Business data distribution method and device, computer equipment and readable storage medium

Abstract

The invention discloses a business data distribution method and device, computer equipment and a readable storage medium. Distribution information associated with each data description text can be automatically determined. And accurate distribution of each piece of service data is realized according to the distribution information. The method comprises the steps of obtaining at least one piece of to-be-distributed business data, and determining a data description text corresponding to each piece of to-be-distributed business data; based on a language model, determining distribution information associated with each data description text; each piece of distribution information serves as a classification label, and the distribution labels are adopted to label the corresponding to-be-distributed business data; in the service data distribution process, the current service link is determined, target to-be-distributed service data of which the classification label is matched with the service link is acquired from at least one piece of to-be-distributed service data, and a service data packet corresponding to the service link is updated by adopting the target to-be-distributed service data, and storing the service data packet to a storage position corresponding to the service link.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a business data distribution method, device, computer equipment and readable storage medium. Background Art

[0002] With the continuous advancement of computer technology, it has become a mainstream trend to use electronic devices to upload various types of information required for financial leasing business. In the financial leasing business, customers need to provide personal information including identity cards, bank statements, etc., and for corporate customers, it is also necessary to collect various types of certificates, tax returns, financial records, annual reports, accounts receivable and other diversified information. In addition, in addition to the lessee, it is also necessary to collect relevant information of multiple roles such as guarantors, related parties, spouses, etc. These materials are in various forms, including various electronic documents such as doc texts, exls tables, pdf documents, as well as electronic photos, screenshots of financial and tax software pages, and scanned copies of bank statements. What is more complicated is that the account manager also needs to collect photos and video materials of various personnel photos, enterprise production status, product displays and office environment on site.

[0003] After the customer information is collected, it will immediately enter various process links such as credit review, which requires the account manager to accurately allocate the collected information according to the various stages of the business. However, in the existing technical methods, the account manager needs to manually allocate the collected information in hundreds of forms to each link of the business definition according to the content of the information. This manual processing method is not only time-consuming and labor-intensive, but also prone to omissions that lead to the loss of certain information, which in turn causes business interruptions, hinders the improvement of work efficiency, causes high operating costs, and seriously affects customer satisfaction. Summary of the invention

[0004] In view of this, the present application provides a business data distribution method, device, computer equipment and readable storage medium, the main purpose of which is to solve the problem that the current manual processing method is not only time-consuming and labor-intensive, but also easily leads to the loss of certain data due to omissions, thereby causing business interruptions, hindering the improvement of work efficiency, causing high operating costs, and seriously affecting customer satisfaction.

[0005] According to a first aspect of the present application, a method for allocating business information is provided, the method comprising:

[0006] Acquire at least one service profile to be allocated, and determine a profile description text corresponding to each service profile to be allocated;

[0007] Inputting the data description text corresponding to each of the to-be-allocated business data into a pre-trained language model, and determining the allocation information associated with each of the data description texts based on the language model, wherein the allocation information is used to indicate the business link associated with the to-be-allocated business data corresponding to the data description text;

[0008] Taking each of the allocation information as a classification label, and using the allocation label to mark the corresponding business data to be allocated;

[0009] During the process of business data allocation, the current business link is determined, and the target business data to be allocated whose classification label matches the business link is obtained from at least one of the business data to be allocated, and the business data package corresponding to the business link is updated using the target business data to be allocated, and the business data package is stored in the storage location corresponding to the business link.

[0010] According to a second aspect of the present application, a service data distribution device is provided, the device comprising:

[0011] An acquisition module, used to acquire at least one service profile to be allocated, and determine a profile description text corresponding to each service profile to be allocated;

[0012] A determination module, used for inputting a data description text corresponding to each of the to-be-allocated business data into a pre-trained language model, and determining allocation information associated with each of the data description texts based on the language model, wherein the allocation information is used for indicating a business link associated with the to-be-allocated business data corresponding to the data description text;

[0013] A setting module, used to use each of the allocation information as a classification label, and use the allocation label to mark the corresponding business data to be allocated;

[0014] The allocation module is used to determine the current business link during the process of business data allocation, obtain the target business data to be allocated whose classification label matches the business link from the at least one business data to be allocated, and use the target business data to be allocated to update the business data package corresponding to the business link, and store the business data package in the storage location corresponding to the business link.

[0015] According to a third aspect of the present application, a computer device is provided, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of any one of the methods described in the first aspect when executing the computer program.

[0016] According to a fourth aspect of the present application, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of any one of the methods described in the first aspect are implemented.

