A data intelligent reporting method and related device

Abstract

The application provides a data intelligent filling method and related equipment, and relates to the technical field of data intelligent processing. Through the steps of obtaining data input information, identification processing, semantic inference, clarification correction and mapping execution, the problems that the prior art cannot effectively process diversified input, lacks context utilization, cannot actively clarify and cannot intelligently map are solved, the accuracy and intelligent degree of data filling are improved, the semantic ambiguity and information missing situation can be actively processed, intelligent data mapping across systems is realized, and the reliability and applicability of data intelligent filling are enhanced.

Description

Technical Field

[0001] This application relates to the field of data intelligence processing technology, and more specifically, to a data intelligence data entry method and related equipment. Background Technology

[0002] In scenarios involving automated data entry through processing images from instant messaging software, existing technologies typically rely on recognizing and extracting information from tables with clear and well-structured tables. However, in practical applications, the images processed by the system come from diverse sources. This includes structural ambiguities arising from the simultaneous use of multiple versions of standardized table templates, as well as unstructured data such as plain text or handwritten text that lacks any table structure. For example, data input from instant messaging software is often plain text or image information entered by the user. Furthermore, the image quality itself varies greatly, exhibiting issues such as blurriness, uneven lighting, and tilted angles.

[0003] There is currently no effective technical solution to the above problems. Summary of the Invention

[0004] In view of this, this application provides a data intelligent filling method and related equipment to improve the accuracy and intelligence of data filling, proactively handle semantic ambiguity and information missing situations, realize intelligent data mapping across systems, and enhance the reliability and applicability of data intelligent filling.

[0005] This application provides a method for intelligent data entry, including the following steps:

[0006] Retrieve data input information sent by instant messaging software;

[0007] The input data is processed to obtain the recognized text.

[0008] Based on the recognized text, combined with the real-time communication context information corresponding to the data input information and the preset business association information, business semantic inference is performed to obtain the initial business semantic information;

[0009] If there is information to be determined in the initial business semantic information, a clarification request corresponding to the initial business semantic information is generated and sent to the sender of the data input information through instant messaging software.

[0010] Based on the clarification information returned by the sender in response to the clarification request, the initial business semantic information is corrected to obtain the target business semantic information;

[0011] Map the target business semantic information to the corresponding fields of the target data system, and perform data writing operations in the target data system, which is the data system corresponding to the business-related information;

[0012] Based on the result of the data writing operation, a success message corresponding to the data input information is sent to the target user via instant messaging software; the target user includes the sender, the original group to which the sender belongs, or a pre-bound user.

[0013] Through this technical solution, this application can achieve intelligent recognition, semantic understanding, uncertainty handling and automated filling of diverse data inputs in instant messaging software. It effectively solves the problems of low accuracy and low efficiency of data filling in the existing technology when processing unstructured data and complex scenarios, and significantly improves the automation level and user experience of intelligent data filling.

[0014] Optionally, the steps for performing business semantic inference based on the recognized text, combined with the instant messaging context information corresponding to the data input information and the preset business association information, to obtain initial business semantic information include:

[0015] Obtain the instant messaging context information corresponding to the data input information;

[0016] Keyword extraction and topic classification are performed on instant messaging context information to obtain effective context information;

[0017] The identified text is matched with preset business terms to obtain preliminary business entity identification results;

[0018] By combining effective contextual information, the preliminary business entity identification results are adjusted to obtain the final business entity identification result;

[0019] Based on the preset business association information, the logical relationship between business entities in the business entity identification result is analyzed to obtain the semantic association inference result;

[0020] Initial business semantic information is obtained based on the business entity identification results and semantic association inference results.

[0021] By parsing the logical relationships between business entities based on pre-defined business association information, the system can deeply understand the business meaning of the input data. It constructs a complete business semantic graph from fragmented entities, enabling refined business semantic inference based on this graph. This allows the system to obtain more accurate and complete business semantic information from the initial stage, significantly reducing the frequency of generating clarification requests when there is undetermined information in the initial business semantic information. This means the system can perform data entry more efficiently, reduce the number of interactions with the sender, shorten the data processing cycle, and significantly improve the automation and overall efficiency of data entry. Simultaneously, the improved accuracy of the initial business semantic information also reduces the complexity and error rate of subsequent data correction, thereby enhancing the reliability of intelligent data entry.

[0022] Optionally, after obtaining initial business semantic information by performing business semantic inference based on the recognized text, combined with the instant messaging context information corresponding to the data input information and preset business association information, the process further includes:

[0023] Based on preset confidence thresholds and business rules, the accuracy of initial business semantic information is evaluated to obtain evaluation results;

[0024] If the evaluation result is accurate, the initial business semantic information will be identified as the target business semantic information.

[0025] Through the above technical solution, this application introduces an accuracy assessment mechanism for the initial business semantic information after performing business semantic inference. This mechanism can effectively identify and filter out inaccurate or ambiguous semantic information, thereby avoiding unnecessary clarification requests or erroneous data writing when information reliability is insufficient. When the assessment result shows that the initial business semantic information is accurate, the system can directly confirm it as the target business semantic information, significantly reducing the number of interactions with users and improving the automation level and overall efficiency of data entry. At the same time, through dual verification using preset confidence thresholds and business rules, it ensures that the data finally written into the target data system has higher accuracy and reliability, reducing the risk of subsequent business processing anomalies due to data errors.

[0026] Optionally, if there is information to be determined in the initial business semantic information, the step of generating a clarification request corresponding to the initial business semantic information includes:

[0027] If the evaluation result is pending, it is determined that there is undetermined information in the initial business semantic information;

[0028] Based on the information type corresponding to the information to be determined, identify the matters to be clarified in the initial business semantic information;

[0029] Based on the matters to be clarified, select the matching sentence structure from the preset clarification request sentence structures;

[0030] Fill the matters to be clarified into the matching sentence structure to generate a clarification request.

[0031] This solution intelligently generates targeted clarification requests when there is undetermined information in the initial business semantic information and proactively sends them to the sender to obtain necessary supplementary information. This enables the system to handle ambiguous or incomplete inputs that were previously impossible to process automatically, significantly improving the robustness and accuracy of intelligent data entry.

[0032] Optionally, the step of correcting the initial business semantic information based on the clarification information provided by the sender in response to the clarification request to obtain the target business semantic information includes:

[0033] Receive feedback from the sender regarding the clarification request;

[0034] Based on the keywords or structured information in the feedback, the initial business semantic information is matched and updated to obtain the target business semantic information.

[0035] Optionally, the steps of recognizing the input data to obtain the recognized text include:

[0036] When the input data is image information, image preprocessing is performed based on the input data to obtain a preprocessed image;

[0037] Optical character recognition is performed on the preprocessed image to obtain the recognized text.

[0038] Optionally, the steps of mapping the target business semantic information to the corresponding fields of the target data system and performing data writing operations include:

[0039] Based on the mapping rule information stored in the preset mapping rule library, determine the target field identifier and entity type corresponding to the target business semantic information;

[0040] Identify the fields in the target data system that correspond to the target field identifier as the corresponding fields;

[0041] Based on entity type, target business semantic information is mapped to corresponding fields to obtain mapped business data;

[0042] The target data system calls the data operation interface to perform the write operation of the mapped business data.

[0043] Secondly, this application provides a data intelligent entry system, comprising:

[0044] A monitoring agent is used to acquire data input information sent by instant messaging software.

[0045] The intelligent agent is used to identify and process input data to obtain identified text.

[0046] The inferential agent is used to perform business semantic inference based on the recognized text, combined with the real-time communication context information corresponding to the data input information and the preset business association information, to obtain the initial business semantic information.

[0047] The clarification agent is used to generate a clarification request corresponding to the initial business semantic information when there is something to be determined in the initial business semantic information, and send the clarification request to the sender of the data input information through instant messaging software.

[0048] The correction agent is used to correct the initial business semantic information based on the clarification information returned by the sender in response to the clarification request, so as to obtain the target business semantic information.

[0049] An execution agent is used to map target business semantic information to the corresponding fields of the target data system and perform data writing operations in the target data system, which is the data system corresponding to the business-related information.

[0050] The sending agent is used to send a success message corresponding to the data input information to the target user via instant messaging software based on the operation result of the data writing operation. The target user includes the sender, the original group to which the sender belongs, or a pre-bound user.

