Business object management method and device, equipment and readable medium

Abstract

The embodiment of the invention discloses a business object management method and device, equipment and a readable medium. The method comprises the following steps: acquiring transaction remark texts of business transactions generated in a business object business process; obtaining the resource quantity of the virtual resources transferred in the execution process of each business transaction; screening out a target transaction remark text from the plurality of transaction remark texts based on the resource quantity corresponding to the business transaction of each transaction remark text explanation; and performing type identification processing on the target transaction remark text to obtain a business type to which the business object operation business belongs. According to the technical scheme provided by the embodiment of the invention, the accurate service type can be obtained.

Description

Technical Field

[0001] The present application relates to the field of computer and communication technology, and in particular to a business object management method, a data push device, an electronic device, a computer-readable storage medium, and a computer program product. Background Art

[0002] With the advancement of online communication technology, transactions are no longer limited to physical locations. More and more businesses are being conducted online and conducting transactions with other parties related to these businesses. Online platforms and applications must clearly identify the business type of each business entity to prevent illicit activities. Currently, the primary method for obtaining the business type of a business entity is to fill in the form themselves. However, this method can result in the business type being left blank or incorrectly entered. Therefore, accurately identifying the business type is a pressing issue. Summary of the Invention

[0003] The embodiments of the present application provide a business object management method, a business object management device, an electronic device, a computer-readable storage medium, and a computer program product, which can obtain accurate business types.

[0004] Other features and advantages of the present application will become apparent from the following detailed description, or may be learned in part by practice of the present application.

[0005] According to one aspect of an embodiment of the present application, a business object management method is provided, the method comprising:

[0006] Obtaining transaction notes for each business transaction generated during the business operation of the business object, where the transaction notes for each business transaction are used to explain the business transaction;

[0007] Acquiring the amount of virtual resources transferred during the execution of each business transaction;

[0008] Based on the resource amounts corresponding to the business transactions explained in each transaction memo text, a target transaction memo text is selected from the multiple transaction memo texts; wherein the resource amount corresponding to the target transaction memo text is greater than the resource amounts corresponding to the other transaction memo texts;

[0009] The target transaction remark text is subjected to type identification processing to obtain the business type to which the business object's operating business belongs.

[0010] According to one aspect of an embodiment of the present application, a business object management device is provided, the device including an acquisition unit, a screening unit, and a processing unit, wherein:

[0011] The acquisition unit is used to acquire transaction remark text of each business transaction generated during the business operation of the business object, wherein the transaction remark text of each business transaction is used to explain the each business transaction;

[0012] The acquisition unit is further configured to acquire the amount of virtual resources transferred during the execution of each business transaction;

[0013] The screening unit is configured to screen out a target transaction memo text from the plurality of transaction memo texts based on the resource amounts corresponding to the business transactions explained in the respective transaction memo texts; wherein the resource amount corresponding to the target transaction memo text is greater than the resource amounts corresponding to the other transaction memo texts;

[0014] The processing unit is used to perform type identification processing on the target transaction note text to obtain the business type to which the business object's operating business belongs.

[0015] According to one aspect of an embodiment of the present application, an embodiment of the present application provides an electronic device, comprising one or more processors; a storage device for storing one or more computer programs, which, when executed by the one or more processors, enables the electronic device to implement the business object management method as described above.

[0016] According to one aspect of an embodiment of the present application, an embodiment of the present application provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor of an electronic device, the electronic device executes the business object management method as described above.

[0017] According to one aspect of an embodiment of the present application, an embodiment of the present application provides a computer program product, including a computer program, which is stored in a computer-readable storage medium. A processor of an electronic device reads and executes the computer program from the computer-readable storage medium, so that the electronic device performs the business object management method as described above.

[0018] In the technical solution provided in the embodiment of the present application, by performing type identification processing on the target transaction note text, it is possible to classify and predict a business type that is more consistent with the actual business operations of the business object based on the actual business operations of the business object reflected in the target transaction note text. At the same time, the target transaction note text in the embodiment of the present application is a transaction note text with a relatively large corresponding resource amount among multiple transaction note texts; and the greater the amount of virtual resources transferred during the execution of the business transaction, the more important the business transaction is to the business object, and accordingly, the transaction note text of the business transaction can more directly reflect the actual business activities of the business object. Therefore, performing type identification processing based on the target transaction note text is conducive to further obtaining a more accurate business type.

[0019] It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The accompanying drawings are incorporated into and constitute a part of the specification, illustrating embodiments consistent with the present application and, together with the specification, serving to explain the principles of the present application. Obviously, the drawings described below are merely some embodiments of the present application, and a person of ordinary skill in the art can derive other drawings based on these drawings without inventive effort. In the drawings:

[0021] Figure 1 This is a schematic diagram of the structure of a business object management system provided by an embodiment of the present application;

[0022] Figure 2 This is a flowchart of a business object management method provided by an embodiment of the present application;

[0023] Figure 3 This is a schematic diagram of the word segmentation encoding process of a transaction note text provided in an embodiment of the present application;

[0024] Figure 4 This is a schematic diagram of feature encoding using an attention mechanism provided in an embodiment of the present application;

[0025] Figure 5 This is a flowchart of another business object management method provided by an embodiment of the present application;

[0026] Figure 6 This is a schematic diagram of the model structure of a business type identification model provided in an embodiment of the present application;

[0027] Figure 7 This is a schematic diagram of a training process of a business type recognition model provided in an embodiment of the present application;

[0028] Figure 8 is a structural block diagram of a business object management device shown in an exemplary embodiment of the present application;

[0029] Figure 9 A schematic diagram of the structure of a computer system suitable for implementing an electronic device according to an embodiment of the present application is shown. DETAILED DESCRIPTION

[0030] Exemplary embodiments will be described in detail herein, with examples illustrated in the accompanying drawings. When the following description refers to the drawings, identical numerals in different figures represent identical or similar elements unless otherwise indicated. The embodiments described in the following exemplary embodiments are not intended to represent all embodiments consistent with the present application. Rather, they are merely examples of apparatus and methods consistent with certain aspects of the present application, as detailed in the appended claims.

[0031] The block diagrams shown in the accompanying drawings are merely functional entities and do not necessarily correspond to physically separate entities. That is, these functional entities may be implemented in software, in one or more hardware modules or integrated circuits, or in different networks and / or processor devices and / or microcontroller devices.

[0032] The flowcharts shown in the accompanying drawings are for illustrative purposes only and do not necessarily include all contents and operations, nor must they be executed in the order described. For example, some operations may be decomposed, while others may be combined or partially combined. Therefore, the actual execution order may vary depending on the actual situation.