[0017] By means of the above technical scheme, the present application provides a method, device, computer equipment and readable storage medium for allocating business data. The present application first obtains at least one business data to be allocated, and determines the data description text corresponding to each business data to be allocated. The data description text corresponding to each business data to be allocated is input into a pre-trained language model. Further, based on the language model, the allocation information associated with each data description text is determined, wherein the allocation information is used to indicate the business link associated with the business data to be allocated corresponding to the data description text. Each allocation information is used as a classification label, and the corresponding business data to be allocated is marked with the allocation label. Finally, in the process of allocating business data, the current business link is determined, and the target business data to be allocated whose classification label matches the business link is obtained in at least one business data to be allocated, and the target business data to be allocated is used to update the business data package corresponding to the business link, and the business data package is stored in the storage location corresponding to the business link. In an embodiment of the present application, the data description text corresponding to each business data to be allocated is analyzed by the language model, and the allocation information associated with each data description text is automatically determined. Then, accurate allocation of each business data is achieved according to the allocation information. In addition, allocating at least one to-be-allocated business data by allocating the allocation information can ensure that each business link obtains the correct data, thus avoiding data mismatch or omission.

[0018] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Various other advantages and benefits will become apparent to those of ordinary skill in the art by reading the detailed description of the preferred embodiments below. The accompanying drawings are only for the purpose of illustrating the preferred embodiments and are not to be considered as limiting the present application. Also, the same reference symbols are used throughout the accompanying drawings to represent the same components. In the accompanying drawings:

[0020] Figure 1 A schematic diagram of a business data distribution method provided in an embodiment of the present application is shown;

[0021] Figure 2 A schematic diagram of a business data distribution method provided in an embodiment of the present application is shown;

[0022] Figure 3 A schematic diagram of a business data distribution method provided in an embodiment of the present application is shown;

[0023] Figure 4 A schematic diagram of a business data distribution method provided in an embodiment of the present application is shown;

[0024] Figure 5 A schematic diagram of a business data distribution method provided in an embodiment of the present application is shown;

[0025] Figure 6 A schematic diagram of the structure of a business data distribution device provided in an embodiment of the present application is shown;

[0026] Figure 7 A schematic diagram of the device structure of a computer device provided in an embodiment of the present application is shown. DETAILED DESCRIPTION

[0027] The exemplary embodiments of the present application will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present application are shown in the accompanying drawings, it should be understood that the present application can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided in order to enable a more thorough understanding of the present application and to fully convey the scope of the present application to those skilled in the art.

[0028] With the continuous advancement of computer technology, it has become a mainstream trend to use electronic devices to upload various types of information required for financial leasing business. In the financial leasing business, customers need to provide personal information including ID cards, bank statements, etc., and for corporate customers, they also need to collect various types of certificates, tax returns, financial records, annual reports, accounts receivable and other diversified information. In addition, in addition to the lessee, it is also necessary to collect relevant information of multiple roles such as guarantors, related parties, spouses, etc. These materials are in various forms, including various electronic documents such as doc texts, exls tables, pdf documents, as well as electronic photos, screenshots of financial and tax software pages, and scanned copies of bank statements. What is more complicated is that the account manager also needs to collect photos and video materials of various personnel, enterprise production status, product display and office environment on the enterprise site. At present, after the customer information is collected, it immediately enters various process links such as credit review, which requires the account manager to accurately allocate the collected information according to the various stages of the business. However, in the existing technical methods, the account manager needs to manually accurately allocate the collected information in hundreds of forms to each link of the business definition according to the content of the information. This manual processing method is not only time-consuming and labor-intensive, but also prone to the loss of certain information due to omissions, which in turn causes business interruptions, hinders the improvement of work efficiency, causes high operating costs, and seriously affects customer satisfaction. The business information distribution method provided in the embodiment of the present invention is applicable to a business information distribution system. Relevant technicians can access the business information distribution system based on the visual interface of the terminal device and upload at least one business information to be distributed. The terminal device can be, but is not limited to, various personal computers, laptops, smart phones, and tablet computers.

[0029] See also Figure 1 As shown, Figure 1 A flow chart of a method for distributing business information provided by an embodiment of the present invention includes the following steps:

[0030] S10. Acquire at least one service profile to be allocated, and determine a profile description text corresponding to each service profile to be allocated.

[0031] In the embodiments of the present application, Figure 2 As shown, in order to ensure the smooth progress of the business process, the business data allocation system will first identify the data type of the business data to be allocated. Figure 3As shown, according to the data type, the method of generating the data description text is determined. When the data type is a document data type, the following step S12 is executed to determine the data description text corresponding to the business data to be allocated according to the document content. When the data type is a non-document data type, the following step S13 is executed to determine the data description text corresponding to the business data to be allocated according to the multimodal model LMM. In the actual operation process, at least one business data to be allocated uploaded by the user includes various files, reports, forms or other related documents. After obtaining these materials, the business data allocation system will analyze each business data to be allocated to determine the data description text corresponding to each business data to be allocated, wherein the data description text is used to explain the main content, purpose, scope of application of the data and its specific role in the business process. For example, if the business data to be allocated is a market analysis report, the data description text should include the subject of the report, the main market analyzed, the data source used, the analysis method, and the main conclusions and suggestions of the report. In addition, the data description text can also include information such as the creation date, author, version number, etc. of the data to facilitate tracking and management. In this way, not only is the orderly distribution of information ensured, but the retrievability and availability of information are also improved, providing strong support for the entire business team.