[0051] Thirdly, this application provides an electronic device, including a processor and a memory, wherein the memory stores computer-readable instructions, and when the computer-readable instructions are executed by the processor, the steps of a data intelligent filling method as described in any of the preceding claims are performed.

[0052] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, performs the steps of a data intelligent entry method as described in any of the preceding claims.

[0053] As can be seen from the above, the data intelligent filling method and related equipment provided in this application solve the problems of existing technologies that cannot effectively handle diverse inputs, lack context utilization, cannot actively clarify, and are difficult to intelligently map by acquiring data input information, recognizing and processing, semantic inference, clarification and correction, and mapping execution steps. It can improve the accuracy and intelligence of data filling, actively handle semantic ambiguity and information missing situations, realize intelligent data mapping across systems, and enhance the reliability and applicability of intelligent data filling.

[0054] Other features and advantages of this application will be set forth in the following description and will be apparent in part from the description or may be learned by practicing embodiments of this application. The objectives and other advantages of this application may be realized and obtained by means of the structures particularly pointed out in the written description and the accompanying drawings. Attached Figure Description

[0055] Figure 1 A flowchart illustrating a data intelligent entry method provided in an embodiment of this application.

[0056] Figure 2 This is a schematic diagram of the structure of the intelligent data entry system provided in the embodiments of this application.

[0057] Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application.

[0058] Labeling Explanation: 21. Monitoring Agent; 22. Identifying Agent; 23. Inferring Agent; 24. Clarifying Agent; 25. Correcting Agent; 26. Executing Agent; 27. Sending Agent; 13. Electronic Device; 1301. Processor; 1302. Memory; 1303. Communication Bus. Detailed Implementation

[0059] 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 a part of the embodiments of this application, and not all of the embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely represents selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0060] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0061] Under complex and ever-changing input conditions, how to accurately extract the correct semantic information from the data input information sent by instant messaging software and fill it into the corresponding fields of the target data system without error is a key technical challenge we are currently facing.

[0062] To address the problem of poor accuracy in intelligent data entry caused by the inability to extract correct semantic information from data input information sent by instant messaging software in existing technologies, this application provides an intelligent data entry method and related equipment that can proactively handle semantic ambiguity and information loss, and realize intelligent data mapping across systems to ensure the reliability and applicability of intelligent data entry.

[0063] Reference Figure 1 This application provides a data intelligent entry method, including the following steps:

[0064] Step S1: Obtain data input information sent by the instant messaging software;

[0065] Step S2: The input data is processed to obtain the recognized text;

[0066] Step S3: Based on the recognized text, combined with the instant messaging context information corresponding to the data input information and the preset business association information, perform business semantic inference to obtain the initial business semantic information;

[0067] Step S4: If there is information to be determined in the initial business semantic information, generate a clarification request corresponding to the initial business semantic information and send the clarification request to the sender of the data input information through instant messaging software.

[0068] Step S5: Based on the clarification information returned by the sender in response to the clarification request, the initial business semantic information is corrected to obtain the target business semantic information;

[0069] Step S6: Map the target business semantic information to the corresponding fields of the target data system, and perform a data writing operation in the target data system, where the target data system is the data system corresponding to the business association information;

[0070] Step S7: Based on the result of the data writing operation, send a success message corresponding to the data input information to the target user through instant messaging software; wherein, the target user includes the sender, the original group to which the sender belongs, or a pre-bound user.

[0071] Instant messaging software refers to applications that allow users to communicate in real time via the internet, such as WeChat, WeChat Work, DingTalk, and Lark. In this application, instant messaging software is not only a carrier for users to send data input information, but also a channel for the system to send clarification requests and success messages.

[0072] Data input information: refers to information containing business data sent by users through instant messaging software. It can be in the form of images (such as images containing tables, text, or handwriting) or plain text messages.

[0073] Instant messaging context information refers to the instant messaging environment information related to the data input information, such as the group name where the message belongs, the sender information, and the content of recently sent messages in that group. This information helps the system understand the context and business background of the data input information.

[0074] Business-related information refers to a pre-defined and stored set of knowledge used to assist the system in understanding and inferring the semantics of business data. This set of knowledge may include enterprise business terms, such as professional terms used in a specific industry or within the enterprise, product names, department names, etc., as well as semantic association rules, which define the logical relationships and inference patterns between these business terms.

[0075] As an example in this application, business association information can be specifically represented as a structured database containing multiple tables. For example, one table can store business terms, including fields such as "product model" (e.g., T-800, T-1000), "region name" (e.g., Area A, Area B), and "unit of measurement" (e.g., piece, unit, box, set). Another table can store semantic association rules, such as defining that when "value + unit of measurement" follows "product model," the value represents the sales quantity of the product; or the value following the keyword "sales amount" represents the sales amount. These rules can be stored in a Structured Query Language (SQL) statement or a format parsable by a specific rule engine.

[0076] When the system recognizes the text corresponding to the currently acquired data input as "Sales Report for Area A Today: 150 units of T-800 model, sales revenue of 30,000," the system will query the enterprise's business vocabulary to identify "Area A" as a regional entity, "T-800 model" as a product model entity, "unit" as a unit of measurement, "ten thousand" as a unit of monetary value, and "sales revenue" as a keyword. Subsequently, the system will apply semantic association rules to infer that "150" is the sales quantity of the "T-800 model" product, and "30,000" is the sales revenue.

[0077] Initial business semantic information: refers to the meaning of business data that the system initially infers based on the recognized text, instant messaging context information, and business-related information.

[0078] Target data system: refers to the data storage and management system corresponding to business-related information, such as Enterprise Resource Planning (ERP) systems, Customer Relationship Management (CRM) systems, or relational databases. It is the ultimate destination for intelligent data entry.

[0079] This application proposes a data intelligent entry method, which aims to solve the technical problem of accurately extracting semantic information from images in instant messaging software and correctly entering it into the target data system under complex and ever-changing input conditions.

[0080] First, this intelligent data entry method involves acquiring data input information sent from instant messaging software. In practical applications, users can send information containing business data through various instant messaging software (such as WeChat, WeChat Work, DingTalk, Lark, etc.). This data input information can be in image format, such as screenshots containing sales reports or inventory lists, or in plain text format, such as directly entered sales data or customer information. The intelligent data entry system continuously listens for and receives this data input information from instant messaging software. For example, when a user sends an image containing sales data in a WeChat Work group, the intelligent data entry system can capture that image information. As another implementation method, the system can also proactively retrieve messages from specific groups or users by integrating the open APIs of instant messaging software, thereby obtaining data input information.

[0081] Next, the input data is processed to obtain the recognized text. If the received input data is a plain text message, no image processing or Optical Character Recognition (OCR) is required; the system directly extracts the text content as the recognized text. If the input data is an image, the system needs to process the image to extract the text. For example, traditional image processing techniques such as grayscale conversion, binarization, and noise removal can be used to improve image quality. The processed image is then fed into the OCR engine for text recognition to obtain the recognized text. Alternatively, the system can directly call the OCR interface provided by a cloud service, uploading the original image to the cloud for recognition to obtain the recognized text.

[0082] Subsequently, based on the identified text, combined with the instant messaging context information corresponding to the data input information and preset business association information, business semantic inference is performed to obtain initial business semantic information. Upon receiving the identified text, the instant messaging context information related to the data input information is first obtained. For example, the group name (such as "Sales Daily Report Group," "Project Progress Group") and the content of several recently sent messages within that group can be obtained; this information provides important clues for understanding the context of the current data. Simultaneously, preset business association information, such as an enterprise business vocabulary (including product models, region names, promotional activity numbers, etc.) and semantic association rules, are utilized. By matching the identified text with the enterprise business vocabulary, business entities can be initially identified. For example, when "T-800" appears in the identified text, it can be identified as the "product model" entity according to the vocabulary. Combined with effective context information, the initial business entity identification results can be adjusted; for example, in the "Sales Daily Report Group," "150" is more likely to represent sales quantity than other values. Based on semantic association rules in business-related information, the logical relationships between business entities can be analyzed. For example, when "numerical value + unit of measurement" is followed by "product model", the numerical value represents the sales quantity of the product. Finally, based on the business entity identification results and semantic association inference results, initial business semantic information is obtained.