[0033] In the embodiments of the present application, the term "module" or "unit" refers to a computer program or a part of a computer program that has a predetermined function and works together with other related parts to achieve a predetermined goal, and can be implemented in whole or in part by using software, hardware (such as processing circuits or memories) or a combination thereof. Similarly, a processor (or multiple processors or memories) can be used to implement one or more modules or units. In addition, each module or unit can be part of an overall module or unit that includes the function of the module or unit.

[0034] It should also be noted that the term "plurality" used in this application refers to two or more. "And / or" describes the relationship between associated objects, indicating that three possible relationships exist. For example, "A and / or B" can mean: A exists alone, A and B exist simultaneously, or B exists alone. The character " / " generally indicates that the associated objects are in an "or" relationship.

[0035] With the continuous development of internet technology, artificial intelligence (AI) has also advanced significantly. AI refers to the theories, methods, techniques, and application systems that use digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to achieve optimal results. In other words, AI is a comprehensive field of computer science; it primarily aims to understand the essence of intelligence and produce new intelligent machines that can react in a manner similar to human intelligence, enabling them to possess diverse capabilities such as perception, reasoning, and decision-making.

[0036] Artificial intelligence (AI) technology is a comprehensive discipline encompassing a wide range of fields, encompassing both hardware and software technologies. Foundational AI technologies generally include sensors, specialized AI chips, cloud computing, distributed storage, big data processing, pre-trained models, operating / interaction systems, and mechatronics. Pre-trained models, also known as large models or basic models, can be fine-tuned and widely applied to downstream tasks across various AI domains. AI software technologies primarily encompass computer vision, speech processing, natural language processing (NLP), and machine learning / deep learning.

[0037] Natural language processing (NLP) is a key area of ​​research in computer science and artificial intelligence. It studies theories and methods that enable effective communication between humans and computers using natural language. Natural language processing involves natural language, the language we use daily, and is closely related to linguistics. It also involves computer science and mathematics. Pre-trained models, a key technology for model training in artificial intelligence, are derived from large language models (LLMs) in the field of natural language processing (NLP). After fine-tuning, LLMs can be widely applied to downstream tasks. Specifically, natural language processing technologies typically include text processing, semantic understanding, machine translation, robotic question-answering, knowledge graphs, and other technologies.

[0038] Research on many techniques within natural language processing requires the use of machine learning. Machine learning is a multidisciplinary field that encompasses probability theory, statistics, approximation theory, convex analysis, and algorithmic complexity theory. It specifically studies how computers can simulate or implement human learning behaviors to acquire new knowledge or skills and reorganize existing knowledge structures to continuously improve their performance. Machine learning is the core of AI and the fundamental path to intelligent computing. Deep learning, on the other hand, utilizes deep neural network systems for machine learning. Machine learning / deep learning typically encompasses various techniques, including artificial neural networks, adversarial learning, reinforcement learning (RL), supervised learning, and unsupervised learning. Supervised learning refers to the use of training samples with known (labeled) categories to train models, while unsupervised learning refers to the use of training samples with unknown (unlabeled) categories to train models.

[0039] At the same time, with the rapid development of network communication technology and smart devices, transactions are no longer limited to physical business premises. More and more businesses are conducting various operations online and completing transactions with other parties related to these businesses. To prevent illegal activities from occurring on online platforms or applications, when businesses connect their businesses to online platforms or applications, the platforms or applications clearly define the business type of the business they operate. However, currently, online platforms or applications rely on businesses to enter the business type themselves. While this method is relatively convenient, it can lead to cases where the business type is either left blank or incorrectly entered.

[0040] Based on this, an embodiment of the present application provides a business object management solution, which takes into account that the transaction note text that explains the business transaction can reflect the actual business activities, more realistic transaction methods and transaction purposes of the business object. Therefore, it will use the transaction note text of each business transaction generated in the business operation process of the business object to identify the business type to which the business object's business operation belongs.

[0041] A business object refers to the object of business operation. A business type is a combination of multiple businesses. For example, the business type for milk tea, noodles, and Western food could be catering; while the business type for trailers and renovations could be construction.

[0042] Furthermore, since the greater the amount of virtual resources transferred during a business transaction, the more important the transaction is to the business subject, the more directly the transaction's transaction notes reflect the business subject's actual operating activities. Therefore, we can further filter out transaction notes corresponding to transactions with larger resource amounts, and then use these filtered transaction notes to identify the business type of the business subject's operations.

[0043] Based on the above business object management solution, this embodiment of the application provides a business object management system, which can be found in Figure 1 , Figure 1 The business object management system shown may include a terminal device 101 and a server 102. The number of terminal devices 101 may include multiple, and the number of servers 102 may include multiple. A communication connection is established between any terminal device and any server. For example, the terminal device 101 may include any one or more of a smartphone, a tablet computer, a laptop computer, a desktop computer, an intelligent voice interaction device, a smart home appliance, a vehicle-mounted terminal, an aircraft, and a smart wearable device. Various types of clients such as a live broadcast client, a social client, a shopping client, a payment client, and a map client may be installed in the terminal device 101. The server 102 may be a server or server cluster that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms. Any terminal device 101 and any server 102 may be directly or indirectly connected in communication via wired or wireless communication, and this application does not impose any restrictions on this.

[0044] In some embodiments, the above business object management method can be performed by Figure 1 The server 102 in the business object management system shown executes the following specific execution process: Server 102 may first obtain the transaction memo text for each business transaction generated during the business object's operation. Furthermore, Server 102 may obtain the resource amount of virtual resources transferred during each business transaction. Next, Server 102 may select a target transaction memo text from multiple transaction memo texts based on the resource amount corresponding to the business transaction explained in each transaction memo text; the resource amount corresponding to the target transaction memo text is greater than the resource amount corresponding to the other transaction memo texts. Finally, Server 102 may perform type identification processing on the target transaction memo text to determine the business type to which the business object's operation belongs.

[0045] Optionally, the above business object management method can also be performed by Figure 1 The terminal device 101 in the business object management system shown is executed. The specific execution process can refer to the specific execution process of the server 102, which will not be repeated here.

[0046] In other embodiments, the business object management method can be run in a business object management system, which can include a terminal device and a server. Figure 1 The business object management system shown is implemented jointly by a terminal device 101 and a server 102. Specifically, the terminal device 101 collects transaction notes from each business transaction generated during the business object's operation, as well as the amount of virtual resources transferred during each business transaction, and sends these to the server 102. The server 102 then selects a target transaction note from multiple transaction notes based on the amount of resources corresponding to the business transactions explained in each transaction note. The server 102 then identifies the target transaction note and determines the business type of the business object's operation.

[0047] It should be noted that the embodiments of the present application can be applied to various scenarios that require text review, including but not limited to cloud technology, AI (Artificial Intelligence), smart cities, smart transportation, assisted driving and other scenarios. It can also be used for live broadcast applications, shopping applications, payment applications and any other applications that can access business or assist business transactions, without limitation.