[0032] S11. For each piece of business data to be allocated, identify the data type of the business data to be allocated.

[0033] In the actual operation process, the types of business data to be allocated include various electronic documents such as doc text, exls tables, pdf documents, electronic photos, photos or screenshots of various financial and tax software pages, bank statements and scans of various banks, and various personnel group photos, enterprise production status, products, and office photos and video materials collected by the account manager on the enterprise site. In the embodiment of the present application, various electronic documents such as doc text, exls tables, pdf documents, etc. are set as document data types, and other data types are set as non-document types. Further, for non-document types, electronic photos, photos or screenshots of various financial and tax software pages, bank statements and scans of various banks, etc., picture data types and video data types are set as the first category of data types in the non-document data types. The group photos of various personnel, enterprise production status, products, and office photos and video materials collected by the account manager on the enterprise site are set as the second category of data types in the non-document data types, wherein the second category of data types carries business scenario description text.

[0034] In this step, the business information distribution system can pre-set corresponding file suffixes for document information types and non-document information types. After receiving the business information to be distributed, the suffix of the business information to be distributed is identified, the suffix is ​​compared with the preset suffix, and the preset suffix that matches the suffix is ​​determined. According to the information type associated with the preset suffix, it is determined whether the business information to be distributed is a document information type or a non-document information type. Next, if the business information to be distributed is identified as a non-document information type after preliminary screening, it is identified whether there is a business scenario description text in the information to be distributed. If there is no business scenario description text, it is determined that the business information to be distributed is a first category information type. If there is a business scenario description text, it is determined that the business information to be distributed is a second category information type.

[0035] S12. When the data type of the business data to be allocated is a document data type, the business data to be allocated is identified based on natural language processing technology, the plain text content in the business data to be allocated is extracted, and the format and noise information in the plain text content are removed to obtain a data description text corresponding to the business data to be allocated.

[0036] In this step, when the data type of the business data to be allocated is a document data type, the system will first read the file of the document data type based on natural language processing technology, such as DOC, DOCX, XLS, XLSX, PDF, etc., and convert its content into a plain text format. Furthermore, considering that the extracted text may contain a large amount of format information (such as tags, special characters, blanks, etc.) and noise, regular expressions, HTML / XML parsers, etc. will be used to clean the text to obtain a purer text content. After the above cleaning and processing, the business data allocation system will obtain a clean and structured text description. This text will serve as the basis for subsequent classification and application to ensure that the system can efficiently and accurately allocate business data.

[0037] S13: When the data type of the business data to be allocated is a non-document data type, determine a data description text corresponding to the business data to be allocated based on a multimodal model.

[0038] In this step, if Figure 4As shown, when the data type of the business data to be allocated is a non-document data type, it is impossible to directly obtain the data description text of the business data to be allocated. It is necessary to use the multimodal model LMM to identify the business data to be allocated, and generate a data description text for describing the content of the business data to be allocated based on the multimodal model. The multimodal model in the embodiment of the present application is a model that can simultaneously process and analyze multiple types of data. It can simultaneously process data of multiple modes such as images and texts. For image data, the model can extract visual features such as key points, edges, and color distribution for analysis. For video data, the model can combine the features of images and audio for comprehensive analysis, and then generate a data description text corresponding to the business data to be allocated. When the business data to be allocated is a non-document data type, the data description text is used to summarize the main content and characteristics of the data, and the generated description text will be used as a summary of the data for subsequent business data allocation. The specific process of generating the data description text is described in steps S131 and S132:

[0039] S131. When the data type of the business data to be allocated is a first category data type, determine a data description text corresponding to the business data to be allocated based on a multimodal model.

