to: determine the analysis key points of each auxiliary file according to the type of each auxiliary file and the pre-set analysis key points corresponding to each file type;

[0071] For each auxiliary file, key information analysis is performed on the current auxiliary file according to the analysis points of the current auxiliary file to obtain the first information of the current auxiliary file, and the first information of the roadshow minutes is determined according to the first information of each auxiliary file.

[0072] When the roadshow is a roadshow in the financial field, before obtaining the recorded data of the roadshow, the second generating module 303 is specifically used to: obtain financial field data through various pre-set interfaces;

[0073] When the predetermined general large model does not meet the preset conditions, the general large model is optimized according to the financial field data to obtain the minutes generation model that meets the preset conditions.

[0074] Optionally, the second generating module 303 is further used to: perform entity relationship analysis on the financial field data, and establish a financial knowledge base according to the analysis results;

[0075] Establishing a financial knowledge embedding layer based on the financial knowledge base and the general large model, and generating an embedding vector of the general large model according to the financial indication embedding layer;

[0076] The embedding vector is input into the general large model, and the model parameters of the general large model are adjusted according to the output information of the general large model to obtain the minutes generation model.

[0077] Optionally, the second generating module 303 is further used to: perform entity relationship analysis on the financial field data, and establish a financial knowledge base according to the analysis results;

[0078] Establishing a financial knowledge embedding layer based on the financial knowledge base and the general large model, and generating an embedding vector of the general large model according to the financial indication embedding layer;

[0079] The embedding vector is input into the general large model, and the model parameters of the general large model are adjusted according to the output information of the general large model to obtain the minutes generation model.

[0080] Optionally, the data display module 304 is specifically configured to: determine a display template corresponding to the roadshow according to the roadshow content of the roadshow, and determine candidate display information according to the roadshow minutes and the display template;

[0081] The candidate display information is displayed to the user, and if modification information sent by the user is received within a preset time interval, the candidate display information is modified according to the modification information to obtain the target display information.

[0082] The minutes generating device provided in the embodiment of the present invention can execute the minutes generating method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method. The contents not described in detail in this embodiment can refer to the description in any method embodiment of the present invention.

[0083] An embodiment of the present invention also provides a computer program product.

[0084] Various embodiments 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), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include: being implemented in one or more computer program products, which can include one or more computer programs, which 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, which 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.

[0085] Figure 4 A schematic diagram of the structure of an electronic device provided by an embodiment of the present invention, referring to Figure 4 , Figure 4 The electronic device 12 shown is only an example and should not limit the functions and scope of use of the embodiments of the present application. Figure 4 As shown, the electronic device 12 is in the form of a general purpose computing device. The components of the electronic device 12 may include, but are not limited to: one or more processors or processing units 16, a system memory 28, and a bus 18 that connects various system components (including the system memory 28 and the processing unit 16).

[0086] Bus 18 represents one or more of several types of bus structures, including a memory bus or memory controller, a peripheral bus, an accelerated graphics port, a processor or a local bus using any of a variety of bus architectures. By way of example, these architectures include, but are not limited to, an Industry Standard Architecture (ISA) bus, a Micro Channel Architecture (MAC) bus, an Enhanced ISA bus, a Video Electronics Standards Association (VESA) local bus, and a Peripheral Component Interconnect (PCI) bus.

[0087] The electronic device 12 typically includes a variety of computer system readable media. These media can be any available media that can be accessed by the electronic device 12, including volatile and non-volatile media, removable and non-removable media.

[0088] The system memory 28 may include computer system readable media in the form of volatile memory, such as random access memory (RAM) 30 and / or cache memory 32. The electronic device 12 may further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, the storage system 34 may be used to read and write non-removable, non-volatile magnetic media ( Figure 4 not shown, usually called a "hard drive"). Although Figure 4 Not shown, a disk drive for reading and writing to a removable non-volatile disk (e.g., a "floppy disk"), and an optical drive for reading and writing to a removable non-volatile optical disk (e.g., a CD-ROM, a DVD-ROM, or other optical medium) may be provided. In these cases, each drive may be connected to the bus 18 via one or more data medium interfaces. The memory 28 may include at least one program product having a set (e.g., at least one) of program modules that are configured to perform the functions of the various embodiments of the present application.

[0089] A program / utility 40 having a set (at least one) of program modules 46 may be stored, for example, in the memory 28, such program modules 46 including, but not limited to, an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment. The program modules 46 generally perform the functions and / or methods of the embodiments described herein.

[0090] The electronic device 12 may also communicate with one or more external devices 14 (e.g., keyboards, pointing devices, displays 24, etc.), may communicate with one or more devices that enable a user to interact with the electronic device 12, and / or may communicate with any device that enables the electronic device 12 to communicate with one or more other computing devices (e.g., a network card, a modem, etc.). Such communication may be performed via an input / output (I / O) interface 22. Furthermore, the electronic device 12 may also communicate with one or more networks (e.g., a local area network (LAN), a wide area network (WAN), and / or a public network, such as the Internet) via a network adapter 20. As shown, the network adapter 20 communicates with other modules of the electronic device 12 via the bus 18. It should be understood that although Figure 4 Not shown, other hardware and / or software modules may be used in conjunction with the electronic device 12, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0091] The processing unit 16 executes various functional applications and data processing by running the program stored in the system memory 28, for example, implementing a minutes generation method provided in an embodiment of the present invention: obtaining recorded data of a roadshow; wherein the recorded data includes at least an audio file; generating prompt words for the recorded data based on a preset prompt word template and a semantic analysis algorithm to obtain key information of a predetermined minutes generation model; inputting the key information and the recorded data into the minutes generation model to obtain a roadshow minutes output by the minutes generation model; wherein the minutes generation model is a vertical large model for processing financial field data; obtaining target display information according to the roadshow minutes, and displaying the target display information to the user.