[0083] If the initial business semantic information contains information that needs to be determined, a clarification request corresponding to the initial business semantic information is generated and sent to the sender of the data input information via instant messaging software. During the business semantic inference process, the intelligent data entry system may find some information to be ambiguous, incomplete, or unclear, such as an unclear numerical value or an entity that cannot be uniquely identified. In this case, the clarification module intervenes and automatically generates a specific clarification request message based on the identified information that needs to be determined. For example, if "150 items" are identified but it is unclear which product they refer to, the intelligent data entry system will generate a request: "Which product are you referring to with the '150 items'?" This clarification request can be sent to the sender of the original data as a system message via the instant messaging software's API. Alternatively, the clarification request can also be sent to the sender via the chatbot function of the instant messaging software in the form of a preset template.

[0084] Based on the clarification information provided by the sender in response to the clarification request, the initial business semantic information is corrected to obtain the target business semantic information. The data intelligent entry system continuously monitors replies from senders in instant messaging software. Once a user's selection or input is received, the correction module parses the reply. If the reply is one of the preset options, it directly adopts that option to correct the initial business semantic information. If the reply is free text, the correction module performs keyword and pattern matching again, using this explicit information to complete or correct previously ambiguous data. For example, if the user replies "T-800", then "150 pieces" is associated with "T-800", thereby correcting the initial business semantic information and obtaining more accurate target business semantic information.

[0085] The execution module maps target business semantic information to corresponding fields in the target data system and performs data write operations within that system. The target data system is the data system corresponding to the business-related information. Once all key business entities are confirmed to be clear and complete, forming target business semantic information, the execution module intervenes. Based on mapping rules stored in a pre-defined mapping rule base, the execution module determines the target field identifier and entity type corresponding to the target business semantic information. For example, "Product Model" is "T-800", "Sales Quantity" is "150", and "Sales Amount" is "30000". The execution module precisely maps these business entities to the corresponding fields in the target template of the headquarters data center. For example, if the target system is a relational database, the execution module constructs an SQL INSERT statement to fill the parsed data into the corresponding tables and columns. If the target system is an Enterprise Resource Planning (ERP) system, the execution module calls the API interface provided by the ERP system to submit the data in a structured manner.

[0086] Finally, based on the data writing operation result, a success message corresponding to the data input information is sent to the target user via instant messaging software. After the data is successfully entered into the target system, the sending module will send a confirmation message of success to the target user through the instant messaging software's API. The target user can include the sender of the original data, the original group to which the sender belongs, or other pre-bound related users. For example, in a sales daily report group, after the data is successfully entered, the system will send a message to the group saying "Sales daily report has been successfully entered".

[0087] The following example will provide a more detailed explanation of the above technical solution:

[0088] Suppose that in a company's "Area A Sales Daily Report Group," user A sends an image containing handwritten sales data, including "150 T-800 models, sales revenue of 30,000." Because it is handwritten and the image may be slightly blurry, traditional systems would struggle to accurately recognize and process it.

[0089] First, the system obtains the image data input information sent by user A in an instant messaging software (such as WeChat for Business).

[0090] Next, the system performs recognition processing on the image. Since it's an image, image preprocessing is required, such as grayscale conversion, binarization, noise removal, and tilt correction, to improve image quality. Then, the preprocessed image is fed into the OCR engine for optical character recognition, yielding the recognized text, such as "T-800 model 150 units, sales revenue 30,000".

[0091] Subsequently, the system performs business semantic inference based on the recognized text, combined with instant messaging context information and preset business association information. It first obtains the instant messaging context information, such as recognizing the message as coming from the "Area A Sales Daily Report Group". Then, it matches the recognized text with a preset enterprise business vocabulary, identifying "T-800" as the product model entity, "150" as a numerical value, "piece" as the unit of measurement, "sales amount" as a keyword, "3" as a numerical value, and "ten thousand" as the monetary unit. Combining the context information ("Area A Sales Daily Report Group") and semantic association rules ("numerical value + unit of measurement" followed by "product model" indicates sales quantity, and the numerical value following the keyword "sales amount" indicates sales amount), the system initially infers the initial business semantic information: product model is T-800, sales quantity is 150 pieces, and sales amount is 30,000 yuan. However, due to the low confidence level of handwritten recognition, the system may determine that the number "30,000" in "sales amount" contains information that needs to be determined, such as whether it is "30,000" or "30,008".

[0092] In cases where there is undetermined information in the initial business semantic information, the clarification module generates a clarification request corresponding to the initial business semantic information. For example, the system automatically generates a clarification request: "Please confirm whether the sales amount is '30000' or '30008'?" and sends it to user A via WeChat.

[0093] After receiving the clarification request, User A replies with "30000". The correction module receives User A's clarification information and corrects the initial business semantic information, determining the sales amount to be 30,000 yuan, thus obtaining the target business semantic information: product model is T-800, sales quantity is 150 units, and sales amount is 30,000 yuan.

[0094] Then, the execution module maps the target business semantic information to the corresponding fields in the target data system. Based on preset mapping rules, the system determines that "Product Model" maps to the "Product Code" field in the target data system (e.g., the company's ERP system), "Sales Quantity" maps to the "Sales Quantity" field, and "Sales Amount" maps to the "Sales Amount" field. The execution module then calls the data operation interface provided by the ERP system to perform data write operations, writing "T-800", "150", and "30000" to their respective fields.

[0095] Finally, based on the results of the data writing operation, the sending module sends a success message corresponding to the data input information to the target user (e.g., all members in the "Area A Sales Daily Report Group") via WeChat Work, such as "Area A Sales Daily Report has been successfully submitted: 150 T-800 models, sales amount of 30,000 yuan".

[0096] The overall technical concept of this application lies in establishing a system capable of proactively understanding the semantics behind data and intelligently interacting with users to clarify information. This allows for accurate and complete acquisition and data entry even under complex and varied input conditions. Compared to existing technologies that rely on recognizing and extracting information from clearly structured table images, this application introduces real-time communication context information and business-related information for business semantic inference, effectively handling diverse data input formats such as multi-version tables, unstructured text, and handwritten text.

[0097] For example, in the above example, even with handwritten and slightly blurry images, this application can extract and recognize text through image preprocessing and OCR technology, and accurately infer sales data by combining the contextual information of "Area A Sales Daily Group" with preset corporate business vocabulary and semantic association rules.

[0098] Furthermore, this application introduces a clarification mechanism during the semantic inference process. When there is undetermined information in the initial business semantic information, it can proactively generate a clarification request and interact with the sender to obtain clear clarification information to correct the business semantics. This interactive correction process significantly improves the accuracy of data entry and effectively avoids misentry caused by ambiguous or unclear information. In the example above, the system can identify the potential uncertainty in the "sales amount" figure and proactively initiate clarification to user A, ultimately ensuring the accurate entry of sales amount. Traditional methods often cannot handle this uncertainty, which may lead to data errors or require manual intervention.

[0099] By precisely mapping the corrected target business semantic information to the corresponding fields of the target data system and performing data writing operations, this application achieves seamless conversion and automated data entry from unstructured or semi-structured data to structured business data. Finally, by sending a success message to the target user, user experience and information transparency are improved. These technical features of this application work together to form an efficient and robust intelligent data entry solution, significantly improving the automation level and accuracy of data processing in complex business scenarios.

[0100] In some implementations, the step of performing business semantic inference based on the identified text, combined with the instant messaging context information corresponding to the data input information and preset business association information, to obtain initial business semantic information includes:

[0101] Obtain the instant messaging context information corresponding to the data input information;

[0102] Keyword extraction and topic classification are performed on instant messaging context information to obtain effective context information;

[0103] The identified text is matched with preset business terms to obtain preliminary business entity identification results;

[0104] By combining effective contextual information, the preliminary business entity identification results are adjusted to obtain the final business entity identification results.

[0105] Based on preset business association information (such as enterprise business terms and semantic association rules), the logical relationship between business entities in the business entity recognition results is analyzed to obtain semantic association inference results;

[0106] Initial business semantic information is obtained based on the business entity identification results and semantic association inference results.