[0048] It should be noted that in the specific implementation of this application, if the transaction note text, resource quantity and other related data or information involve object-related, when the embodiment of this application is applied to a specific product or technology, it is necessary to obtain the object's permission or consent, and the collection, use and processing of the relevant data or information must comply with the relevant laws, regulations and standards of the relevant countries and regions.

[0049] The following describes in detail the various implementation details of the technical solutions of the embodiments of the present application:

[0050] like Figure 2 As shown, Figure 2 This is a flow chart of a business object management method according to an embodiment of the present application. The method can be applied to Figure 1 The business object management system shown in FIG. 1 may be executed by a terminal device or a server, or may be executed by both a terminal device and a server. In the embodiment of the present application, the method executed by the server is used as an example for explanation. The business object management method may include S201 to S204, which are described in detail as follows:

[0051] S201. Obtain transaction remark text of each business transaction generated during the business operation of the business object.

[0052] In this embodiment of the present application, the transaction notes for each business transaction are used to explain the transaction. The process by which a person or group purchases the business operated by the business entity is the process of generating and executing a business transaction. For example, if the business entity is a bakery, then the process by which a customer selects a loaf of bread and pays the business entity is the completion of a business transaction.

[0053] In addition, business transactions generated in the course of a business object's business operations may include those generated in the business object's recent business operations, or may include all business transactions generated since the business object began operating its business, or may include business transactions generated within a certain period of time after the business object began operating its business, and there is no limitation here.

[0054] Optionally, the transaction memo text acquisition process may specifically include acquiring the transaction memo text for each business transaction generated by the business object during a preset time period from the current time. The preset time period may be manually set or may be set by a server or terminal device in the business object management system, and is not limited here.

[0055] In actual applications, since different businesses have different off-seasons and peak seasons, different preset time periods can be set for the off-season and peak seasons; for example, the length of the preset time period set for the off-season can be longer than the length of the preset time period set for the peak season. Optionally, since different businesses have different levels of business activity, different preset time periods can also be set for different businesses; for example, the length of the preset time period set for businesses with high business activity can be shorter than the length of the preset time period set for businesses with low business activity.

[0056] Specifically, the business activities of business objects are changing dynamically, and the transaction notes text of business transactions generated recently can better represent the actual current business activities of the business objects. Therefore, the transaction notes text of business transactions generated recently can be obtained so that the business type of the business object's business can be identified and predicted through the transaction notes text that best reflects the actual situation of the business object's current business operations, which is conducive to improving the accuracy of business type recognition.

[0057] Optionally, in order to further understand the dynamic changes in the business activities of the business object, the transaction note text of each business transaction generated by the business object during the business operation process can also be periodically obtained.

[0058] Specifically, the process of periodically obtaining transaction memo text may include: if it is detected that the time difference between the current time and the previous period node time reaches a specified time, the current time may be used as the current period node time, and the transaction memo text for each business transaction generated by the business object during the period from the previous period node time to the current time may be obtained. The specified time may be set manually or by a server or terminal device in the business object management system, and is not limited here.

[0059] S202: Acquire the amount of virtual resources transferred during the execution of each business transaction.

[0060] In the embodiment of the present application, the virtual resource may be resource data that can serve as a general equivalent, or may be an item that can serve as a general equivalent and can be paid in electronic form, which is not limited here.

[0061] Then, the virtual resources transferred during the execution of each business transaction may include: virtual resources transferred to the business object by other objects when they purchase or use the business object to operate their business, and may also include virtual resources transferred to other objects by the business object for normal business operations, which are not limited here.

[0062] Accordingly, the amount of virtual resources is the quantified value of virtual resources. For example, if another object purchases a service operated by a business object and pays 36 yuan to the business object through a payment application, then the virtual resource is the currency in the payment application, and the amount of virtual resources transferred in this transaction is 36.

[0063] S203: Filter out a target transaction note text from the multiple transaction note texts based on the resource amount corresponding to the business transaction explained in each transaction note text.

[0064] In an embodiment of the present application, the amount of resources corresponding to the target transaction note text is greater than the amount of resources corresponding to other transaction note texts.

[0065] In one embodiment, the target transaction memo text screening process may include: sorting multiple transaction memo texts in descending order of resource volume to obtain multiple sorted transaction memo texts; and selecting, from the sorted multiple transaction memo texts, a transaction memo text with a sequence number less than or equal to a preset value as the target transaction memo text. The preset value may be manually set or may be set by a server or terminal device in the business object management system, and is not limited here.

[0066] Alternatively, the target transaction note text may be selected by ascending sorting, which is not described in detail here. Alternatively, the target transaction note text may be selected by other methods such as bubble sorting, which is not limited here.

[0067] S204: Perform type identification processing on the target transaction remark text to obtain the business type to which the business object's operating business belongs.

[0068] In an embodiment of the present application, the specific process of type identification processing may include: performing feature extraction processing on the target transaction note text to obtain a text feature vector of the target transaction note text; performing classification processing on the text feature vector to obtain the business type to which the business object's operating business belongs.

[0069] In practice, deep neural networks such as BERT (a language representation model) and convolutional neural networks can be used to extract text features. Since there may be multiple business types, this is a multi-classification problem; therefore, multi-classification functions such as softmax (an activation function for multi-classification) and cross-entropy loss can be used for classification.

[0070] In one embodiment, because the same word may convey different meanings in different sentences, that is, a word in a sentence may be influenced by other words in the sentence, thereby affecting the semantics of the entire sentence. Therefore, in order to extract a feature vector that can more accurately represent the semantics of the target transaction memo text, thereby improving the accuracy of subsequent business type recognition, the specific process of feature extraction processing may include: performing word segmentation processing on the target transaction memo text to obtain multiple fields; performing initial encoding processing on each field to obtain a field feature vector for each field; and performing feature encoding processing on the field feature vector of each field based on the field feature vectors of other fields in the multiple fields to obtain a text feature vector for the target transaction memo text.

[0071] In practical applications, type recognition can be accomplished using the Bert model and the softmax function. The Bert model is a pre-trained model, and its structure is equivalent to the encoder of a Transformer model. The word representations learned by the Bert model vary depending on the context. During pre-training, the Bert model must jointly complete two tasks: 1. Cloze, which predicts missing words in a text; and 2. Determine whether the preceding and following sentences constitute context.

[0072] The target transaction note text, serving as the input to the BERT model, requires word segmentation and initial encoding. This can be done using the BERT model's word segmenter (i.e., tokenizer). Initial encoding of each field yields a field feature vector that is the sum of each field's semantic vector (token embeddings), segment embeddings (used to identify the sentence to which the field belongs), and position embeddings (used to indicate the relative position of the word within the sentence).

