Service key information identification method and device, equipment and medium

Abstract

The embodiment of the invention discloses a business key information identification method and device, equipment and a medium, and relates to the technical field of computers. The method comprises the following steps: acquiring a business information introduction text; in the business information introduction text, determining a target text associated with to-be-identified business key information; and determining an artificial intelligence model corresponding to the to-be-identified information type, and processing the target text based on the artificial intelligence model to obtain an identification result of the business key information. According to the technical scheme, the target text associated with the business key information to be recognized is determined in the business information introduction text, the range of key information recognition is shortened, the effect of reducing the recognition time is achieved, key information recognition is carried out through the artificial intelligence model corresponding to the type of the information to be recognized, and the recognition efficiency is improved. And the identification result is more accurate and reliable.

Description

Technical Field

[0001] This invention relates to the field of computer technology, and in particular to a method, apparatus, device, and medium for identifying business-critical information. Background Technology

[0002] In the bidding and tendering field, a large number of bidding requests are generated daily, and relevant personnel cannot possibly sift through and read through a large number of bidding documents with limited time. To address this issue, a scheme is proposed to extract entity triples from the bidding documents to obtain key information.

[0003] However, current technical solutions can only extract key information in a simple way, and the accuracy of the results obtained is relatively low. Summary of the Invention

[0004] This invention provides a method, apparatus, device, and medium for identifying key business information, which can accurately identify the content of key business information in business information.

[0005] According to one aspect of the present invention, a method for identifying business-critical information is provided, the method comprising:

[0006] Obtain business information introduction text;

[0007] In the business information description text, identify the target text associated with the key business information to be identified;

[0008] The artificial intelligence model corresponding to the type of information to be identified is determined, and the target text is processed based on the artificial intelligence model to obtain the identification result of key business information.

[0009] According to another aspect of the present invention, a device for identifying business-critical information is provided, comprising:

[0010] The business information introduction text acquisition module is used to acquire business information introduction text;

[0011] The target text determination module is used to determine the target text associated with the key business information to be identified in the business information introduction text;

[0012] The business key information identification result determination module is used to determine the artificial intelligence model corresponding to the type of information to be identified, and to process the target text based on the artificial intelligence model to obtain the identification result of the business key information.

[0013] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0014] At least one processor; and

[0015] A memory communicatively connected to the at least one processor; wherein,

[0016] The memory stores a computer program that can be executed by the at least one processor, which enables the at least one processor to perform the business-critical information identification method according to any embodiment of the present invention.

[0017] According to another aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium storing computer instructions, the computer instructions being configured to cause a processor to execute and implement the business-critical information identification method described in any embodiment of the present invention.

[0018] The technical solution of this application embodiment includes: acquiring business information introduction text; determining target text associated with key business information to be identified in the business information introduction text; determining an artificial intelligence model corresponding to the type of information to be identified; and processing the target text based on the artificial intelligence model to obtain the identification result of the key business information. This technical solution shortens the scope of key information identification by determining the target text associated with the key business information to be identified in the business information introduction text, thus reducing the identification time. Furthermore, this technical solution uses an artificial intelligence model corresponding to the type of information to be identified to perform key information identification, making the identification result more accurate and reliable.

[0019] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 This is a flowchart of a method for identifying key business information according to Embodiment 1 of this application;

[0022] Figure 2 This is a flowchart of a method for identifying key business information according to Embodiment 2 of this application;

[0023] Figure 3 This is a flowchart of a key information extraction process for bidding, provided according to Embodiment 2 of this application;

[0024] Figure 4 This is a schematic diagram of the structure of a business key information identification device according to Embodiment 3 of this application;

[0025] Figure 5 This is a schematic diagram of the structure of an electronic device that implements a method for identifying key business information according to an embodiment of this application. Detailed Implementation

[0026] To enable those skilled in the art to better understand the present invention, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0027] It should be noted that the terms "first," "second," "target," etc., used in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0028] Example 1

[0029] Figure 1 This application provides a flowchart of a method for identifying key business information in Embodiment 1. This embodiment is applicable to situations where key information is identified within business information. The method can be executed by a device for identifying key business information, which can be implemented in hardware and / or software and can be configured in an electronic device with data processing capabilities. Figure 1 As shown, the method includes:

[0030] S110, Obtain the business information introduction text.