[0107] To accurately understand the business meaning of input data, the system needs to acquire its context. This can be achieved in several ways. For example, the system can utilize the open application programming interface (API) provided by instant messaging software to obtain the name of the group containing the current input data, as well as the content of the most recent messages (e.g., the last 5 to 10 messages) within that group. Alternatively, the system can maintain an internal message history database. When new input data is received, this database can be queried to retrieve related historical dialogue records, thus constructing complete contextual information. For instance, a pre-trained text classification model (e.g., a Transformer-based model) can be used to identify the topic of the contextual information and categorize it into preset business topic categories (e.g., sales reports, inventory counts, purchase requests, etc.). Alternatively, rule-based pattern matching can be used to identify specific business terms or phrase combinations within the contextual information, thereby inferring its topic. Through these processes, irrelevant information can be effectively filtered out, focusing on core business semantics. To identify specific business entities from the identified text, the system needs to compare them with preset business knowledge. This can be achieved through dictionary-based matching, which precisely or fuzzily matches words in the identified text against a structured business vocabulary. This vocabulary can be stored as key-value pairs; for example, product models like "T-800" and "T-1000" can be categorized under the "Product Model" entity type, or region names like "Area A" and "Area B" can be categorized under the "Region Name" entity type. Alternatively, regular expression-based pattern matching can be used to identify business entities conforming to specific formats, such as dates, amounts, and numbers. Initial business entity identification results may be ambiguous or incomplete, requiring optimization based on context. The system can preset a set of adjustment rules; for example, if the initial identification result is "apple," and the relevant contextual information contains keywords such as "mobile phone" or "press conference," it can be adjusted to "Apple Inc."; if it contains keywords such as "fruit" or "sweetness," it can be adjusted to "apple (fruit)."

[0108] To understand the deep relationships between business entities, the system needs to perform logical inference using pre-defined business knowledge. This can be achieved by pre-setting a series of semantic association rules. For example, when "numerical value + unit of measurement" is followed by "product model," the numerical value represents the sales quantity of the product. These rules can be constructed based on grammatical structure, word order, or business logic. After completing the identification of business entities and the inference of logical relationships between entities, the system needs to integrate this information to form a structured, preliminary business semantic representation. This can be achieved by integrating the identified business entities (e.g., product model, region, quantity, sales revenue) and their inferred logical relationships (e.g., the sales quantity of "T-800" is "150 units") into a structured data object, such as JSON or XML format. The initial business semantic information can be a record containing multiple fields, each corresponding to a business entity or its attribute, and including a description of the relationships between entities, thus providing a clear and operable semantic foundation for subsequent data entry.

[0109] Specifically, by extracting keywords and classifying topics from instant messaging context information, the system can effectively filter noise and focus on core business semantics, thus providing high-quality contextual information for subsequent entity recognition and relationship inference. Matching the recognized text with preset enterprise business vocabulary and adjusting it with effective contextual information significantly improves the accuracy and robustness of business entity recognition, avoiding recognition errors caused by vocabulary diversity or ambiguity. More importantly, by parsing the logical relationships between business entities based on preset business association information, the system can deeply understand the business meaning of the data input information and construct a complete business semantic graph from scattered entities. This refined business semantic inference process allows the system to obtain more accurate and complete business semantic information at the initial stage, thereby significantly reducing the frequency of generating clarification requests when there is undetermined information in the initial business semantic information. This means the system can perform data entry more efficiently, reducing the number of interactions with the sender, shortening the data processing cycle, and significantly improving the automation and overall efficiency of data entry. At the same time, the improved accuracy of the initial business semantic information also reduces the complexity and error rate of subsequent data correction, thus enhancing the reliability of intelligent data entry.

[0110] The following is a concrete example. Suppose the system receives a message from a "Sales Daily Report Group" via instant messaging software. The message content is an image, and after recognition processing, the resulting text is: "Sales Report for Area A Today: 150 units of T-800 model, sales revenue of 30,000." First, the system obtains the instant messaging context information corresponding to the input data. For example, through the instant messaging software's API, the system obtains that the message comes from the "Sales Daily Report Group," and that the most recent messages are related to sales performance and product models. Next, the system performs keyword extraction and topic classification on this instant messaging context information, identifying keywords such as "sales," "daily report," and "performance," and classifying the topic as "sales report," thus obtaining valid context information. Subsequently, the system matches the recognized text "Sales Report for Area A Today: 150 units of T-800 model, sales revenue of 30,000" with preset enterprise business terms. For example, the system's internally maintained vocabulary includes: "Product Model": ["T-800", "T-1000"], "Region Name": ["Area A", "Area B"], "Unit of Measurement": ["piece", "item"], and "Unit of Amount": ["ten thousand"]. Through matching, the system initially identifies "Area A" as a region entity, "T-800" as a product model entity, "150" as a numerical value, "piece" as a unit of measurement, "3" as a numerical value, "ten thousand" as a unit of amount, and "sales amount" as a keyword. Based on this, the system adjusts the initial business entity identification results by combining previously obtained valid contextual information (i.e., the theme is "sales report"). For example, if "150" might be identified as a date in other contexts, in the context of "sales report," combined with the unit of measurement "piece," the system will explicitly adjust it to a "quantity" entity. Furthermore, the system can parse the logical relationships between business entities in the business entity identification results based on preset business association information. For example, the system's preset semantic association rules might include: when "numerical value + unit of measurement" immediately follows "product model," the numerical value represents the sales quantity of the product; the numerical value following the keyword "sales amount" represents the sales amount. Based on these rules, the system infers that the sales quantity of "T-800 model" is "150 units," and the "sales amount" is "30,000." Ultimately, the system obtains initial business semantic information based on the business entity identification results and semantic association inference results. This information can be structured as: {"Business Type": "Daily Sales Report," "Region": "Area A," "Product Sales": [{"Product Model": "T-800," "Sales Quantity": "150 units"}], "Total Sales Amount": "30,000"}. Such structured information provides a clear and accurate semantic foundation for subsequent data entry.

[0111] In some implementations, after performing business semantic inference based on the identified text, combined with the instant messaging context information corresponding to the data input information and preset business association information, to obtain initial business semantic information, the method further includes:

[0112] Based on preset confidence thresholds and business rules, the accuracy of initial business semantic information is evaluated to obtain evaluation results;

[0113] If the evaluation result is accurate, the initial business semantic information will be identified as the target business semantic information.

[0114] The phrase "assessing the accuracy of initial business semantic information based on preset confidence thresholds and business rules" refers to the process of using pre-defined quantitative standards and business logic constraints to check the reliability, completeness, and compliance of business semantic information inferred from input data. The preset confidence threshold can be a numerical value, for example, a business entity is considered accurate when its confidence score is higher than 0.9; or it can be a range, for example, ambiguity is considered present when the confidence differences between multiple identification results are less than a certain value. This threshold can be set and adjusted based on historical data analysis, expert experience, or machine learning model training results. Business rules refer to a set of rules used to verify the logicality and compliance of semantic information in a specific business scenario. For example, in sales data entry, business rules might stipulate that "sales quantity must be a positive integer," "sales amount cannot be lower than cost price," and "product model must exist in the product master data list." These rules can be defined and stored in the form of conditional statements, decision trees, or business process diagrams. Accuracy assessment can include checking the identification confidence of each business entity and verifying whether the logical relationships between business entities conform to preset business rules. For example, if the "product model" and "sales quantity" are identified, the system will check whether the product model exists and whether the sales quantity conforms to the product's sales specifications.

[0115] "Identifying the initial business semantic information as the target business semantic information when the evaluation result is accurate" means that when the evaluation result shows that the initial business semantic information is considered highly reliable, complete, and consistent with business logic, the system formally establishes it as the final target business semantic information for data entry. This implies that the system has sufficient confidence in the currently inferred business semantic information and can directly use it for subsequent data writing operations without further manual intervention or clarification. Once identified as the target business semantic information, it will serve as the basis for subsequent data mapping and writing operations, effectively improving the efficiency of intelligent data entry.

[0116] This solution is closely integrated with the basic business semantic inference process. As mentioned earlier, semantic inference is performed by recognizing text, contextual information, and business-related information to generate initial business semantic information. However, this inference may be uncertain due to the complexity, ambiguity, or polysemy of the input data. This solution adds a crucial quality control step. It doesn't simply accept or reject the initial inference result, but rather conducts a systematic "review." Using a preset confidence threshold, the system can quantitatively determine the reliability of each identified business entity; for example, whether the confidence level of a product name is high enough. Simultaneously, through business rules, the system can logically verify whether the combination of these business entities meets the requirements of the actual business scenario; for example, whether the sales volume is positive, or whether a specific product can only be sold in a specific region. Only when the initial business semantic information simultaneously meets the confidence requirements and the constraints of the business rules—that is, when the evaluation result is accurate—is it "promoted" to target business semantic information and directly enters the subsequent data mapping and writing stages. This mechanism avoids blindly writing data under conditions of high uncertainty, and also avoids making redundant clarification requests when the information is already accurate enough. Thus, while ensuring data quality, it optimizes the efficiency and user experience of the entire data entry process.