[0073] Please see the attached Figure 3 , shows a schematic diagram of the word segmentation encoding process of a transaction note text. Figure 3 As shown in the figure, after the target transaction note text "mybagisdirty,itneedscleaning" is segmented, the input text "[CLS]mybagisdirty[SEP]itneedscleaning[SEP]" can be obtained. Among them, [CLS] before the first character of the inserted text is a classification marker used to indicate the beginning of the text; [SEP] is a separation marker used to separate two sentences in the text. At the same time, Figure 3 As shown, by adding the semantic vector, segment vector and position vector of each field obtained by word segmentation, the field feature vector of each field can be obtained.

[0074] The main structure of the Bert model for target transaction note text input is composed of multiple multi-head self-attention mechanisms. The so-called attention mechanism means that for each word, other words in the sentence are used to express this word, and the weights of other words on this word are different. Figure 4 , shows a feature encoding schematic diagram that introduces the attention mechanism. Figure 4 As shown in the figure, after obtaining the field feature vectors E0 to E N (N is a positive integer) and then E0 to E N As the input of the first layer of network in the Bert model. Each layer of network in the Bert model takes the output of the previous layer of network as the input of the network, and for each field, each layer of network will perform feature encoding on the feature vector of the field output by the previous layer of network based on the feature vector of other fields output by the previous layer of network; finally, the feature vectors T0 to T N , and based on T0 to T N Get the text feature vector of the target transaction note text.

[0075] Multi-head attention refers to multiple attention mechanisms that can capture relationships at different levels of abstraction from different perspectives. The formula (1-1) for the multi-head attention mechanism is as follows:

[0076] MultiHead(Q,K,V)=Concat(Head1;Head2;...;Head N )W 0 (1-1)

[0077] Among them, Q, K, V represent query vector, key vector and value vector respectively, N represents the total number of fields (that is, the total number of headers), Head i represents the output of the i-th head, i is a positive integer; W 0 is the output transformation matrix. i The formula (1-2) is as follows:

[0078] Head i =Attention(QW i Q ,KW i K ,VW i V ) (1-2)

[0079] Among them, W i Q 、W i K and W i Q are the query transformation matrix, key transformation matrix, and value transformation matrix of the i-th head respectively; Attention is the attention calculation function.

[0080] Compared to information such as the business entity's name and its income and expenditure, transaction notes better reflect the entity's actual business activities and more directly indicate the true purpose and method of the transaction. For example, a business entity operating a milk tea shop might have a name like "Gong Xi Fa Cai" (Gong Xi Fa Cai), which has nothing to do with milk tea. A business entity operating a beauty salon might have a name like "Self-employed - Zhang San." It's impossible to determine the type of business an entity operates based on the names of these business entities.

[0081] In an embodiment of the present application, by performing type identification processing on the target transaction note text, a more accurate and more relevant business type can be classified and predicted based on the actual business operations of the business object reflected in the target transaction note text. At the same time, the target transaction note text in the embodiment of the present application is a transaction note text with a relatively large amount of corresponding resources among multiple transaction note texts; and the greater the amount of virtual resources transferred during the execution of the business transaction, the more important the business transaction is to the business object, and accordingly, the transaction note text of the business transaction can more directly reflect the actual business activities of the business object. Therefore, performing type identification processing based on the target transaction note text is conducive to obtaining a more accurate business type.

[0082] In one embodiment of the present application, another business object management method is provided, which can be applied to Figure 1 The business object management system shown in the figure can be executed by a terminal device or a server, or by a terminal device and a server together. In the embodiment of the present application, the method is described by taking the server as an example. Figure 5 As shown, a flow chart of another business object management method is shown. Figure 2 This paper expands on the method shown in .

[0083] Among them, S501 to S508 are described in detail as follows:

[0084] S501. Obtain training transaction remark text of each business transaction generated during the business operation of the training subject, as well as a type label of the business operation of the training subject.

[0085] In the embodiment of the present application, each training transaction remark text is used to explain each business transaction generated during the business operation of the training subject. The type label is the actual business type of the business operated by the training subject.

[0086] S502: Perform feature extraction on the training transaction remark text to obtain a remark feature vector.

[0087] In the embodiment of the present application, the specific implementation of the feature extraction process can refer to the specific implementation of the feature extraction process in step S204, which will not be repeated here.

[0088] S503 : Perform feature extraction processing on each preset service type to obtain a type feature vector of each preset service type.

[0089] In the embodiment of the present application, the specific implementation of the feature extraction process can be referred to the specific implementation of the feature extraction process in step S204, which will not be repeated here.

[0090] S504 , training is performed based on the remark feature vector, the type feature vectors of each preset business type, and the type label to obtain a business type recognition model.

[0091] In an embodiment of the present application, the training process of the business type identification model may include: classifying the remark feature vector to obtain the type prediction result of the business operated by the training object; obtaining the type simulation result of the business operated by the training object based on the type feature vector, remark feature vector and type label of each preset business type; calculating the probability distribution difference between the type simulation result and the type prediction result, and training in the direction of reducing the probability distribution difference to obtain a business type identification model.

[0092] The specific implementation of the classification process can be found in the specific implementation of the classification process in step S204 and will not be described in detail here. The type prediction result includes the probability that the training subject's business belongs to each preset business type. The type simulation result also includes the probability that the training subject's business belongs to each preset business type. However, the type prediction result and the type simulation result will generally differ.

[0093] In one embodiment, the process of obtaining type simulation results may include: obtaining the similarity between the type feature vector and the remark feature vector of each preset business type, and generating a label confusion vector based on the multiple similarities obtained; performing label encoding processing on the type label to obtain a label feature vector; and performing feature fusion on the label confusion vector and the label feature vector to obtain the type simulation result of the business operation of the training object.

[0094] Among them, each element in the label confusion vector is used to characterize each similarity. The specific process of label encoding processing can be: one-hot encoding (One-hotEncoding) based on multiple preset business types and the preset business type indicated by the type label of the training object to obtain a label feature vector. For example, the preset business types may include catering, construction, automobiles, and retail, and the type label of the training object is construction. Then, after label encoding processing is performed on the type label of the training object, the obtained label feature vector can be [0,1,0,0]. Optionally, the type label can also be encoded using encoding methods such as count encoding and type consolidation encoding, which are not limited here.

[0095] Optionally, the feature fusion process between the label confusion vector and the label feature vector may include: fusing the label confusion vector with the label feature vector to obtain a fused vector; and classifying the fused vector to obtain a type simulation result. The fusion method may be feature addition, weighted addition, or obtaining an average feature vector, etc., which is not limited here.