[0040] Specifically, when the data type of the business data to be allocated is the first category data type, the main content corresponding to the business data to be allocated can be identified by relying on the general recognition ability of the multimodal model, without relying on a specific business background for understanding. Therefore, when the data type of the business data to be allocated is the first category data type, the business data to be allocated can be directly extracted based on the multimodal model to obtain an image feature vector, and the data description text is determined according to the image feature vector. Specifically, based on the multimodal model, the business data to be allocated is extracted to obtain an image feature vector, which can effectively capture the key information and features in the image and provide a basis for subsequent processing. Based on the multimodal model, multiple preset data description texts are feature extracted to obtain a text feature vector corresponding to each preset data description text, and each text feature vector can represent the core content and semantic information of its corresponding text. On this basis, the vector similarity between the image feature vector and each text feature vector is calculated to evaluate the matching degree between the image content and each text description. Determine the text feature vector whose vector similarity is higher than the other vector similarities, and use the preset data description text corresponding to the text feature vector as the data description text corresponding to the business data to be allocated. For example, the preset data description texts include "This is a multifunctional small mechanical tool suitable for home and small business use", "Advanced CNC machine tools, suitable for high-precision processing needs", and "This equipment is a light engineering machinery suitable for short-term construction projects". Use the pre-trained multimodal model to extract the image features of the equipment image and the text features of each preset data description text. Use the cosine similarity formula to calculate the similarity between each text feature vector and the image feature vector. Find the highest value from the calculated similarity and correspond it to the corresponding text description, such as "Advanced CNC machine tools, suitable for high-precision processing needs". Use this preset data description text as the data description text corresponding to the business data to be assigned.

[0041] S132: When the data type of the business data to be allocated is a data type of the second category, determine a data description text corresponding to the business data to be allocated based on the multimodal model.

[0042] Specifically, when the data type of the business data to be assigned is the second category data type, the content of these data has obvious business scenario characteristics (such as group photos of personnel, factory equipment, etc.), and requires stronger business understanding and background knowledge to accurately classify. Therefore, it is necessary to obtain the business scenario description text associated with the business data to be assigned. The business scenario description text elaborates on the specific content and background of the business scenario in detail, providing important context information for subsequent processing steps. Based on the multimodal model, feature extraction is performed on the business data to be assigned and the business scenario description text respectively to obtain image feature vectors and text feature vectors. Based on the multimodal model, cross-attention calculation is performed on the image feature vector and the text feature vector. Cross-attention calculation is a method that can measure the degree of correlation between different modal features. Through this calculation, the model can obtain attention weights, which reflect the degree of mutual influence between the image feature vector and the text feature vector. Based on the obtained attention weights, the multimodal model will fuse the text feature vector and the image feature vector to generate a fused feature vector. Based on the multimodal model, feature extraction is performed on multiple preset data description texts to obtain the text feature vector corresponding to each preset data description text, and the vector similarity between the fused feature vector and each text feature vector is calculated to determine the text feature vector whose vector similarity is higher than other vector similarities, and the preset data description text corresponding to the text feature vector is used as the data description text corresponding to the business data to be allocated. The embodiment of the present application generates a fused feature vector so that the multimodal model can comprehensively consider image and text information, providing a more comprehensive basis for data allocation. In practical applications, this method can be widely used in multiple fields such as financial leasing services, medical diagnosis, and legal document processing. For example, in the field of financial leasing services, various types of certificates, taxes, finance, annual reports, and accounts receivable submitted by customers waiting for allocation of business data can be matched with the preset business description text, so as to quickly and accurately determine the data description text corresponding to the business data to be allocated.

[0043] S20. Input the data description text corresponding to each business data to be allocated into a pre-trained language model, and determine the allocation information associated with each data description text based on the language model, wherein the allocation information is used to indicate the business link associated with the business data to be allocated corresponding to the data description text.

[0044] In an embodiment of the present application, the language model is an LLM language model. Before using the language model to recognize the data description text, the language model needs to be pre-trained. Specifically, first collect a certain amount of business data samples and their corresponding allocation information, and then use these samples to optimize the pre-trained language model. During the optimization process, supervised learning can be used to allow the model to learn the mapping relationship between different data description texts and business links, so that the language model can determine the allocation information associated with the data description text. It should be noted that the allocation information is used to indicate the business link associated with the business data to be allocated corresponding to the data description text. After the training is completed, such as Figure 5 As shown, the trained language model is used to identify the data description text corresponding to each business data to be allocated, and the allocation information associated with each data description text is determined.

[0045] S21. Inputting a data description text and a plurality of preset allocation information into a language model, wherein each preset allocation information includes a plurality of classification keywords.

[0046] S22. Based on the language model, determine the word embedding vector corresponding to the data description text, and determine the keyword embedding vector of each classification keyword, and calculate the similarity between the word embedding vector and each keyword embedding vector, and associate each similarity with the corresponding keyword embedding vector.

[0047] S23. Based on the language model, determine multiple keyword embedding vectors corresponding to each preset allocation information, and determine the similarity corresponding to the preset allocation information according to multiple similarities associated with the multiple keyword embedding vectors, and determine the preset allocation information with a similarity higher than other similarities, and associate the preset allocation information with the material description text.