[0117] For example, after the system obtains initial business semantic information through semantic inference, such as "Product Model: T-800, Sales Quantity: 150 units, Sales Amount: 30,000," the system immediately initiates an accuracy assessment process. First, the system checks the recognition confidence of each business entity. For instance, it can directly obtain the recognition confidence of characters or words output by the Optical Character Recognition (OCR) engine, and use this obtained confidence as the recognition confidence of the business entity. Specifically, when recognizing text in an image, the OCR engine not only outputs the recognized text content but also provides a confidence score for each recognized character, word, or text block. For example, a clear "T-800" might obtain a recognition confidence of 0.98, while a blurry handwritten digit "3" might only have a recognition confidence of 0.60. Assuming that for "Product Model: T-800," if its recognition confidence is 0.98, which is higher than the preset confidence threshold of 0.95, then the entity is considered to be accurately recognized. For "Sales Quantity: 150 units", if the recognition confidence score for the number "150" is 0.97, the recognition confidence score for the unit "unit" is 0.99, and the combined confidence score is also higher than the threshold, then the entity is considered accurately identified. For "Sales Amount: 30,000", if the recognition confidence score for "3" is 0.96, the recognition confidence score for "10,000" is 0.98, and the combined confidence score is also higher than the threshold, then the entity is considered accurately identified. Secondly, the system applies preset business rules for logical verification. For example, business rules might include: "Sales quantity must be an integer greater than 0". The system checks if "150" meets this rule, and the result is yes. "Product model must exist in the current valid product list". The system queries the product database and confirms that "T-800" is a valid product model. "Sales amount must logically match the sales quantity and product unit price (allowing a certain margin of error)". The system can calculate the theoretical sales revenue (150 units * 200 yuan / unit = 30,000 yuan, or 30,000 yuan) based on the preset unit price of "T-800" (e.g., 200 yuan / unit), and compare it with the identified "30,000 yuan". If all business entities pass the confidence threshold check and all relevant business rule verifications pass, the evaluation result will be considered accurate. In this case, the system directly identifies the initial business semantic information of "Product Model: T-800, Sales Quantity: 150 units, Sales Revenue: 30,000 yuan" as the target business semantic information, without needing to send a clarification request to the sender, and can directly proceed to subsequent data mapping and writing operations.

[0118] In some implementations, when there is information to be determined in the initial business semantic information, the step of generating a clarification request corresponding to the initial business semantic information includes:

[0119] If the evaluation result is pending, it is determined that there is undetermined information in the initial business semantic information;

[0120] Based on the information type corresponding to the information to be determined, identify the matters to be clarified in the initial business semantic information;

[0121] Based on the matters to be clarified, select the matching sentence structure from the preset clarification request sentence structures;

[0122] Fill the matters to be clarified into the matching sentence structure to generate a clarification request.

[0123] The working principle of this scheme is that after the system performs an accuracy assessment of the initial business semantic information, if the assessment result shows a pending state, it indicates that the current information is insufficient to support accurate data entry. In this case, the system will proactively intervene and initiate a clarification request generation process. First, based on the assessment result, the system will identify which specific information points in the initial business semantic information are pending determination. This step aims to identify uncertainties or ambiguities in the initial business semantic information that prevent the system from directly identifying them as target business semantic information. After the initial business semantic information is assessed for accuracy, if the assessment result fails to reach a preset confidence threshold, the assessment result is determined to be a pending result. For example, if the identification confidence of a key business entity (such as product model or sales quantity) is lower than the preset confidence threshold, the assessment result corresponding to that business entity can be considered a pending result, and thus, based on this pending result, it can be concluded that there is pending information in the initial business semantic information. One implementation method is that the system maintains a confidence model; when the overall confidence of the initial business semantic information is lower than a certain preset threshold (e.g., 70%), it is determined to be a pending result. Another approach is to pre-define a series of business rules, such as "the product model must appear together with the sales quantity". If the initial business semantic information only identifies the sales quantity but not the product model, it is determined to be a result to be determined.

[0124] Next, the system will accurately identify the specific matters requiring user clarification based on the specific information type corresponding to these matters (e.g., product information, quantity information, or amount information). This step aims to precisely locate the specific parts of the initial business semantic information that require user clarification. The information to be determined may involve multiple business entity types, such as product model, sales quantity, date, amount, customer name, etc. The system needs to identify which business entity(s) are uncertain based on these information types, thus forming matters requiring clarification. One implementation method is to maintain a mapping table between business entity types and uncertainty patterns internally. For example, when the OCR engine recognizes the "sales quantity" area, if it gives multiple possible recognition results (e.g., "150" and "158"), and if the confidence levels of these results are relatively close (e.g., the confidence levels of these results are all between 0.85 and 0.95), then the system can determine that the "sales quantity" entity has fuzzy recognition and identify it as "sales quantity needs clarification". Another approach is to use Natural Language Processing (NLP) technology to analyze words or phrases related to uncertainty in the initial business semantic information, and combine this with a pre-defined business vocabulary to identify the specific business entities to be clarified and their content.

[0125] Subsequently, to ensure that the generated clarification requests are clear, standardized, and easy for users to understand and respond to, the system intelligently selects the most suitable sentence template from a pre-stored library of clarification request sentence structures designed for different business scenarios and information types, based on the identified matters to be clarified. This step aims to ensure that the generated clarification requests are clear, standardized, and easy for users to understand and respond to. The system selects the most appropriate sentence structure from a predefined sentence template library based on the nature of the identified matters to be clarified. These sentence structures are designed for different types of matters to be clarified, aiming to guide users to provide the required information. One implementation is that the system maintains a database containing multiple sentence templates. For example, the template for "product model to be clarified" could be "Which product are you referring to with '[content to be clarified]'?"; the template for "sales amount to be clarified" could be "Please confirm whether the sales amount is '[Option 1]' or '[Option 2]'?". The system can match and select based on the type of matters to be clarified (such as product model, sales amount). Another implementation is to use a rule-based matching engine to dynamically select the sentence structure that best fits the context based on the keywords or semantic tags of the matters to be clarified.

[0126] Finally, the system precisely fills the identified content to be clarified (e.g., vague product names or uncertain numbers) into the selected sentence structure, thereby dynamically generating a complete, specific clarification request that can be directly sent to the sender of the data input information. This step embeds the identified content to be clarified into the selected sentence structure, thus forming a complete, specific clarification request message that can be sent to the user. This process typically involves variable substitution or string formatting. One alternative implementation is to use placeholder substitution techniques, such as replacing "[content to be clarified]" in the sentence structure with the actual identified vague information (e.g., "150 items"). Another alternative implementation is to pass the matter to be clarified as a parameter to the sentence generation function through a programming interface, allowing the function to dynamically construct the complete clarification request text.

[0127] Through this mechanism, the system can proactively interact with users to obtain necessary supplementary information, effectively solving the data entry problem caused by ambiguous or incomplete information, and significantly improving the accuracy and automation of data entry. This solution, closely integrated with the aforementioned accuracy assessment steps, forms a closed-loop intelligent processing flow, ensuring that even when faced with complex and ever-changing data input, the system can ultimately obtain accurate target business semantic information through a proactive clarification mechanism, thereby smoothly completing the data writing operation.

[0128] The following example illustrates this. Suppose the system receives an instant messaging message containing an image. After recognition processing, the initial business semantic information identifies "sales quantity as 150 units." However, due to image quality issues or insufficient contextual information, the system cannot determine which product these "150 units" specifically refer to. In this case, when evaluating the accuracy of the initial business semantic information, the evaluation result will show as "pending determination" because the entity "sales quantity" lacks a corresponding "product model" entity, which does not conform to the preset business rules. The system will then determine that there is pending information in the initial business semantic information, specifically the missing "product model." Next, based on the information type corresponding to the pending information (i.e., the "product model" type), the system identifies the matter to be clarified as "product model." Subsequently, the system will select a sentence structure from a preset clarification request sentence structure library that matches "product model pending clarification," for example: "Which product is the '[content to be clarified]' you mentioned about?" Finally, the system fills the identified ambiguous information "150 items" or its associated contextual information into the sentence structure, generating a specific clarification request, such as: "Which product are you referring to with the '150 items'?"