[0096] Specifically, in a multi-classification model, training labels are generally represented by One-hotEncoding; this representation has many defects, which are mainly reflected in: the relationship between the real label and other labels is ignored, resulting in the model training process being unable to learn a lot of useful knowledge. For example: "bird" and "airplane" are two relatively similar items, so in this case, the model should feel that the two are closer when predicting and give a smaller loss value. However, One-hotEncoding cannot reflect this difference, especially when facing easily confused classification tasks and noisy (mislabeled) data sets, it is more easily affected. Therefore, the embodiment of the present application adds the type feature vectors of each preset business type to the model training, so that the business type recognition model can learn the relationship between each preset business type and the type label, which is beneficial to improving the robustness of the business type recognition model, thereby improving the recognition accuracy of the business type recognition model, and achieving the purpose of obtaining more accurate business types.

[0097] S505: Acquire transaction remark text of each business transaction generated during the business operation of the business object.

[0098] In the embodiment of the present application, the specific implementation of step S505 can refer to the specific implementation of step S201 and will not be repeated here.

[0099] S506: Obtain the resource amount of the virtual resources transferred during the execution of each business transaction.

[0100] In the embodiment of the present application, the specific implementation of step S505 can refer to the specific implementation of step S201 and will not be repeated here.

[0101] S507: Filter out a target transaction note text from the multiple transaction note texts based on the resource amount corresponding to the business transaction explained in each transaction note text.

[0102] Since transaction memo texts are usually randomly input by business objects or transaction objects of business objects, they often contain some meaningless or unhelpful words and phrases that do not contribute to expressing the business objects' business activities. Therefore, in an embodiment of the present application, the process of screening target transaction memo texts may further include: first, discarding the transaction memo texts containing specified characters from multiple transaction memo texts to obtain multiple updated transaction memo texts; then, screening transaction memo texts with a text length greater than a preset length from the multiple updated transaction memo texts as candidate transaction memo texts; and finally, screening the target transaction memo text from the candidate transaction memo texts based on the resource amount corresponding to the candidate transaction memo texts.

[0103] The designated characters and preset length may be manually set or configured by the server or terminal device in the business object management system, and are not limited here. Specifically, the designated characters may be derived from a collection of meaningless or irrelevant words and phrases related to the business object's business activities. For example, the designated characters may include numbers, symbols, stop words, etc.; stop words include "pay," "order," "pay," and other words that are irrelevant to the business object's business activities.

[0104] Furthermore, considering that text that is too short cannot accurately express the business object's operating activities and transaction purposes, and is often irrelevant to the business object's operating activities, transaction memo text with relatively short text lengths may not be included in the transaction memo text used to identify business types. Specifically, since many relatively short transaction memo texts are often the object names of business objects or transaction objects, the preset length can be set based on the text length of the object names of business objects or transaction objects. For example, if it is found that object names are often three characters long, the preset length can be set to 3.

[0105] In addition, the specific implementation of step S505 can also refer to the specific implementation of step S201, which will not be described in detail here.

[0106] Optionally, during the training phase of the business type recognition model, the training transaction memo text obtained can be used to discard specified characters and filter out training transaction memo texts whose length is less than or equal to a preset length; and then the training transaction memo text finally obtained by screening can be used as the training transaction memo text used in the model training process.

[0107] S508: Perform type recognition processing on the target transaction remark text through the business type recognition model to obtain the business type to which the business object's operating business belongs.

[0108] In this embodiment of the present application, the specific process of type identification may include: invoking a business type identification model to perform feature extraction on the target transaction note text to obtain a text feature vector for the target transaction note text; and classifying the text feature vector to obtain the business type to which the business object's business belongs. The specific implementation of the feature extraction and classification processes can be found in the specific implementation of step S204 and will not be further described here.

[0109] For specific implementation, please refer to the attached Figure 6 , shows a schematic diagram of the model structure of a business type identification model. Figure 6 As shown, the business type recognition model includes a basic prediction module 601 and a label optimization module 602 (also known as a Label Confusion Model, LCM). The basic prediction module 601 includes a text feature extraction network 603 and a classification function 604, and the label optimization module 602 includes a label feature extraction network 605, a similarity calculation layer 606, and a feature fusion layer 607.

[0110] After word segmentation and feature encoding processing of the training transaction notes text, the field feature vectors W1 to W corresponding to the training transaction notes text can be obtained. N , where N is a positive integer, representing the total number of fields obtained after training transaction notes text segmentation. Then, the field feature vectors W1 to W N The text feature extraction network 603 is input to obtain a comment feature vector 608 output by the text feature extraction network 603. The text feature extraction network 603 may include at least one of a convolutional neural network (CNN), a recurrent neural network (RNN), or a Bert (BERT) network, but this is not limited here. The comment feature vector 608 is then input to the classification function 604 to obtain a type prediction result 609 (i.e., a predicted label distribution (PLD)).

[0111] Taking the text feature extraction network 603 as a Bert model and the classification function 604 as a softmax function as an example, the expression formula of the basic prediction module 601 is shown in formulas (2-1) and (2-2):

[0112] v (i) =f I (w) = f I (w1,w2,...,w N )=[v1 (i) ,v2 (i) ,...,vN (i) ] (2-1)

[0113] y (p) =softmax(v (i) ) (2-2)

[0114] Among them, v (i) is the remark feature vector, I is the mapping function of the text feature extraction network; (p) is the type prediction result.

[0115] At the same time, if Figure 6 As shown, multiple preset business types such as type 1 to type M can be input into the label feature extraction network 605 respectively, thereby obtaining the type feature vector 610 of each preset business type output by the label feature extraction network 605. Specifically, the label feature extraction network 605 can be a deep neural network (DNN).

[0116] Then, the type feature vector and remark feature vector 608 of each preset business type are input into the similarity calculation layer to calculate the similarity between the type feature vector and remark feature vector 608 of each preset business type, and the label confusion vector 611 is obtained by the calculated multiple similarities. After that, the label feature vector 612 of the type label obtained after the one-hot encoding and the label confusion vector 611 are input into the feature fusion layer 607, thereby obtaining the type simulation result 613 (i.e., simulated label distribution, SLD) output by the feature fusion layer 607.

[0117] Finally, if Figure 6 As shown, the KL divergence (Kullback-Leibler divergence, also known as relative entropy, is a measurement method for measuring the difference between two probability distributions) between the type prediction result 609 and the type simulation result 613 can be calculated; then, the business type recognition model is trained in the direction of reducing the KL divergence.

[0118] Furthermore, after the business type identification model is trained, the business type identification model can be used to perform type identification processing on the business operated by the business objects connected to the Internet platform or application.