[0031] Among them, the business information introduction text reflects relevant information about a certain business. For example, taking the bidding business as an example, the business information introduction text can be the bidding document, etc.

[0032] S120, in the business information introduction text, determine the target text associated with the key business information to be identified.

[0033] The key business information to be identified is the information that needs to be identified in this key information identification process. Since the business information description text may contain comprehensive information about the business, including but not limited to: business background, business start time, business requirements, and project completion information, etc., the key business information to be identified may only be one of these elements, such as the start time or business requirements. In this case, target texts associated with the key business information to be identified can be identified first within the business information description text. The key business information can then be identified by targeting these target texts, thereby narrowing the identification scope and shortening the identification time.

[0034] S130, determine the artificial intelligence model corresponding to the type of information to be identified, process the target text based on the artificial intelligence model, and obtain the identification result of key business information.

[0035] The type of information to be identified refers to the type of key information to be identified, such as entity recognition type, content summary type, etc.; the entity recognition type refers to identifying specific information, such as identifying location, time, etc.; the content summary type refers to summarizing certain content.

[0036] Specifically, different artificial intelligence models are suitable for processing different types of information. For example, entity extraction models are suitable for recognizing specific information, such as time and location; fine-tuned large models are suitable for processing long texts and summarizing content; and semantic discrimination and matching models can extract and supplement key information.

[0037] Therefore, in this application embodiment, multiple artificial intelligence models are pre-trained, a suitable artificial intelligence model is determined according to the type of information to be identified, and then the target text is processed by the model to obtain the identification result of key business information.

[0038] The technical solution of this application embodiment includes: acquiring business information introduction text; determining target text associated with key business information to be identified in the business information introduction text; determining an artificial intelligence model corresponding to the type of information to be identified; and processing the target text based on the artificial intelligence model to obtain the identification result of the key business information. This technical solution shortens the scope of key information identification by determining the target text associated with the key business information to be identified in the business information introduction text, thus reducing the identification time. Furthermore, this technical solution uses an artificial intelligence model corresponding to the type of information to be identified to perform key information identification, making the identification result more accurate and reliable.

[0039] Example 2

[0040] Figure 2 This is a flowchart of a method for identifying key business information provided in Embodiment 2 of this application. This embodiment is an optimization based on the above embodiment.

[0041] like Figure 2 As shown, the method in this embodiment of the application specifically includes the following steps:

[0042] S210: Obtain the business information introduction document and convert the non-text introduction document into text format to obtain the business information introduction text.

[0043] Specifically, after obtaining the business information introduction document, if it is a PDF file or an image file, it is converted into text to obtain the business information introduction text.

[0044] S220, determine the chapter information of the business information introduction text through the chapter extraction model.

[0045] Specifically, a file format knowledge base can be obtained, and then a chapter extraction model can be used to identify the chapters of the business information text based on the information in the file format knowledge base, thereby obtaining information such as the name of each chapter and the page number corresponding to each chapter.

[0046] S230, Based on the chapter information, determine the target text associated with the key business information to be identified in the business information introduction text.

[0047] Since chapter information can reflect the content covered in a chapter, the key business information to be identified may be information from one or more chapters. Therefore, the target text associated with the key business information to be identified can be quickly determined in the business information introduction text based on the chapter information. This setting has the effect of shortening the time for determining the target text.

[0048] In this embodiment of the application, optionally, based on the chapter information, determining the target text associated with the business key information to be identified in the business information introduction text includes: obtaining a model fine-tuned by the business information; determining the target chapter in the chapter information that matches the business key information to be identified through the model; obtaining a pre-established business key information knowledge base; matching the content of the target chapter with the business key information knowledge base through a pre-trained domain matching dual-tower similarity model, and determining the part of the content of the target chapter that meets the matching degree condition as the target text.