[0129] As a specific implementation method, the system can also provide preset options for users to quickly select based on the matter to be clarified. For example, for product clarification, it may provide a list of recently sold products (such as "T-800", "T-1000") as options to reduce the burden of user responses. As another example, if "sales amount" is recognized but the number is ambiguous, the system will generate a request: "Please confirm whether the sales amount is '30000' or '30008'?".

[0130] Through the above technical solution, this solution ensures that the system can intervene in a timely and effective manner when uncertainty is detected, avoids the writing of erroneous data, and reduces the workload of manual verification and correction, thereby improving the automation level and operational efficiency of the entire data entry process.

[0131] In some implementations, the step of correcting the initial business semantic information based on the clarification information provided by the sender in response to the clarification request to obtain the target business semantic information includes:

[0132] Receive feedback from the sender regarding the clarification request;

[0133] Based on the keywords or structured information in the feedback, the initial business semantic information is matched and updated to obtain the target business semantic information.

[0134] Specifically, in the stage of receiving feedback from the sender regarding the clarification request, the system continuously listens for and obtains the sender's response to the previously sent clarification request by establishing a valid connection with the instant messaging software. This can be achieved through various technical means. For example, the system can periodically query the latest messages of a specific session or user through the instant messaging software's application programming interface (API) to obtain feedback information; or, the system can be configured to receive real-time message notifications (webhooks) pushed by the instant messaging software, and process them immediately once new feedback messages arrive.

[0135] In the step of matching and updating the initial business semantic information based on keywords or structured information in the feedback information to obtain the target business semantic information, the system intelligently parses the received feedback information. If the feedback information is structured, such as when the sender selects from a preset list of options, the system can directly map the selection to the corresponding field to be determined in the initial business semantic information and update it. If the feedback information is in free text form, the system will use Natural Language Processing (NLP) techniques, including but not limited to keyword extraction, named entity recognition, and pattern matching, to identify key information related to the matter to be clarified (e.g., specific product model, accurate numerical value, confirmation date, etc.). The identified key information is then used to accurately match and correct uncertain or ambiguous parts in the initial business semantic information, thereby forming complete and accurate target business semantic information.

[0136] This application's solution establishes an interactive closed loop with the sender, enabling the system to directly obtain clarification from the user regarding information to be determined, thus effectively resolving potential ambiguities and vagueness during semantic inference. After issuing a clarification request, the system continuously listens for responses from the sender. Once a response is received, whether it's a structured selection from preset options or free text input by the user, the system intelligently parses it. For structured feedback, the system directly and accurately updates the initial business semantic information; for free text feedback, the system uses advanced text analysis technology to extract key business entities or values ​​and matches and corrects them against the items to be determined in the initial business semantic information. This mechanism ensures that the initial business semantic information can be accurately completed or corrected, transforming the information to be determined into clear target business semantic information, providing a highly accurate data foundation for subsequent data writing operations. This, closely integrated with the aforementioned steps of obtaining data input information, identification and processing, business semantic inference, and generating clarification requests, constitutes a robust intelligent data entry process, significantly improving the entire system's ability to handle complex and uncertain data input.

[0137] The following is a concrete example to illustrate this. Suppose that during the business semantic inference process described above, the system recognizes the quantity information "150 units" in the input data, but cannot specify the corresponding product model. Therefore, it generates and sends a clarification request: "Which product are you referring to with the '150 units'?". The system then continuously monitors for replies from the sender via instant messaging. Once it receives a clarification reply from the sender, such as "T-800," the system parses the feedback. In this case, "T-800" is identified as a keyword, directly corresponding to the product model, the matter to be clarified. The system then updates the product model field associated with "150 units" in the initial business semantic information to "T-800" based on this keyword, thus obtaining the clear target business semantic information: "Product model is T-800, sales quantity is 150 units." For example, if the clarification request concerns a vague sales figure, and the sender replies "sales amount is 30,000," the system will identify the key value "30,000" and fill it into the corresponding sales amount field in the initial business semantic information, completing the information correction. This closed-loop correction mechanism through human-computer interaction effectively avoids erroneous mappings caused by improper information matching, significantly improves the accuracy and completeness of business semantic information, provides a reliable data source for subsequent data writing operations, and thus ensures the overall quality and efficiency of intelligent data entry.

[0138] In some implementations, the step of processing the input data to obtain the recognized text includes:

[0139] When the input data is image information, image preprocessing is performed based on the input data to obtain a preprocessed image;

[0140] Optical character recognition is performed on the preprocessed image to obtain the recognized text.

[0141] Specifically, when the system receives image-formatted input data, such as image files (JPG, PNG, BMP, etc.) sent via instant messaging software, this solution will initiate a dedicated image data processing flow. This indicates that the system can distinguish between different types of data input and select the appropriate processing path based on the input type, ensuring the targeted nature and effectiveness of subsequent processing. Image preprocessing refers to a series of image enhancement and correction operations performed on the original image before optical character recognition (OCR). Its purpose is to improve image quality, eliminate or mitigate adverse factors such as noise, blur, uneven lighting, tilt, and perspective distortion, thereby improving the accuracy and robustness of subsequent OCR. Image preprocessing may include, but is not limited to, grayscale conversion, binarization, denoising, tilt correction, and perspective transformation correction. The preprocessed image is the image obtained after image preprocessing operations; its quality is improved compared to the image in the original data input information, making it more suitable for OCR. This image typically has clearer text edges, a more uniform background, correct text orientation, and more regular geometric shapes. Optical Character Recognition (OCR) is a technology that converts handwritten or printed text in an image into machine-coded text. After obtaining a pre-processed image, the system uses an OCR engine to recognize and extract the text content within the image. OCR technology can convert pixel information in an image into editable and searchable text data, and typically provides the recognized text content and its location within the image. The recognized text is the output of the optical character recognition operation, i.e., the text content extracted from the pre-processed image. This text data is presented in a machine-readable format, facilitating further analysis and processing by the system, and forming the basis for subsequent business semantic inference.

[0142] In some implementations, the steps of mapping target business semantic information to corresponding fields in the target data system and performing data write operations include:

[0143] Based on the mapping rule information stored in the preset mapping rule library, determine the target field identifier and entity type corresponding to the target business semantic information; (wherein, the mapping rule information is used to determine the target field identifier and entity type corresponding to the target business semantic information).

[0144] Identify the fields in the target data system that correspond to the target field identifier as the corresponding fields;

[0145] Based on entity type, target business semantic information is mapped to corresponding fields to obtain mapped business data;

[0146] By calling the data operation interface (application interface or direct database operation interface) provided by the target data system, the write operation of the mapped business data is performed.

[0147] The pre-defined mapping rule base is a knowledge base that stores the correspondence between business semantic information and fields in the target data system. It can be a structured database table, for example, containing columns such as "business entity name," "target field name," "field type," and "data format conversion rules"; or a configuration file, such as an XML or JSON rule file, defining the mapping logic for different business scenarios. Mapping rule information refers to the specific entries or logic in the mapping rule base that guide the mapping process. This information can include direct correspondences between field names, data type conversion rules, data format validation rules, and mapping conditions for specific business logic. The target field identifier is the name or code used to uniquely identify a field in the target data system. For example, in a relational database, it can be a combination of table and column names; in an API interface, it can be a parameter name in the request body. The entity type refers to the category or format of the data represented by the target business semantic information. For example, "product model" might correspond to a string type, "sales quantity" might correspond to an integer type, and "sales amount" might correspond to a floating-point number type. Entity types may also include more detailed classifications, such as date, time, and boolean values. The first step involves identifying the field in the target data system that corresponds to the target field identifier. This step aims to precisely locate the specific field in the actual target data system where data needs to be entered, based on the target field identifier determined in the previous step. For relational databases, the system can find the corresponding physical field based on the target field identifier by querying the database metadata or pre-loaded database table structure definitions. For systems that write data through API interfaces, the target field identifier usually directly corresponds to the parameter name in the API request body. The system will match the identifier with the input parameters of the API interface according to the API documentation or preset interface definitions to determine the "corresponding field" where the data should be entered. The next step is to map the target business semantic information to the corresponding field based on the entity type, obtaining mapped business data. This step converts the semantically parsed business information into a format and value acceptable to the target data system, ensuring data format compatibility and correctness. If the entity type is "date" and the target data system requires the "YYYY-MM-DD" format, the system will convert the original date information to the target format. If the entity type is "amount" and the target system requires storage in "cents", the system will convert "30,000" to "3,000,000". For enumeration types or code values, the system will map the business semantic information to the corresponding code within the target system according to the entity type and preset conversion rules.By calling the data operation interface provided by the target data system, the write operation of the mapped business data is performed. The data operation interface is the data interaction channel provided by the target data system to the outside world. It can be an application programming interface (API), such as an ERP system providing a RESTful API, allowing external systems to submit data via HTTP requests; or it can be a direct database operation interface, which refers to interacting directly with the database through a database connection driver to execute SQL statements. Performing a write operation means persisting the mapped business data to the target data system. This can be adding a new record or updating an existing record.