[0048] In steps S21-S23, the preset allocation information is a statement pre-stored by relevant technical personnel to describe the business links and material summaries, for example, "Link A-on-site due diligence operation photos". It should be noted that the preset allocation information includes multiple allocation keywords. Taking "Link A-on-site due diligence operation photos" as an example, the allocation keywords can be "financial leasing, equipment, site, operation". In the actual operation process, for example, the material description text is "a photo of a financial leasing project site, with large production equipment, a running excavator, and several workers operating the equipment". The LLM model maps the material description text to a vector representation, namely a word embedding vector. The model maps the classification keywords "financial leasing, equipment, site, operation" in the preset allocation information to a vector representation to obtain a keyword embedding vector. By calculating the similarity between the word embedding vector and each keyword embedding vector "financial leasing, equipment, site, operation", each similarity is associated with the corresponding keyword embedding vector. Finally, the LLM model combines the similarity of the associated embedding vectors of each keyword "financial leasing, equipment, site, and operation" to determine the similarity between "Link A-site due diligence operation photos" and the document description text. At the same time, the LLM model determines the similarity between other preset allocation information and the document description text. If the similarity corresponding to the preset allocation information "Link A-site due diligence operation photos" is higher than the similarity corresponding to other preset allocation information, the preset allocation information "Link A-site due diligence operation photos" is associated with the document description text. In summary, by combining pre-trained language models and machine learning algorithms, the problem of business document allocation can be effectively solved, work efficiency can be improved, and the waste of human resources can be reduced.

[0049] S30: taking each piece of allocation information as a classification label, and using the allocation label to mark the corresponding business data to be allocated.

[0050] Specifically, the specific content of each assigned information is determined and converted into a classification label. The classification label can accurately reflect the core meaning of the assigned information, so as to facilitate the subsequent business data labeling and retrieval. Furthermore, in order to ensure that each business data can match its corresponding classification label, so as to facilitate the subsequent classification management and data analysis, the business data allocation system will apply these labels to the corresponding business data to be allocated. Specifically, this can be achieved by adding label fields to the business data, using color coding or other visual markings. In addition, in order to ensure the accuracy of the labeling, the business data allocation system can also set up a quality inspection link. Specifically, it can verify whether the label is correctly applied to each data, and check whether there are any missing or incorrect labels. Through this step, the consistency and accuracy of the data can be ensured, providing strong support for subsequent data analysis and business decisions. After the labeling is completed, the updated business data can be archived in the corresponding database or file system in a timely manner to ensure the integrity and traceability of the information, and also facilitate the subsequent data retrieval and analysis.

[0051] S40. During the process of business data allocation, determine the current business link, obtain target business data to be allocated whose classification label matches the business link from at least one business data to be allocated, and use the target business data to be allocated to update the business data package corresponding to the business link, and store the business data package in the storage location corresponding to the business link.

[0052] In the embodiment of the present application, during the process of business data allocation, the current business link is determined. The business link can be divided according to multiple stages of the business process, such as project establishment, due diligence, contract signing, execution management, etc. Next, the target business data to be allocated whose classification label matches the business link is obtained from at least one business data to be allocated, and the business data package corresponding to the business link is updated using the target business data to be allocated. For example, if the current business link is "project establishment", the classification labels corresponding to all business data to be allocated are read, and the target classification label of "project establishment" is determined in the classification label content, and the business data to be allocated corresponding to the target classification label is used as the target business data to be allocated, and the business data package corresponding to the business link is updated using the target business data to be allocated. Finally, the business data package is stored in the storage location corresponding to the business link, completing the allocation process of the business data to be allocated. In addition, after the business data package is stored in the storage location of the corresponding business link, the business data allocation system can obtain the business data list corresponding to each business link, and use the list to check the business data package to determine whether there is any missing data. Specifically, multiple target data identifiers in the business data list are identified, and the corresponding data identifier of each data to be allocated in the business data package is determined. Then, each data identifier to be allocated is matched and identified in turn using the target data identifier. If a data identifier to be allocated that matches the target data identifier is found, the next target data identifier is obtained for identification. If no matching data identifier to be allocated is found, a test result is generated, indicating that there is data omission, and the result is associated with the target data identifier. Further, when the business data allocation system detects that there is a test result indicating that data is missing, it determines the data identifier associated with the test result, and generates and displays a prompt message of data omission based on it, and then sends the prompt message to the terminal device of the relevant technician to prompt the relevant technician that there is data omission and which data is missing in the business link. In this way, the data integrity of each business link can be ensured to avoid business delays or errors caused by missing data. The automation of the verification process improves work efficiency and reduces the tediousness and error probability of manual verification. In addition, when missing data is detected, the system can quickly locate the problem and provide clear prompt information, which helps business personnel to supplement the missing data in a timely manner and ensure the smooth progress of the business process.