[0049] Specifically, when determining the target text, a model fine-tuned with business information can be obtained first. This model can understand the nouns under this business, and then the chapter information can be processed by this model to identify the target chapters that match the key business information to be identified. Next, a pre-established knowledge base of key business information is obtained, and a pre-trained domain matching dual-tower similarity model is used to match the content of the target chapters with the knowledge base of key business information. The parts of the content of the target chapters that meet the matching degree conditions are determined as the target text. This approach achieves initial screening of the business information introduction text, and the content of the obtained target text is the content of the business information introduction text that is more consistent with the key business information, thereby improving the efficiency of subsequent key business information identification.

[0050] S240, determine the artificial intelligence model corresponding to the type of information to be identified, process the target text based on the artificial intelligence model, and obtain the identification result of key business information.

[0051] In this embodiment of the application, optionally, determining the artificial intelligence model corresponding to the type of information to be identified includes: determining the artificial intelligence model corresponding to the type of information to be identified in a pre-established model tool library.

[0052] Specifically, models suitable for recognizing various information types can be pre-established and stored in a model tool library. After determining the type of information to be identified corresponding to the key business information, an artificial intelligence model corresponding to the type of information to be identified can be determined in the model tool library, so as to identify the key business information based on the model.

[0053] In this embodiment of the application, optionally, after processing the target text based on the artificial intelligence model to obtain the identification result of key business information, the method further includes: performing short text verification and / or long text verification on the identification result to determine the identification result that fails the verification; among the target text corresponding to the identification result that fails the verification, determining the first text that meets the similarity requirement with the business domain knowledge base, and obtaining the second text filtered by the rule base established by expert experience; determining supplementary content based on the first text and the second text, and determining the final identification information according to the content that passes the verification in the identification result of key business information and the supplementary content.

[0054] Specifically, after obtaining the identification results of key business information, short texts (such as certain time and location information) can be validated, and long texts (such as a summary) can be validated. If all validations pass, the identification results of key business information are determined as the final identification information. If some validations fail, there may be identification errors in the failed parts. In this case, a first text that meets the similarity requirements of the business domain knowledge base can be determined from the target text corresponding to the failed identification results; and a second text filtered by a rule base established based on expert experience can be obtained from the target text corresponding to the failed identification results. Supplementary content is determined based on the first and second texts, and the final identification information is determined based on the validated content in the key business information identification results and the supplementary content. This scheme ensures the correctness of the content of the final identification information.

[0055] Optionally, in this embodiment of the application, the method further includes: performing preliminary screening of the final identification information based on preset rules, inputting the preliminary screened final identification information into multiple expert discrimination models, performing weighted summation of the outputs of each expert discrimination model to obtain a calculation result; if the calculation result is greater than a preset threshold, then determining that the final identification information is reliable.

[0056] Specifically, after obtaining the final identification information, its reliability can be assessed. This can be achieved by initially screening the final identification information based on preset rules. If the initial screening passes, the final identification information is input into multiple expert judgment models. These expert judgment models are three types: Prompt judgment based on a large model, domain category classification judgment, and knowledge base semantic matching judgment. The outputs of each expert judgment model are weighted and summed to obtain a calculation result. If the calculation result is greater than a preset threshold, the final identification information is determined to be reliable, and the final identification information can be fed back to the user.

[0057] Optionally, in this embodiment of the application, the method further includes: performing similarity recognition in the business information introduction text based on the final identification information to obtain the location of text that meets the similarity condition with the final identification information, so that the user can locate the business information introduction text corresponding to the final identification information based on the location of the text.

[0058] Specifically, after obtaining the final identification information, the location of the original business information description text corresponding to the final identification information can be marked, allowing users to view the corresponding original text. This greatly improves the user experience.

[0059] The technical solution of this application embodiment determines the target text associated with the key business information to be identified in the business information introduction text through chapter information, thereby achieving the effect of quickly identifying the text associated with the identification of key business information. Then, based on the artificial intelligence model, the target text is used to identify key business information, and accurate identification results are obtained.

[0060] In a specific example, we will use a bidding process as an example. Figure 3 This is a flowchart for extracting key information from a tender. The following is based on... Figure 3 Explain the specific implementation process of this solution.