[0148] This application's solution addresses the issues of low mapping accuracy and efficiency during data entry by introducing a systematic mapping and execution mechanism. It ensures seamless and error-free transformation of semantic information into operable data when dealing with diverse data systems and entity types. This solution is closely integrated with the aforementioned steps of acquiring data input information, identification and processing, business semantic inference, clarification requests, and information correction. In particular, after the intelligent data entry system performs in-depth analysis of the data input information and, when necessary, uses a clarification mechanism to ensure the clarity and completeness of the business semantic information, this solution, as the final stage of the data entry process, inherits the results of the preceding steps. It accurately transforms the highly refined and confirmed target business semantic information obtained in the preceding steps into actual data in the target data system through a systematic mapping and writing mechanism. This connection ensures the automation, intelligence, and high reliability of the entire chain from raw data input to final data entry. In this way, this solution not only solves the accuracy problem of data mapping and writing but also enables the entire data entry process to adapt to complex and ever-changing input scenarios and diverse target data systems, significantly improving the efficiency and quality of data processing.

[0149] The following is a concrete example to illustrate this. When the intelligent data entry system identifies key business entities, such as "Product Model" as "T-800", "Sales Quantity" as "150", and "Sales Amount" as "30000", the system will execute data entry. First, it queries a pre-defined mapping rule library. This library might be a JSON file configured internally within the system, defining how "Product Model" should be mapped to the "product_model" field in the target data system (e.g., a database table named "Sales Records"), with the entity type being string; how "Sales Quantity" should be mapped to the "quantity" field, with the entity type being integer; and how "Sales Amount" should be mapped to the "amount" field, with the entity type being floating-point. Based on these mapping rules, the system determines that the target field identifier for "Product Model" is "Sales Records.product_model" (entity type string); the target field identifier for "Sales Quantity" is "Sales Records.quantity" (entity type integer); and the target field identifier for "Sales Amount" is "Sales Records.amount" (entity type floating-point). Subsequently, based on these target field identifiers, the system locates the three corresponding fields "product_model," "quantity," and "amount" in the "Sales Records" table within the target data system (e.g., a MySQL database). Next, based on the entity type, the system transforms the target business semantic information. For example, "T-800" is directly mapped as a string; "150" as an integer; and "30000" as a floating-point number. If the original semantics of "sales amount" is "30,000," but the target system requires a floating-point number in "yuan," the system will convert it to "30000.00." These transformed data collectively constitute the mapped business data.Finally, the intelligent data entry system calls the direct database operation interface provided by the target data system, such as using a JDBC driver, to construct an SQL INSERT statement. For example, the constructed SQL INSERT statement could be "INSERT INTO sales_record (product_model, quantity, amount)VALUES ('T-800', 150, 30000.00)". Here, the string "INSERT INTO sales_record" indicates that a new data entry named "sales record" (sales_record) is added to the database table (product_model, quantity, amount). VALUES specifies the specific data values ​​to be inserted, which can be assigned sequentially according to the order of the fields within the parentheses. Subsequently, based on the constructed statement, the system can execute the corresponding write operation through the direct database operation interface, thereby accurately filling the parsed business data into the sales record table in the headquarters data center.

[0150] Secondly, refer to Figure 2 This application provides a data intelligent entry system, including:

[0151] The monitoring agent 21 is used to obtain data input information sent by instant messaging software;

[0152] The intelligent agent 22 is used to recognize and process the input data to obtain the recognized text;

[0153] Inferential agent 23 is used to perform business semantic inference based on the recognized text, combined with the real-time communication context information corresponding to the data input information and the preset business association information, to obtain initial business semantic information;

[0154] Clarification agent 24 is used to generate a clarification request corresponding to the initial business semantic information when there is information to be determined in the initial business semantic information, and send the clarification request to the sender of the data input information through instant messaging software.

[0155] The correcting agent 25 is used to correct the initial business semantic information based on the clarification information fed back by the sender in response to the clarification request, so as to obtain the target business semantic information.

[0156] The execution agent 26 is used to map the target business semantic information to the corresponding fields of the target data system and perform data writing operations in the target data system, which is the data system corresponding to the business-related information.

[0157] The sending agent 27 is used to send a success message corresponding to the data input information to the target user through instant messaging software based on the operation result of the data writing operation. The target user includes the sender, the original group to which the sender belongs, or a pre-bound user.

[0158] The data intelligent entry system of this application achieves closed-loop optimization of data processing through the division of labor and cooperation of multiple intelligent agents. In the context of data intelligent entry, an intelligent agent refers to a software module or program with specific functions that can independently or collaboratively complete data processing tasks. Specifically, the monitoring intelligent agent 21 is used to acquire data input information sent by instant messaging software, ensuring that the system can capture diverse input sources from platforms such as WeChat, WeChat Work, DingTalk, or Lark, including business data in the form of images or plain text. The recognition intelligent agent 22 is used to recognize and process the data input information to obtain the recognized text; when the input is an image, the recognition intelligent agent 22 improves the image quality through image preprocessing techniques (such as grayscale, binarization, and tilt correction) and calls optical character recognition technology to extract the text content, thereby effectively resisting image quality problems such as blurriness and uneven lighting; when the input is plain text, the text content is directly obtained as the recognized text. The inference agent 23 is used to perform business semantic inference based on the recognized text, combined with the instant messaging context information corresponding to the data input information and preset business association information, to obtain initial business semantic information. This inference agent 23 resolves ambiguities in multi-version table structures and performs semantic parsing of unstructured data by parsing context information such as the group name, sender information, and recent conversation content, and matching it with the enterprise business vocabulary and semantic association rules. When the initial business semantic information is uncertain, the clarification agent 24 generates a clarification request corresponding to the initial business semantic information and sends it to the sender of the data input information via instant messaging software, thereby proactively resolving uncertainties caused by low recognition confidence or semantic ambiguity. The correction agent 25 corrects the initial business semantic information based on the clarification information provided by the sender in response to the clarification request, to obtain target business semantic information. This correction agent 25 dynamically adjusts the accuracy of the business entity by parsing preset options or free text input from user feedback. The executing agent 26 maps the target business semantic information to the corresponding fields of the target data system and performs data writing operations in the target data system, which is the data system corresponding to the business-related information. Based on preset mapping rules, the executing agent 26 accurately fills the business entities into the corresponding fields of the enterprise resource planning system, customer relationship management system, or relational database. The sending agent 27, based on the operation result of the data writing operation, sends a success message corresponding to the data input information to the target user via instant messaging software. The target user includes the sender, the sender's original group, or a pre-bound user, thus completing the operation feedback loop.

[0159] The data intelligent entry system provided in this embodiment is used to perform the steps in the data intelligent entry method provided in the first aspect above. The principle of the data intelligent entry system provided in this embodiment is the same as that of the data intelligent entry method provided in the first aspect above, and will not be discussed in detail here.

[0160] Please refer to Figure 3 , Figure 3 This application provides a schematic diagram of the structure of an electronic device 13, comprising a processor 1301 and a memory 1302. The processor 1301 and the memory 1302 are interconnected and communicate with each other via a communication bus 1303 and / or other forms of connection mechanism (not shown). The memory 1302 stores computer-readable instructions executable by the processor 1301. When the electronic device is running, the processor 1301 executes the computer-readable instructions to perform the method in any optional implementation of the above embodiments, thereby achieving the following functions: acquiring data input information sent by instant messaging software; performing recognition processing on the data input information to obtain recognized text; and based on the recognized text, combining the instant messaging context information corresponding to the data input information and preset business association information. The process involves: performing business semantic inference to obtain initial business semantic information; generating a clarification request corresponding to the initial business semantic information if there is undetermined information in the initial business semantic information, and sending the clarification request to the sender of the data input information via instant messaging software; correcting the initial business semantic information based on the clarification information returned by the sender in response to the clarification request to obtain target business semantic information; mapping the target business semantic information to the corresponding fields of the target data system, and performing a data write operation in the target data system, which is the data system corresponding to the business-related information; and sending a successful submission message corresponding to the data input information to the target user via instant messaging software based on the result of the data write operation. The target user includes the sender, the original group to which the sender belongs, or a pre-bound user.