[0119] In this example, the label feature extraction network 605 in the label optimization module learns the representation of each label (i.e., the preset business type) and calculates similarity with the vector representation of the input corpus, thereby obtaining a distribution reflecting the degree of confusion / similarity between the labels. Finally, by calculating the KL divergence, the original label distribution can be optimized based on the obtained confusion distribution, allowing the business type recognition model to more accurately distinguish similar business types, thereby improving the recognition accuracy and robustness of the business type recognition model.

[0120] For specific implementation, please refer to the attached Figure 7 , shows a schematic diagram of the training process of a business type recognition model. Figure 7 As shown, by inputting transaction notes into the basic prediction module of the business type recognition model, a type prediction result is obtained. Then, by inputting type labels into the label optimization module of the business type recognition model, a type simulation result is obtained. The business type recognition model is then trained using the type prediction and simulation results. Finally, the trained business type recognition model is used to identify the business type of the business object's operations.

[0121] In one embodiment, after identifying the business type of the business object's business operation, the business type registered by the business object can also be obtained; if the identified business type does not match the registered business type, the business object's business operation is managed and controlled.

[0122] The business type registered by a business object includes the business type registered by the business object during the process of accessing the internet platform or application. The business type registered by the business object can be based on the business object's business license or can be filled in by the business object itself, without limitation.

[0123] The management and control operations may include stopping the business object's business operations, stopping business transactions for the business object's business operations, outputting type verification prompt information for the business object's business operations, monitoring business transactions for the business object's business operations, and other one or more operations for supervising the business object's business operations, which are not limited here.

[0124] In practical applications, the business type identification model can be applied in financial risk control scenarios. In these scenarios, payment institutions need to verify various information about business entities, and the business type of the entity's actual operations is a crucial piece of information. The business type identified by the business type identification model can better reflect the entity's actual business type, facilitating payment institutions' financial risk management of the entity.

[0125] In one embodiment, transaction data of business transactions generated during the business operation of the business object can also be obtained; then, the transaction data is analyzed according to the preset feature dimensions to obtain transaction feature information of the business object under the preset feature dimensions; then, a business type that matches the transaction feature information is selected from multiple preset business types; if the selected business type does not match the identified business type, the business operation of the business object is managed and controlled.

[0126] Transaction data may include transaction objects, transaction amounts, transaction circumstances, transaction quantities, and other transaction-related data, without limitation. Preset feature dimensions can be derived based on transaction patterns across various business types. Preset feature dimensions may also be manually set or configured by a server or terminal device within the aforementioned business object management system, without limitation.

[0127] For example, different businesses may have different total transaction amounts within the same timeframe. For example, the total transaction amount for the catering industry within a month typically ranges from 10 million to 100,000 yuan, while the total transaction amount for the construction industry within a month typically ranges from 200,000 to 20 million yuan. Therefore, the total transaction amount within the past month can be used as a preset feature dimension. Accordingly, if the total transaction amount for a business object's business operations within the past month is 5,560 yuan, then the transaction feature information for this business object under the preset feature dimension of total transaction amount within the past month is 5,560 yuan. Furthermore, the target business type that matches this business object's business operations can be catering.

[0128] In practical applications, the business type recognition model can be applied to suspicious transaction monitoring. For example, the business object's transaction data can be used to extract transaction characteristics, identifying the target business type as catering. Simultaneously, the business type recognition model can be used to identify the type of transaction notes generated during the business object's operations, resulting in the business type being building materials. These two business types are completely different, so the business object can be reported to the regulatory authorities as suspicious.

[0129] In one embodiment, taking into account that some business objects engaged in illegal and non-compliant businesses may disguise themselves to a certain extent when they first access the Internet platform or application, and that business objects may quietly change their business operations, the business type recognition model can also be used periodically to perform type recognition processing on the transaction note text of business transactions generated during the business operations of the business objects to obtain multiple business types; if multiple business types are detected to be mismatched, the business operations of the business objects are managed and controlled.

[0130] Specifically, while the business activities of business entities may change, the types of business they operate under legal and compliant conditions remain constant. Therefore, by periodically identifying the types of business entities they operate under, we can pay attention to those whose business types have changed, thereby further preventing illegal and non-compliant businesses from infiltrating internet platforms or applications.

[0131] Since there are many different types of business, and some of them are very similar (such as the catering industry and the baking industry), conventional neural network models cannot classify these business types well when faced with business types with smaller differences. Therefore, in an embodiment of the present application, by introducing various preset business types in the model training stage, each preset business type and type label can form a distribution that reflects the degree of confusion / similarity between the labels. In this way, the business type recognition model can learn the characteristics between confusing / similar business types during the training stage, so that the business type recognition model can more accurately distinguish similar business types, thereby achieving the purpose of improving the recognition accuracy and robustness of the business type recognition model. Therefore, the business type obtained by type recognition processing by the business type recognition model is more accurate.

[0132] Here, an embodiment of the device of the present application is introduced, which can be used to execute the business object management method in the above embodiment of the present application. For details not disclosed in the embodiment of the device of the present application, please refer to the embodiment of the business object management method in the above embodiment of the present application.

[0133] The embodiment of the present application provides a business object management device, such as Figure 8 As shown, the apparatus includes an acquisition unit 801, a screening unit 802, and a processing unit 803, wherein:

[0134] The acquisition unit 801 is used to acquire the transaction note text of each business transaction generated during the business operation of the business object. The transaction note text of each business transaction is used to explain each business transaction;

[0135] The acquisition unit 801 is further configured to acquire the amount of virtual resources transferred during the execution of each business transaction;

[0136] A screening unit 802 is configured to screen out a target transaction memo text from the plurality of transaction memo texts based on the resource amounts corresponding to the business transactions explained in the respective transaction memo texts; wherein the resource amount corresponding to the target transaction memo text is greater than the resource amounts corresponding to the other transaction memo texts;

[0137] The processing unit 803 is configured to perform type identification processing on the target transaction remark text to obtain the business type to which the business object's operating business belongs.

[0138] In one embodiment of the present application, based on the aforementioned scheme, the target transaction remark text is subjected to type identification processing through a business type identification model to obtain the business type to which the business object's business operation belongs; the business object management device may also include a model training unit 804, and the model training unit 804 is used to obtain the training transaction remark text of each business transaction generated in the training object's business operation process, as well as the type label of the training object's business operation; the training transaction remark text is subjected to feature extraction processing to obtain a remark feature vector; each preset business type is subjected to feature extraction processing to obtain a type feature vector of each preset business type; training is performed based on the remark feature vector, the type feature vector of each preset business type, and the type label to obtain a business type identification model.