[0061] (1) Obtain the tender documents from the file server. The tender documents can be in PDF, image, or WORD format. (2) After obtaining the PDF, image, and WORD files, provide user prompts for files that are not in these three formats. Convert the files in these three formats. Perform layout analysis, text recognition, and layout restoration on the image and PDF files, and restore them to text files for subsequent processing and extraction. At the same time, load the tender document format knowledge base, use the chapter extraction model combined with chapter rules to extract the chapter name and chapter starting page number, and return them together with the text document for subsequent steps.

[0062] (3) This step is mainly to narrow the range of key information to be extracted, reduce redundant input to subsequent models, improve extraction accuracy, and improve model recognition speed, and pre-select the key information area to be extracted. This step first uses a large model that has been fine-tuned on the tender document data to match the chapters of the tender document with the content of the key information to be extracted, initially assigning the types of key information to be extracted to the corresponding chapters, and splitting the chapter text content according to the chapter starting page number in step (2); then loading the domain knowledge base corresponding to the type to be extracted, the knowledge base can contain text statements, keywords, rules, etc., and then using the trained domain matching double-tower similarity model to match the chapter content and the knowledge base. The matching degree is modified from Triplet Loss, and the similarity distance of the same category and the similarity distance of different categories are calculated. The formula is as follows: Where margin is a constant greater than 0, which can be set manually. To calculate the distance to the current knowledge base sample, To calculate the distance to other knowledge base samples, matching texts with a Similarity greater than 0 are retained. After deduplication, the extracted category, the chapter and page number of the category, and the position of the candidate paragraph are output for the next step.

[0063] (4) Obtain information such as the category and the chapter where the category is located in step (3), and realize information extraction through a 1+N+N architecture. Here, 1 refers to the extraction category and the type of extraction model to be called through one model, N refers to the N model methods that can be used for the first information extraction, and the second N refers to the extraction of information through result verification. If no information that meets the requirements is found, then N models are used for supplementary extraction. N model toolkits are constructed. Through the study of information extraction categories, the types of information extraction are confirmed, such as whole paragraph information extraction, multi-paragraph information extraction and summary output, and entity extraction. Based on the above types, the model toolkit is confirmed to include semantic discrimination matching model, entity recognition model, large model, etc. By manually annotating the extracted content, the entity extraction model is trained to extract key information such as name, time, and place; by fine-tuning the large model to extract the whole paragraph information, and after extracting multiple paragraphs, the information is merged and summarized for output; by training the semantic discrimination model and the domain matching model, the category domain knowledge base is matched to perform supplementary extraction of key information. After training the above models, a tool library is created. The pre-built large model determines the category and type of content to be extracted. For long text extraction and content summary extraction, the large model is fine-tuned; for entity extraction, an entity recognition model is used. Corresponding tools are assigned to perform information extraction. After the first stage of extraction is complete, a prompt is built using the large model to validate the short text. A pass rate is 100%, and a fail rate is 0. For long text, a domain matching model is used to validate the content, verifying the relationship between the extracted content and the category. The validation output is a cosine similarity score, calculated using the formula: After calculating the matching degree, the second stage – the supplementary extraction stage – is initiated to extract the content categories and types with matching degrees less than a threshold. This supplementary extraction utilizes a domain knowledge base for bidding categories. For sentences segmented by periods, the cosine similarity is calculated sentence-by-sentence with content in the domain knowledge base, extracting information with similarities greater than a threshold (0.85 in this case). In parallel, a rule base is constructed based on type expert experience. Relevant category information is extracted using these rules. Finally, the domain matching and expert rule results are merged as the supplementary extraction results from the first stage. Combining the content extracted from the first and second stages, the final output includes the extracted information category, extracted information text, the chapter and page number where the extracted content is located, etc., which are then passed to the next step for post-processing and verification of the output results.