[0053] The method provided in the embodiment of the present application first obtains at least one business data to be allocated, and determines the data description text corresponding to each business data to be allocated. The data description text corresponding to each business data to be allocated is input into the pre-trained language model. Further, based on the language model, the allocation information associated with each data description text is determined, wherein the allocation information is used to indicate the business link associated with the business data to be allocated corresponding to the data description text. Each allocation information is used as a classification label, and the corresponding business data to be allocated is marked with the allocation label. Finally, in the process of business data allocation, the current business link is determined, and the target business data to be allocated whose classification label matches the business link is obtained in at least one business data to be allocated, and the business data package corresponding to the business link is updated with the target business data to be allocated, and the business data package is stored in the storage location corresponding to the business link. In the embodiment of the present application, the data description text corresponding to each business data to be allocated is analyzed by the language model, and the allocation information associated with each data description text is automatically determined. Then, accurate allocation of each business data is realized according to the allocation information. In addition, allocating the allocation information to at least one business data to be allocated can ensure that each business link obtains the correct data and avoids data mismatch or omission.

[0054] Further, as Figure 1 In a specific implementation of the method, the present application provides a business data distribution device, such as Figure 6 As shown, the device includes: an acquisition module 601, a determination module 602, a setting module 603, and an allocation module 604.

[0055] The acquisition module 601 is used to acquire at least one service profile to be allocated and determine a profile description text corresponding to each service profile to be allocated;

[0056] The determination module 602 is used to input the data description text corresponding to each of the to-be-allocated business data into a pre-trained language model, and determine the allocation information associated with each of the data description texts based on the language model, wherein the allocation information is used to indicate the business link associated with the to-be-allocated business data corresponding to the data description text;

[0057] The setting module 603 is used to use each of the allocation information as a classification label, and use the allocation label to mark the corresponding business data to be allocated;

[0058] The allocation module 604 is used to determine the current business link during the process of business data allocation, obtain the target business data to be allocated whose classification label matches the business link from the at least one business data to be allocated, and use the target business data to be allocated to update the business data package corresponding to the business link, and store the business data package in the storage location corresponding to the business link.

[0059] In a specific application scenario, the acquisition module 601 is used to identify the material type of each of the business data to be allocated, the material type including a document material type and a non-document material type, the non-document material type including a first category material type and a second category material type, wherein the first category material type is a picture material type and a video material type, and the second category material type is a picture material type with a business scenario description text and a video material type with a business scenario description text; when the material type of the business data to be allocated is the document material type, the business data to be allocated is identified based on natural language processing technology, the plain text content in the business data to be allocated is extracted, and the format and noise information in the plain text content are removed to obtain the material description text corresponding to the business data to be allocated; when the material type of the business data to be allocated is the non-document material type, the material description text corresponding to the business data to be allocated is determined based on a multimodal model.

[0060] In a specific application scenario, the acquisition module 601 is used to perform feature extraction on the business data to be assigned based on the multimodal model to obtain an image feature vector when the data type of the business data to be assigned is the first category data type; perform feature extraction on multiple preset data description texts based on the multimodal model to obtain a text feature vector corresponding to each preset data description text, and calculate the vector similarity between the image feature vector and each of the text feature vectors, determine a text feature vector whose vector similarity is higher than other vector similarities, and use the preset data description text corresponding to the text feature vector as the data description text corresponding to the business data to be assigned.

[0061] In a specific application scenario, the acquisition module 601 is used to obtain the business scenario description text associated with the business data to be allocated when the data type of the business data to be allocated is the second category data type; based on the multimodal model, feature extraction is performed on the business data to be allocated and the business scenario description text respectively to obtain an image feature vector and a text feature vector; based on the multimodal model, cross-attention calculation is performed on the image feature vector and the text feature vector to obtain an attention weight, and the text feature vector and the image feature vector are fused according to the attention weight to obtain a fused feature vector; based on the multimodal model, feature extraction is performed on multiple preset data description texts to obtain a text feature vector corresponding to each preset data description text, and the vector similarity between the fused feature vector and each of the text feature vectors is calculated, and the text feature vector having a vector similarity higher than other vector similarities is determined, and the preset data description text corresponding to the text feature vector is used as the data description text corresponding to the business data to be allocated.

[0062] In a specific application scenario, the determination module 602 is used to input the material description text and multiple preset allocation information into the language model, each of the preset allocation information includes multiple classification keywords; based on the language model, determine the word embedding vector corresponding to the material description text, and determine the keyword embedding vector of each classification keyword, and calculate the similarity between the word embedding vector and each keyword embedding vector, and associate each similarity with the corresponding keyword embedding vector; based on the language model, determine the multiple keyword embedding vectors corresponding to each of the preset allocation information, and determine the similarity corresponding to the preset allocation information based on the multiple similarities associated with the multiple keyword embedding vectors, and determine the preset allocation information with a similarity higher than other similarities, and associate the preset allocation information with the material description text.