[0139] In one embodiment of the present application, based on the aforementioned scheme, when the model training unit 804 performs training based on the remark feature vector, the type feature vectors of each preset business type, and the type label to obtain a business type recognition model, it can be specifically used to: classify the remark feature vector to obtain a type prediction result of the training object's business; obtain a type simulation result of the training object's business based on the type feature vector, the remark feature vector, and the type label of each preset business type; calculate the probability distribution difference between the type simulation result and the type prediction result, and train in the direction of reducing the probability distribution difference to obtain a business type recognition model.

[0140] In one embodiment of the present application, based on the aforementioned scheme, when the model training unit 804 obtains the type simulation result of the training object's business operation based on the type feature vector, remark feature vector and type label of each preset business type, it can be specifically used to: obtain the similarity between the type feature vector and the remark feature vector of each preset business type, and generate a label confusion vector based on the obtained multiple similarities; perform label encoding processing on the type label to obtain a label feature vector; perform feature fusion on the label confusion vector and the label feature vector to obtain the type simulation result of the training object's business operation.

[0141] In one embodiment of the present application, based on the aforementioned scheme, when the screening unit 802 screens out a target transaction memo text from multiple transaction memo texts based on the amount of resources corresponding to the business exchanges explained in each transaction memo text, it can be specifically used to: discard the specified characters in the multiple transaction memo texts containing the specified characters to obtain multiple updated transaction memo texts; screen out transaction memo texts with a text length greater than a preset length from the multiple updated transaction memo texts as candidate transaction memo texts; and screen out the target transaction memo text from the candidate transaction memo texts based on the amount of resources corresponding to the candidate transaction memo texts.

[0142] In one embodiment of the present application, based on the aforementioned solution, the processing unit 803 may also be used to: obtain the business type registered by the business object; if the identified business type does not match the registered business type, perform management and control operations on the business object's business operations.

[0143] In one embodiment of the present application, based on the aforementioned scheme, the processing unit 803 can also be used to: obtain transaction data of business transactions generated in the business operation process of the business object; analyze the transaction data according to preset feature dimensions to obtain transaction feature information of the business object under the preset feature dimensions; select a business type that matches the transaction feature information from multiple preset business types; if the selected business type does not match the identified business type, perform management and control operations on the business object's business operations.

[0144] In one embodiment of the present application, based on the aforementioned scheme, the processing unit 803 can also be used to: periodically perform type identification processing on the transaction note text of the business transactions generated during the business operation of the business object to obtain multiple business types; if it is detected that multiple business types do not match, then perform management and control operations on the business object's business operations.

[0145] In one embodiment of the present application, based on the aforementioned scheme, when the processing unit 803 performs type identification processing on the target transaction memo text to obtain the business type to which the business object's operating business belongs, it can be specifically used to: perform word segmentation processing on the target transaction memo text to obtain multiple fields; perform initial encoding processing on each field to obtain a field feature vector of each field; perform feature encoding processing on the field feature vector of each field based on the field feature vectors of other fields in the multiple fields to obtain a text feature vector of the target transaction memo text; perform classification processing on the text feature vector to obtain the business type to which the business object's operating business belongs.

[0146] It should be noted that the apparatus provided in the above embodiment and the method provided in the above embodiment belong to the same concept, wherein the specific manner in which each module and unit performs operations has been described in detail in the method embodiment and will not be repeated here.

[0147] The device provided by the above embodiment can be located in a terminal device or in a server. The device provided by the embodiment of the present application can classify and predict a business type that is more consistent with the actual business operations of the business object based on the actual business operations of the business object reflected in the target transaction note text by performing type identification processing on the target transaction note text. At the same time, the target transaction note text in the embodiment of the present application is a transaction note text with a relatively large corresponding resource amount among multiple transaction note texts; and the greater the amount of virtual resources transferred during the execution of the business transaction, the more important the business transaction is to the business object, and accordingly, the transaction note text of the business transaction can more directly reflect the actual business activities of the business object. Therefore, performing type identification processing based on the target transaction note text is conducive to obtaining a more accurate business type.

[0148] An embodiment of the present application also provides an electronic device, comprising one or more processors and a storage device, wherein the storage device is used to store one or more computer programs, and when the one or more computer programs are executed by one or more processors, the electronic device implements the above-mentioned business object management method.

[0149] Figure 9 A schematic diagram of the structure of a computer system suitable for implementing an electronic device according to an embodiment of the present application is shown.

[0150] It should be noted that Figure 9 The computer system 900 of the electronic device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0151] like Figure 9 As shown, the computer system 900 includes a processor (Central Processing Unit, CPU) 901, which can perform various appropriate actions and processes according to the program stored in the read-only memory (Read-Only Memory, ROM) 902 or the program loaded from the storage part 908 into the random access memory (Random Access Memory, RAM) 903, such as executing the method in the above embodiment. Various programs and data required for system operation are also stored in the RAM 903. The CPU 901, ROM 902 and RAM 903 are connected to each other via a bus 904. An input / output (I / O) interface 905 is also connected to the bus 904.

[0152] In some embodiments, the following components are connected to the I / O interface 905: an input section 906 including a keyboard, a mouse, and the like; an output section 907 including devices such as a cathode ray tube (CRT), a liquid crystal display (LCD), and a speaker; a storage section 908 including a hard disk and the like; and a communication section 909 including a network interface card such as a LAN (Local Area Network) card or a modem. The communication section 909 performs communication processing via a network such as the Internet. A drive 910 is also connected to the I / O interface 905 as needed. Removable media 911, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, and the like, are installed in the drive 910 as needed so that computer programs read therefrom can be installed into the storage section 908 as needed.

[0153] In particular, according to an embodiment of the present application, the process described above with reference to the flowchart can be implemented as a computer program. For example, an embodiment of the present application includes a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a computer program for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network via the communication section 909, and / or installed from a removable medium 911. When the computer program is executed by the processor (CPU) 901, the various functions defined in the system of the present application are executed.

[0154] It should be noted that the computer-readable medium shown in the embodiment of the present application can be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium can be, for example, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media can include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (Erasable Programmable Read Only Memory), a flash memory, an optical fiber, a portable compact disk read-only memory (Compact Disc Read-Only Memory, CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present application, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, wherein a computer-readable computer program is carried. This propagated data signal can take a variety of forms, including but not limited to an electromagnetic signal, an optical signal, or any suitable combination of the above. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium that can transmit, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device. A computer program embodied on a computer-readable medium may be transmitted using any suitable medium, including but not limited to wireless, wired, or any suitable combination thereof.

[0155] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architectures, functions and operations of the devices, methods and computer program products according to various embodiments of the present application. Among them, each box in the flowchart or block diagram can represent a module, program segment, or part of the code, and the above-mentioned module, program segment, or part of the code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in an order different from that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flowchart, and the combination of boxes in the block diagram or flowchart, can be implemented with a dedicated hardware-based system that performs the specified function or operation, or can be implemented with a combination of dedicated hardware and a computer program.