[0064] (5) After extracting information content and information category from the model in step (4), the information is verified and information with low matching degree with the category is removed to improve the extraction accuracy. This step is mainly divided into two stages. In the first stage, the information is initially filtered and extracted through the category experience knowledge base and category rules, and through experience rules such as keywords and sentence structure. After the initial filtering, an expert judgment model is constructed. The judgment is voted on by M expert models, and the results are summed according to the weights of different models. The extracted information with a value greater than the threshold is retained. In this case, three expert models are constructed based on the large model: Prompt judgment, domain category classification judgment, and knowledge base semantic matching judgment. After weighting the three models, the results are summed to calculate the total score. The formula for calculating the total score is as follows: ,in, These are the weights of the expert model results: the weight for the large model's discrimination result is 0.45, the weight for the domain category classification model is 0.3, and the weight for knowledge base semantic matching is 0.25. The model results are as follows: the large model's discrimination result is a binary value of 0 or 1, where 1 is a positive result and 0 is a negative result; the domain model result is the domain classification score, calculated using softmax. The softmax formula is: , K represents the expected classification result for the i-th text, where K is the number of classification results; cosine similarity is used for semantic matching in the knowledge base.

[0065] (6) Since generative methods such as large models are used in this case study, the extracted information may contain summaries and logical judgments that do not completely match the original text. Therefore, it is necessary to provide prompts for the relevant content so that users can distinguish them. At the same time, the extracted information should correspond to the original text for users to compare and verify. Therefore, the final output steps are as follows: First, obtain the post-processing extraction results, the chapter and page number of the results, etc. For all types of extracted text, calculate the original text matching degree of the corresponding page number according to the page number. The original text matching degree adopts the Levenshtein method, and its calculation steps are as follows:

[0066] 1. Initialize a matrix with the number of rows equal to the length of the first string plus 1 and the number of columns equal to the length of the second string plus 1;

[0067] 2. Fill the first row of the matrix with increments starting from 0.

[0068] 3. Fill the first column of the matrix with increments starting from 0.

[0069] 4. For the other positions in the matrix, fill according to the following rules:

[0070] - If two characters are equal, the value at that position is equal to the value at the top left corner.

[0071] - If two characters are not equal, the value at that position is equal to the value at the top left corner plus 1.

[0072] Finally, the value in the bottom right corner of the matrix is ​​the Levenshtein distance, which is the similarity between the strings. To convert the similarity to a range of 0 to 1, the matching score is calculated using the following formula:

[0073] ;

[0074] in, The distance is Levenshtein. The length of the string used for matching.

[0075] Calculate all extracted information and statements in the corresponding page numbers, return the start and end positions of the original text statement corresponding to the maximum matching degree, and finally return the extracted information category, extracted information text, original text page number, start and end positions of the original text matching statement and original text matching degree, thus completing the entire tender document extraction process.

[0076] Based on the steps (1)-(6) above, this proposal has constructed a tender document parsing system that extracts 47 types of key information across seven major categories, including project information and qualification review. The accuracy rate for entity extraction is 97.75%, and the accuracy rate for whole-segment extraction is 92%. This system can efficiently and accurately help users extract key information from tender documents, condensing hundreds of pages of tender document information into a few pages for structured display. Based on the original text matching results, users can quickly jump to view the original text content. At the same time, based on the OCR document restoration scheme, users can quickly obtain the table content in the tender document, realizing diverse extraction schemes for text information extraction and table extraction.

[0077] The above description uses a multi-model fusion information extraction method as an illustrative application, not for limiting purposes. This proposed solution can be widely applied and transferred to various information extraction scenarios. Furthermore, this proposal enables more flexible and efficient information extraction across different scenarios. The multi-model and multi-layered architecture ensures model robustness, meeting users' needs for high-accuracy information extraction in various scenarios.

[0078] This technical solution addresses the limitations of existing models in extracting entity triples, extending it to information extraction schemes that combine entity, whole paragraph, and multi-paragraph information. It also addresses the limitations of existing solutions using static word embedding methods by employing dynamic word embedding in category matching text matching, resulting in more accurate similarity matching scenarios. This application's technical solution adopts a 1+N+N multi-level, step-by-step extraction approach, enhancing model robustness and generalization while improving accuracy in complex extraction scenarios. This application's technical solution uses model fusion, leveraging the different performance preferences of various models to comprehensively process the results, resulting in higher accuracy and stability. This application's technical solution integrates domain knowledge and expert knowledge into the information extraction method, using domain knowledge matching and expert knowledge matching for preprocessing and post-processing of the text to be extracted, reducing the amount of text processed by the model, improving data processing efficiency, reducing redundant information transmission, and enhancing information extraction accuracy. This application's technical solution considers both original text information extraction and the extraction of results requiring logical judgment, using a large model to fuse and summarize information extracted from multiple paragraphs.