[0161] This application provides a computer-readable storage medium storing a computer program thereon. When the computer program is executed by a processor, it executes the method in any optional implementation of the above embodiments to achieve the following functions: acquiring data input information sent by instant messaging software; processing the data input information to obtain identified text; based on the identified text, combining the instant messaging context information corresponding to the data input information and preset business association information, performing business semantic inference to obtain initial business semantic information; if there is information to be determined in the initial business semantic information, generating a clarification request corresponding to the initial business semantic information, and sending the clarification request to the sender of the data input information through instant messaging software; correcting the initial business semantic information according to the clarification information fed back by the sender in response to the clarification request to obtain target business semantic information; mapping the target business semantic information to the corresponding field of the target data system, and performing a data writing operation in the target data system, where the target data system is the data system corresponding to the business association information; based on the operation result of the data writing operation, sending a successful submission message corresponding to the data input information to the target user through instant messaging software; wherein, the target user includes the sender, the original group to which the sender belongs, or a pre-bound user. The computer-readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0162] In the embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and there may be other division methods in actual implementation. Furthermore, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Additionally, the coupling or direct coupling or communication connection shown or discussed may be through some communication interface; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0163] Furthermore, the units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0164] Furthermore, the functional modules in the various embodiments of this application can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0165] In this document, relational terms such as first and second are used only to distinguish one entity or operation from another entity or operation, without necessarily requiring or implying any such actual relationship or order between these entities or operations.

[0166] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A data intelligent entry method, characterized in that, include: Retrieve data input information sent by instant messaging software; The input data is processed to obtain the recognized text. Based on the recognized text, combined with the real-time communication context information corresponding to the data input information and the preset business association information, business semantic inference is performed to obtain the initial business semantic information; If there is information to be determined in the initial business semantic information, a clarification request corresponding to the initial business semantic information is generated and sent to the sender of the data input information through instant messaging software. Based on the clarification information returned by the sender in response to the clarification request, the initial business semantic information is corrected to obtain the target business semantic information; Map the target business semantic information to the corresponding fields of the target data system, and perform data writing operations in the target data system, which is the data system corresponding to the business-related information; Based on the result of the data writing operation, a success message corresponding to the data input information is sent to the target user via instant messaging software; the target user includes the sender, the original group to which the sender belongs, or a pre-bound user; Based on the recognized text, combined with the real-time communication context information corresponding to the data input information and the preset business association information, the steps to perform business semantic inference and obtain the initial business semantic information include: Obtain the instant messaging context information corresponding to the data input information; Keyword extraction and topic classification are performed on instant messaging context information to obtain effective context information; The identified text is matched with preset business terms to obtain preliminary business entity identification results; By combining effective contextual information, the preliminary business entity identification results are adjusted to obtain the final business entity identification result; Based on the preset business association information, the logical relationship between business entities in the business entity identification result is analyzed to obtain the semantic association inference result; Based on the business entity identification results and semantic association inference results, initial business semantic information is obtained; After obtaining initial business semantic information by performing business semantic inference based on the recognized text, combined with the real-time communication context information corresponding to the data input information and the preset business association information, the process also includes: Based on preset confidence thresholds and business rules, the accuracy of initial business semantic information is evaluated to obtain evaluation results; When there is undetermined information in the initial business semantic information, the steps to generate a clarification request corresponding to the initial business semantic information include: If the evaluation result is pending, it is determined that there is undetermined information in the initial business semantic information; Based on the information type corresponding to the information to be determined, identify the matters to be clarified in the initial business semantic information; Based on the matters to be clarified, select the matching sentence structure from the preset clarification request sentence structures; Fill the matters to be clarified into the matching sentence structure to generate a clarification request; The steps of mapping target business semantic information to the corresponding fields of the target data system and performing data writing operations include: Based on the mapping rule information stored in the preset mapping rule library, determine the target field identifier and entity type corresponding to the target business semantic information; Identify the fields in the target data system that correspond to the target field identifier as the corresponding fields; Based on entity type, target business semantic information is mapped to corresponding fields to obtain mapped business data; The target data system calls the data operation interface to perform the write operation of the mapped business data.

2. The intelligent data entry method according to claim 1, characterized in that, After obtaining initial business semantic information by performing business semantic inference based on the recognized text, combined with the real-time communication context information corresponding to the data input information and the preset business association information, the process also includes: If the evaluation result is accurate, the initial business semantic information will be identified as the target business semantic information.

3. The intelligent data entry method according to claim 1, characterized in that, The steps for correcting the initial business semantic information based on the clarification information returned by the sender in response to the clarification request, to obtain the target business semantic information, include: Receive feedback from the sender regarding the clarification request; Based on the keywords or structured information in the feedback, the initial business semantic information is matched and updated to obtain the target business semantic information.

4. The intelligent data entry method according to claim 1, characterized in that, The steps for processing input data to obtain recognized text include: When the input data is image information, image preprocessing is performed based on the input data to obtain a preprocessed image; Optical character recognition is performed on the preprocessed image to obtain the recognized text.

5. A data intelligent entry system, characterized in that, include: A monitoring agent is used to acquire data input information sent by instant messaging software. The intelligent agent is used to identify and process input data to obtain identified text. The inferential agent is used to perform business semantic inference based on the recognized text, combined with the real-time communication context information corresponding to the data input information and the preset business association information, to obtain the initial business semantic information. The clarification agent is used to generate a clarification request corresponding to the initial business semantic information when there is information to be determined in the initial business semantic information, and to send the clarification request to the sender of the data input information through instant messaging software. The correction agent is used to correct the initial business semantic information based on the clarification information fed back by the sender in response to the clarification request, so as to obtain the target business semantic information. An execution agent is used to map target business semantic information to the corresponding fields of the target data system and perform data writing operations in the target data system, which is the data system corresponding to the business-related information. Sending intelligent agents are used to send a success message corresponding to the data input information to the target user through instant messaging software based on the operation result of the data writing operation. The target user includes the sender, the original group to which the sender belongs, or a pre-bound user. When the inference agent performs business semantic inference based on recognized text, combined with the real-time communication context information corresponding to the data input information and the preset business association information, to obtain the initial business semantic information, the specific execution is as follows: Obtain the instant messaging context information corresponding to the data input information; Keyword extraction and topic classification are performed on instant messaging context information to obtain effective context information; The identified text is matched with preset business terms to obtain preliminary business entity identification results; By combining effective contextual information, the preliminary business entity identification results are adjusted to obtain the final business entity identification result; Based on the preset business association information, the logical relationship between business entities in the business entity identification result is analyzed to obtain the semantic association inference result; Based on the business entity identification results and semantic association inference results, initial business semantic information is obtained; The inference agent, after performing business semantic inference based on recognized text, combined with the real-time communication context information corresponding to the data input information and preset business association information, obtains the initial business semantic information and then performs the following: Based on preset confidence thresholds and business rules, the accuracy of initial business semantic information is evaluated to obtain evaluation results; When the clarifying agent generates a clarification request corresponding to the initial business semantic information, given that there is information to be determined in the initial business semantic information, it specifically executes the following: If the evaluation result is pending, it is determined that there is undetermined information in the initial business semantic information; Based on the information type corresponding to the information to be determined, identify the matters to be clarified in the initial business semantic information; Based on the matters to be clarified, select the matching sentence structure from the preset clarification request sentence structures; Fill the matters to be clarified into the matching sentence structure to generate a clarification request; When the executing agent maps the target business semantic information to the corresponding fields of the target data system and performs data writing operations, it specifically executes the following: Based on the mapping rule information stored in the preset mapping rule library, determine the target field identifier and entity type corresponding to the target business semantic information; Identify the fields in the target data system that correspond to the target field identifier as the corresponding fields; Based on entity type, target business semantic information is mapped to corresponding fields to obtain mapped business data; The target data system calls the data operation interface to perform the write operation of the mapped business data.

6. An electronic device, characterized in that, It includes a processor and a memory, the memory storing computer-readable instructions, which, when executed by the processor, perform the steps of the data intelligent entry method as described in any one of claims 1-4.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it performs the steps of the data intelligent entry method as described in any one of claims 1-4.

Images