[0063] In a specific application scenario, the device further includes: a detection module 605 and a generation module 606 .

[0064] The detection module 605 is used to obtain a business data list corresponding to each business link, and use the business data list to detect the business data package corresponding to the business link;

[0065] The generating module 606 is used to determine the data identification associated with the detection result if the detection result indicates that there is data missing, and to generate and display data missing prompt information according to the data identification.

[0066] In a specific application scenario, the detection module 605 is used to determine multiple target data identifiers in the business data list, and determine the designated data identifier of each business data to be allocated in the business data package; use the target data identifier to identify each designated data identifier one by one; if a designated data identifier matching the target data identifier is identified, obtain the next target data identifier to identify each designated data identifier; if a designated data identifier matching the target data identifier is not identified, generate a detection result indicating that data is missing, and associate the target data identifier with the detection result.

[0067] The device provided in the embodiment of the present application first obtains at least one business data to be allocated, and determines the data description text corresponding to each business data to be allocated. The data description text corresponding to each business data to be allocated is input into the pre-trained language model. Further, based on the language model, the allocation information associated with each data description text is determined, wherein the allocation information is used to indicate the business link associated with the business data to be allocated corresponding to the data description text. Each allocation information is used as a classification label, and the corresponding business data to be allocated is marked with the allocation label. Finally, in the process of business data allocation, the current business link is determined, and the target business data to be allocated whose classification label matches the business link is obtained in at least one business data to be allocated, and the business data package corresponding to the business link is updated with the target business data to be allocated, and the business data package is stored in the storage location corresponding to the business link. In the embodiment of the present application, the data description text corresponding to each business data to be allocated is analyzed by the language model, and the allocation information associated with each data description text is automatically determined. Then, accurate allocation of each business data is realized according to the allocation information. In addition, allocating the allocation information to at least one business data to be allocated can ensure that each business link obtains the correct data and avoids data mismatch or omission.

[0068] It should be noted that for other corresponding descriptions of the functional units involved in the business data distribution device provided in the embodiment of the present application, reference can be made to Figures 1 to 5 The corresponding description in will not be repeated here.

[0069] In an exemplary embodiment, see Figure 7 , and also provides a device, which includes a communication bus, a processor, a memory and a communication interface, and may also include an input and output interface and a display device, wherein each functional unit can communicate with each other through the bus. The memory stores a computer program, and the processor is used to execute the program stored in the memory and execute the business data distribution method in the above embodiment.

[0070] A computer-readable storage medium stores a computer program, which implements the steps of the business data distribution method when executed by a processor.

[0071] Through the description of the above implementation methods, those skilled in the art can clearly understand that the present application can be implemented by hardware, or by means of software plus a necessary general hardware platform. Based on such an understanding, the technical solution of the present application can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.), including a number of instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each implementation scenario of the present application.

[0072] Those skilled in the art will appreciate that the accompanying drawings are merely schematic diagrams of a preferred implementation scenario, and the modules or processes in the accompanying drawings are not necessarily required for implementing the present application.

[0073] Those skilled in the art will appreciate that the modules in the device in the implementation scenario can be distributed in the device in the implementation scenario according to the implementation scenario description, or can be changed accordingly and located in one or more devices different from the implementation scenario. The modules in the above implementation scenario can be combined into one module, or can be further split into multiple sub-modules.

[0074] The above application serial numbers are for description only and do not represent the advantages or disadvantages of the implementation scenarios.

[0075] The above disclosure only discloses several specific implementation scenarios of the present application. However, the present application is not limited thereto, and any changes that can be conceived by technicians in this field should fall within the scope of protection of the present application.

Claims

1. A method for distributing business information, characterized in that: include: Acquire at least one service profile to be allocated, and determine a profile description text corresponding to each service profile to be allocated; Inputting the data description text corresponding to each of the to-be-allocated business data into a pre-trained language model, and determining the allocation information associated with each of the data description texts based on the language model, wherein the allocation information is used to indicate the business link associated with the to-be-allocated business data corresponding to the data description text; Taking each of the allocation information as a classification label, and using the allocation label to mark the corresponding business data to be allocated; During the process of business data allocation, the current business link is determined, and the target business data to be allocated whose classification label matches the business link is obtained from at least one of the business data to be allocated, and the business data package corresponding to the business link is updated using the target business data to be allocated, and the business data package is stored in the storage location corresponding to the business link.