[0079] Example 3

[0080] Figure 4 This is a schematic diagram of a business key information identification device provided in Embodiment 3 of this application. This device can execute the business key information identification method provided in any embodiment of this invention, and has the corresponding functional modules and beneficial effects of the method. For example... Figure 4 As shown, the device includes:

[0081] The business information introduction text acquisition module 310 is used to acquire business information introduction text;

[0082] The target text determination module 320 is used to determine the target text associated with the key business information to be identified in the business information introduction text;

[0083] The business key information identification result determination module 330 is used to determine the artificial intelligence model corresponding to the type of information to be identified, and to process the target text based on the artificial intelligence model to obtain the identification result of the business key information.

[0084] The technical solution of this application embodiment includes: a business information introduction text acquisition module 310, used to acquire business information introduction text; a target text determination module 320, used to determine target text associated with key business information to be identified in the business information introduction text; and a key business information identification result determination module 330, used to determine an artificial intelligence model corresponding to the type of information to be identified, and process the target text based on the artificial intelligence model to obtain the identification result of the key business information. This technical solution shortens the scope of key information identification by determining the target text associated with the key business information to be identified in the business information introduction text, thus reducing the identification time. Furthermore, this technical solution uses an artificial intelligence model corresponding to the type of information to be identified to perform key information identification, making the identification result more accurate and reliable.

[0085] Optionally, in this embodiment of the application, the business information description text acquisition module 310 includes:

[0086] The text conversion unit is used to acquire business information introduction documents and convert non-text introduction documents into text form to obtain business information introduction text;

[0087] The chapter information determination unit is used to determine the chapter information of the business information introduction text through a chapter extraction model;

[0088] Accordingly, the target text determination module 320 includes:

[0089] The target text determination unit is used to determine, based on the chapter information, the target text associated with the business information to be identified in the business information introduction text.

[0090] Optionally, in this embodiment of the application, the target text determination unit is specifically used for:

[0091] Obtain the model after fine-tuning based on business information;

[0092] This model is used to identify target chapters within the chapter information that match the key business information to be identified.

[0093] Acquire a pre-established knowledge base of key business information;

[0094] The content of the target chapter and the business key information knowledge base are matched by a pre-trained domain matching dual-tower similarity model, and the part of the content of the target chapter that meets the matching degree condition is determined as the target text.

[0095] Optionally, in this embodiment of the application, the business key information identification result determination module 330 includes:

[0096] The artificial intelligence model determination unit is used to determine the artificial intelligence model corresponding to the type of information to be identified from a pre-established model tool library.

[0097] Optionally, in this embodiment of the application, the device further includes:

[0098] The verification module is used to perform short text verification and / or long text verification on the recognition result, and to determine the recognition result that fails the verification.

[0099] The first text and second text determination module is used to determine the first text that meets the similarity requirement with the business domain knowledge base among the target texts corresponding to the recognition results that fail the verification, and to obtain the second text filtered by the rule base established by expert experience.

[0100] The final identification information determination module is used to determine supplementary content based on the first text and the second text, and to determine the final identification information based on the content that has passed the verification in the identification results of key business information and the supplementary content.

[0101] Optionally, in this embodiment of the application, the device further includes:

[0102] The weighted calculation module is used to perform preliminary screening of the final identification information based on preset rules, input the preliminary screening final identification information into multiple expert judgment models, and perform weighted summation of the outputs of each expert judgment model to obtain the calculation result;

[0103] A reliability identification module is used to determine that the final identification information is reliable if the calculation result is greater than a preset threshold.

[0104] Optionally, in this embodiment of the application, the device further includes:

[0105] The location module is used to perform similarity recognition in the business information introduction text based on the final identification information to obtain the location of the text that meets the similarity condition with the final identification information, so that the user can locate the business information introduction text corresponding to the final identification information based on the location of the text.

[0106] The business key information identification device provided in this application embodiment can execute the business key information identification method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the method execution.