[0092] The embodiment of the present invention provides a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, a method for generating minutes as provided in all the embodiments of the present invention is implemented: obtaining recorded data of a roadshow; wherein the recorded data includes at least an audio file; generating prompt words for the recorded data based on a preset prompt word template and a semantic analysis algorithm to obtain key information of a predetermined minutes generation model; inputting the key information and the recorded data into the minutes generation model to obtain a roadshow minutes output by the minutes generation model; wherein the minutes generation model is a vertical large model for processing financial field data; obtaining target display information according to the roadshow minutes, and displaying the target display information to a user. The computer-readable medium may be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium may be, for example, but not limited to, an electronic device, device or device of electrical, magnetic, optical, electromagnetic, infrared, or semiconductor, or any combination thereof. More specific examples of computer-readable storage media (a non-exhaustive list) include: an electrical connection with one or more conductors, 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 above. In this document, a computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction-executing electronic device, apparatus, or device.

[0093] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, which carry computer-readable program code. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer-readable signal media may also be any computer-readable medium other than a computer-readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction-executing electronic device, apparatus, or device.

[0094] The program code embodied on the computer readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

[0095] Computer program code for performing the operations of the present invention may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0096] Note that the above are only preferred embodiments of the present invention and the technical principles used. Those skilled in the art will understand that the present invention is not limited to the specific embodiments herein, and that various obvious changes, readjustments and substitutions can be made by those skilled in the art without departing from the scope of protection of the present invention. Therefore, although the present invention has been described in more detail through the above embodiments, the present invention is not limited to the above embodiments, and may include more other equivalent embodiments without departing from the concept of the present invention, and the scope of the present invention is determined by the scope of the appended claims.

Claims

1. A method for generating minutes, characterized in that: The method comprises: Acquire recorded data of the roadshow; wherein the recorded data at least includes an audio file; Generate prompt words for the record data based on a preset prompt word template and a semantic analysis algorithm to obtain key information of a predetermined minutes generation model; Input the key information and the recorded data into the minutes generation model to obtain the roadshow minutes output by the minutes generation model; wherein the minutes generation model is a vertical large model for processing financial field data; Target display information is obtained according to the roadshow minutes, and the target display information is displayed to the user.

2. The method according to claim 1, characterized in that The record data also includes at least one auxiliary file; based on a preset prompt word template and a semantic analysis algorithm, prompt words are generated for the record data to obtain key information of a predetermined minutes generation model, including: Performing text conversion and data preprocessing on the audio file to obtain a text file of the audio file; Perform key information analysis on each auxiliary file according to the type of each auxiliary file to obtain the first information of the roadshow minutes; A multi-dimensional semantic analysis is performed on the text file according to the first information, the prompt word template and the semantic analysis algorithm to obtain key information of at least one dimension of the minutes generation model.

3. The method according to claim 2, characterized in that Analyze the key information of each auxiliary file according to the type of each auxiliary file to obtain the first information of the roadshow minutes, including: Determining the analysis key points of each auxiliary file according to the type of each auxiliary file and the pre-set analysis key points corresponding to each file type; For each auxiliary file, key information analysis is performed on the current auxiliary file according to the analysis points of the current auxiliary file to obtain the first information of the current auxiliary file, and the first information of the roadshow minutes is determined according to the first information of each auxiliary file.

4. The method according to claim 1, characterized in that The roadshow is a roadshow in the financial field. Before obtaining the recorded data of the roadshow, the method further includes: Obtain financial data through various pre-set interfaces; When the predetermined general large model does not meet the preset conditions, the general large model is optimized according to the financial field data to obtain the minutes generation model that meets the preset conditions.

5. The method according to claim 4, characterized in that Optimizing the general large model according to the financial field data includes: Conduct entity relationship analysis on the financial field data, and establish a financial knowledge base based on the analysis results; Establishing a financial knowledge embedding layer based on the financial knowledge base and the general large model, and generating an embedding vector of the general large model according to the financial indication embedding layer; The embedding vector is input into the general large model, and the model parameters of the general large model are adjusted according to the output information of the general large model to obtain the minutes generation model.

6. The method according to claim 1, characterized in that Obtain target display information based on the roadshow minutes, including: Determine a display template corresponding to the roadshow according to the roadshow content, and determine candidate display information according to the roadshow minutes and the display template; The candidate display information is displayed to the user, and if modification information sent by the user is received within a preset time interval, the candidate display information is modified according to the modification information to obtain the target display information.

7. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the computer program implements a method for generating minutes according to any one of claims 1 to 6.

8. A minutes generating device, characterized in that: include: A data acquisition module, used to acquire recorded data of the road show; wherein the recorded data at least includes an audio file; A first generation module, configured to generate prompt words for the record data based on a preset prompt word template and a semantic analysis algorithm, and obtain key information of a predetermined minutes generation model; A second generation module is used to input the key information and the recorded data into the minutes generation model to obtain the roadshow minutes output by the minutes generation model; wherein the minutes generation model is a vertical large model for processing financial field data; The data display module is used to obtain target display information according to the roadshow minutes and display the target display information to the user.

9. An electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the minutes generation method according to any one of claims 1 to 6 is implemented.

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

Images