[0156] The units or modules described in the embodiments of the present application may be implemented in software or hardware, and the units or modules described may also be provided in a processor. The names of these units or modules do not, in certain circumstances, limit the units or modules themselves.

[0157] Another aspect of the present application provides a computer-readable storage medium having a computer program stored thereon. When executed by a processor, the computer program implements the aforementioned business object management method. The computer-readable storage medium may be included in the electronic device described in the above embodiments, or may exist independently and not be incorporated into the electronic device.

[0158] Another aspect of the present application provides a computer program product, comprising a computer program stored in a computer-readable storage medium. A processor of an electronic device reads the computer program from the computer-readable storage medium and executes the computer program, causing the electronic device to perform the business object management method described above in each of the above embodiments.

[0159] It should be noted that, although several modules or units of the device for action execution are mentioned in the above detailed description, this division is not mandatory. In fact, according to the embodiment of the application, the features and functions of two or more modules or units described above can be concretized in one module or unit. On the contrary, the features and functions of one module or unit described above can be further divided into multiple modules or units to be concretized.

[0160] Other embodiments of the present invention will readily occur to those skilled in the art after considering the specification and practicing the embodiments disclosed herein. This application is intended to cover any variations, uses, or adaptations of the present invention that follow the general principles of this application and include common knowledge or customary techniques in the art that are not disclosed herein.

[0161] The above content is only a preferred exemplary embodiment of the present application and is not intended to limit the implementation scheme of the present application. Ordinary technicians in this field can easily make corresponding changes or modifications based on the main concept and spirit of the present application. Therefore, the scope of protection of the present application shall be based on the scope of protection required by the claims.

Claims

1. A business object management method, characterized in that: The method comprises: Obtaining transaction notes for each business transaction generated during the business operation of the business object, where the transaction notes for each business transaction are used to explain the business transaction; Acquiring the amount of virtual resources transferred during the execution of each business transaction; Based on the resource amounts corresponding to the business transactions explained in each transaction memo text, a target transaction memo text is selected from the multiple transaction memo texts; wherein the resource amount corresponding to the target transaction memo text is greater than the resource amounts corresponding to the other transaction memo texts; The target transaction remark text is subjected to type identification processing to obtain the business type to which the business object's operating business belongs.

2. The method according to claim 1, characterized in that Performing type recognition processing on the target transaction note text using a business type recognition model to obtain the business type to which the business object's operating business belongs; The business type recognition model is trained in the following way: Obtaining training transaction remark text for each business transaction generated during the training subject's business operations, as well as a type label for the business operated by the training subject; Performing feature extraction on the training transaction note text to obtain a note feature vector; Performing feature extraction processing on each preset service type to obtain a type feature vector of each preset service type; The business type recognition model is obtained by training based on the remark feature vector, the type feature vectors of each preset business type, and the type label.

3. The method according to claim 2, characterized in that The training based on the remark feature vector, the type feature vectors of each preset business type, and the type label to obtain the business type recognition model includes: Classify the remark feature vector to obtain a prediction result of the type of business operated by the training object; Obtaining a type simulation result of the business operated by the training subject based on the type feature vectors, the remark feature vectors, and the type labels of each preset business type; The probability distribution difference between the type simulation result and the type prediction result is calculated, and training is performed in a direction of reducing the probability distribution difference to obtain the business type recognition model.

4. The method according to claim 3, characterized in that The obtaining of a type simulation result of the business operated by the training object based on the type feature vectors, the remark feature vectors, and the type labels of the respective preset business types includes: Obtaining similarities between the type feature vectors of the respective preset business types and the remark feature vectors, and generating a label confusion vector based on the obtained multiple similarities; Performing label encoding processing on the type label to obtain a label feature vector; The label confusion vector and the label feature vector are subjected to feature fusion to obtain a simulation result of the type of business operated by the training object.

5. The method according to claim 1, wherein The resource amount corresponding to the business transaction based on the explanation of each transaction note text, and screening out the target transaction note text from the multiple transaction note texts, include: For transaction memo texts containing specified characters among the multiple transaction memo texts, a discard operation is performed on the specified characters to obtain multiple updated transaction memo texts; Filtering out transaction memo texts having a text length greater than a preset length from the updated plurality of transaction memo texts as candidate transaction memo texts; The target transaction remark text is screened out from the candidate transaction remark texts based on the resource amount corresponding to the candidate transaction remark texts.

6. The method according to claim 1, characterized in that The method further comprises: Obtain the business type registered by the business object; If the identified business type does not match the registered business type, a management and control operation is performed on the business object's business operations.

7. The method according to claim 1, characterized in that The method further comprises: Obtaining transaction data of business transactions generated during the business operation of the business object; Analyze the transaction data according to preset feature dimensions to obtain transaction feature information of the business object under the preset feature dimensions; Selecting a business type that matches the transaction feature information from a plurality of preset business types; If the selected business type does not match the identified business type, a management and control operation is performed on the business object's business operations.

8. The method according to any one of claims 1 to 7, characterized in that The method further comprises: Periodically performing type identification processing on transaction remark texts of business transactions generated during the business operation of the business object to obtain multiple business types; If it is detected that the multiple business types do not match, a management and control operation is performed on the business object operation business.

9. The method according to any one of claims 1 to 7, characterized in that The type identification processing of the target transaction remark text to obtain the business type of the business object's operating business includes: Performing word segmentation processing on the target transaction remark text to obtain multiple fields; Performing initial encoding processing on each field to obtain a field feature vector of each field; performing feature encoding processing on the field feature vector of each field based on the field feature vectors of other fields in the multiple fields to obtain a text feature vector of the target transaction memo text; The text feature vector is classified to obtain the business type to which the business object operates.

10. A business object management device, characterized in that: The device includes an acquisition unit, a screening unit and a processing unit, wherein: The acquisition unit is used to acquire transaction remark text of each business transaction generated during the business operation of the business object, wherein the transaction remark text of each business transaction is used to explain the each business transaction; The acquisition unit is further configured to acquire the amount of virtual resources transferred during the execution of each business transaction; The screening unit is configured to screen out a target transaction memo text from the plurality of transaction memo texts based on the resource amounts corresponding to the business transactions explained in the respective transaction memo texts; wherein the resource amount corresponding to the target transaction memo text is greater than the resource amounts corresponding to the other transaction memo texts; The processing unit is used to perform type identification processing on the target transaction note text to obtain the business type to which the business object's operating business belongs.

11. A computer-readable medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the business object management method according to any one of claims 1 to 9 is implemented.

12. An electronic device, characterized in that: include: one or more processors; A storage device for storing one or more programs, which, when executed by the one or more processors, enables the one or more processors to implement the business object management method according to any one of claims 1 to 9.

Images