[0107] Example 4

[0108] Figure 5A schematic diagram of an electronic device 10, which can be used to implement embodiments of the present invention, is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (e.g., helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0109] like Figure 5 As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12 or a random access memory (RAM) 13, communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes based on the computer program stored in the ROM 12 or loaded from storage unit 18 into the RAM 13. The RAM 13 can also store various programs and data required for the operation of the electronic device 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0110] Multiple components in electronic device 10 are connected to I / O interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as disk, optical disk, etc.; and communication unit 19, such as network card, modem, wireless transceiver, etc. Communication unit 19 allows electronic device 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0111] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as methods for identifying business-critical information.

[0112] In some embodiments, the method for identifying business-critical information may be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the method for identifying business-critical information described above may be performed. Alternatively, in other embodiments, processor 11 may be configured to perform the method for identifying business-critical information by any other suitable means (e.g., by means of firmware).

[0113] 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), system-on-a-chip (SoCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0114] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0115] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0116] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0117] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or middleware components (e.g., application servers), or frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0118] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through a communication network. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.

[0119] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0120] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method for identifying key business information, characterized in that, include: Obtain business information introduction text; In the business information description text, identify the target text associated with the key business information to be identified; The artificial intelligence model corresponding to the type of information to be identified is determined, and the target text is processed based on the artificial intelligence model to obtain the identification result of key business information.

2. The method according to claim 1, characterized in that, Obtain business information description text, including: Obtain the business information introduction document and convert the non-text document into text format to obtain the business information introduction text; The chapter information of the business information introduction text is determined by the chapter extraction model; Accordingly, in the business information description text, target texts associated with the key business information to be identified are determined, including: Based on the chapter information, the target text associated with the key business information to be identified is determined in the business information introduction text.

3. The method according to claim 2, characterized in that, Based on the aforementioned chapter information, target texts associated with the key business information to be identified are determined within the business information description text, including: Obtain the model after fine-tuning based on business information; This model is used to identify target chapters within the chapter information that match the key business information to be identified. Acquire a pre-established knowledge base of key business information; The content of the target chapter and the business key information knowledge base are matched by a pre-trained domain matching dual-tower similarity model, and the part of the content of the target chapter that meets the matching degree condition is determined as the target text.

4. The method according to claim 1, characterized in that, Determine the artificial intelligence model corresponding to the type of information to be identified, including: In a pre-established model tool library, an artificial intelligence model corresponding to the type of information to be identified is determined.

5. The method according to claim 1, characterized in that, After processing the target text based on the artificial intelligence model to obtain the identification results of key business information, the method further includes: Perform short text verification and / or long text verification on the recognition result to determine the recognition result that fails the verification; Among the target texts corresponding to the recognition results that fail the verification, the first text that meets the similarity requirements with the business domain knowledge base is determined, and the second text is obtained by filtering the rule base established by expert experience. Based on the first text and the second text, supplementary content is determined, and the final identification information is determined based on the content that has passed verification in the identification results of key business information and the supplementary content.

6. The method according to claim 5, characterized in that, The method further includes: The final identification information is initially screened based on preset rules. The final identification information after initial screening is input into multiple expert judgment models. The outputs of each expert judgment model are weighted and summed to obtain the calculation result. If the calculation result is greater than the preset threshold, then the final identification information is determined to be reliable.

7. The method according to claim 5, characterized in that, The method further includes: In the business information introduction text, similarity recognition is performed based on the final identification information to obtain the location of text that meets the similarity condition with the final identification information, so that users can locate the business information introduction text corresponding to the final identification information based on the location of the text.

8. A device for identifying key business information, characterized in that, include: The business information introduction text acquisition module is used to acquire business information introduction text; The target text determination module is used to determine the target text associated with the key business information to be identified in the business information introduction text; The business key information identification result determination module is used to determine the artificial intelligence model corresponding to the type of information to be identified, and to process the target text based on the artificial intelligence model to obtain the identification result of the business key information.

9. An electronic device, characterized in that, The electronic device includes: At least one processor; and A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the identification method for business-critical information as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that, when executed by a processor, implement the method for identifying business-critical information as described in any one of claims 1-7.

Images