2. The method according to claim 1, characterized in that The step of determining the data description text corresponding to each of the service data to be allocated includes: For each of the to-be-allocated business data, identifying the data type of the to-be-allocated business data, the data type includes a document data type and a non-document data type, the non-document data type includes a first category data type and a second category data type, wherein the first category data type is a picture data type and a video data type, and the second category data type is a picture data type with a business scenario description text and a video data type with a business scenario description text; When the data type of the business data to be allocated is the document data type, the business data to be allocated is identified based on natural language processing technology, the plain text content in the business data to be allocated is extracted, and the format and noise information in the plain text content are removed to obtain a data description text corresponding to the business data to be allocated; When the material type of the business material to be allocated is the non-document material type, a material description text corresponding to the business material to be allocated is determined based on a multimodal model.

3. The method according to claim 2, characterized in that The determining, based on the multimodal model, a data description text corresponding to the to-be-allocated business data includes: When the data type of the business data to be allocated is the first category data type, extracting features of the business data to be allocated based on the multimodal model to obtain an image feature vector; Based on the multimodal model, feature extraction is performed on multiple preset data description texts to obtain a text feature vector corresponding to each preset data description text, and the vector similarity between the image feature vector and each of the text feature vectors is calculated to determine a text feature vector whose vector similarity is higher than other vector similarities, and the preset data description text corresponding to the text feature vector is used as the data description text corresponding to the business data to be assigned.

4. The method according to claim 2, characterized in that: The determining, based on the multimodal model, a data description text corresponding to the to-be-allocated business data includes: When the data type of the business data to be allocated is the second category data type, obtaining a business scenario description text associated with the business data to be allocated; Based on the multimodal model, feature extraction is performed on the business information to be assigned and the business scenario description text to obtain an image feature vector and a text feature vector; Performing cross-attention calculation on the image feature vector and the text feature vector based on the multimodal model to obtain an attention weight, and fusing the text feature vector and the image feature vector according to the attention weight to obtain a fused feature vector; Based on the multimodal model, feature extraction is performed on multiple preset data description texts to obtain a text feature vector corresponding to each preset data description text, and the vector similarity between the fused feature vector and each of the text feature vectors is calculated to determine a text feature vector whose vector similarity is higher than other vector similarities, and the preset data description text corresponding to the text feature vector is used as the data description text corresponding to the business data to be assigned.

5. The method according to claim 1, characterized in that The step of inputting the data description text corresponding to each of the to-be-allocated business data into a pre-trained language model, and determining the allocation information associated with each of the data description text based on the language model, comprises: Inputting the material description text and a plurality of preset allocation information into the language model, each of the preset allocation information comprising a plurality of classification keywords; Based on the language model, determine the word embedding vector corresponding to the document description text, and determine the keyword embedding vector of each classification keyword, and calculate the similarity between the word embedding vector and each keyword embedding vector, and associate each similarity with the corresponding keyword embedding vector; Based on the language model, multiple keyword embedding vectors corresponding to each of the preset assignment information are determined, and based on multiple similarities associated with the multiple keyword embedding vectors, the similarity corresponding to the preset assignment information is determined, and the preset assignment information with a similarity higher than other similarities is determined, and the preset assignment information is associated with the material description text.

6. The method according to claim 1, characterized in that After storing the business data package in the storage location corresponding to the business link, the method further includes: Obtaining a business data list corresponding to each of the business links, and using the business data list to detect the business data package corresponding to the business link; If the detection result indicates that data is missing, a data identifier associated with the detection result is determined, and data missing prompt information is generated and displayed according to the data identifier.

7. The method according to claim 6, characterized in that The detecting of the business data package corresponding to the business link by using the business data list includes: Determine a plurality of target data identifiers in the business data list, and determine a designated data identifier of each business data to be allocated in the business data package; Using the target data identifier one by one to identify each designated data identifier; If a designated material identifier matching the target material identifier is identified, the next target material identifier is obtained to identify each designated material identifier; If the designated material identifier that matches the target material identifier is not identified, a detection result indicating that there is a missing material is generated, and the target material identifier is associated with the detection result.

8. A business data distribution device, characterized in that: include: An acquisition module, used to acquire at least one service profile to be allocated, and determine a profile description text corresponding to each service profile to be allocated; A determination module, used for inputting a data description text corresponding to each of the to-be-allocated business data into a pre-trained language model, and determining allocation information associated with each of the data description texts based on the language model, wherein the allocation information is used for indicating a business link associated with the to-be-allocated business data corresponding to the data description text; A setting module, used to use each of the allocation information as a classification label, and use the allocation label to mark the corresponding business data to be allocated; The allocation module is used to determine the current business link during the process of business data allocation, obtain the target business data to be allocated whose classification label matches the business link from the at least one business data to be allocated, and use the target business data to be allocated to update the business data package corresponding to the business link, and store the business data package in the storage location corresponding to the business link.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images