Business type determination method and device, electronic equipment, medium and program product

Abstract

The embodiment of the invention discloses a business type determination method and device, electronic equipment, a medium and a program product, and can be applied to the technical field of artificial intelligence, such as the field of machine learning. According to the embodiment of the invention, subject business data of an operation subject is acquired, the subject business data comprises subject attribute data and transaction data, and the transaction data comprises behavior data and position data; according to an association relationship between the behavior data and the position data, classifying the behavior data and the position data to obtain target behavior data corresponding to the specified transaction mode and transaction position distribution; determining a transaction distribution rule according to the target behavior data and the transaction position distribution; and determining the business type of the operation subject according to the transaction distribution rule and the subject attribute data. According to the scheme, the real business type of the business subject can be determined based on the multi-dimensional business data, so that a business platform can conveniently manage and supervise the business subject.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a method, device, electronic device, medium and program product for determining a business type. Background Art

[0002] In recent years, with the rapid development and popularization of mobile Internet, more and more business platforms have emerged, which have attracted a large number of business entities to settle in. These business platforms can provide a variety of online and offline business environments for the settled business entities, enabling them to more comprehensively meet the needs of consumers and expand their business scope.

[0003] However, in the prior art, business platforms usually determine the business type of business entities based on the settlement information submitted by the business entities. It is difficult for business platforms to verify the actual business type of business entities based on the settlement information, which makes it difficult for business platforms to manage and supervise business entities. Summary of the invention

[0004] The embodiments of the present application provide a business type determination method, device, electronic device, medium and program product, which can determine the true business type of a business entity based on multi-dimensional business data, so that a business platform can manage and supervise the business entity.

[0005] An embodiment of the present application provides a method for determining a business type, comprising: acquiring the main business data of a business entity, the main business data comprising main attribute data and transaction data, the transaction data comprising behavior data and location data; classifying the behavior data and the location data according to the association relationship between the behavior data and the location data to obtain target behavior data corresponding to a specified transaction method, and a transaction location distribution; determining a transaction distribution pattern from the target behavior data and the transaction location distribution; and determining the business type of the business entity according to the transaction distribution pattern and the main attribute data.

[0006] The embodiment of the present application also provides a business type determination device, including: an acquisition unit, used to acquire the main business data of the business entity, the main business data including main attribute data and transaction data, and the transaction data including behavior data and location data; a classification unit, used to classify the behavior data and the location data according to the association relationship between the behavior data and the location data, and obtain target behavior data corresponding to a specified transaction method, and transaction location distribution; a data determination unit, used to determine a transaction distribution pattern from the target behavior data and the transaction location distribution; a type determination unit, used to determine the business type of the business entity according to the transaction distribution pattern and the main attribute data.

[0007] In some embodiments, the acquisition unit is used to: acquire first business data and second business data of the business entity, the first business data being obtained by statistics on current business data in a first statistical period, and the second business data being obtained by statistics on historical business data in a second statistical period, and the first statistical period being shorter than the second statistical period; and obtain the main business data of the business entity from the first business data and the second business data.

[0008] In some embodiments, obtaining the first business data and the second business data of the business entity includes: in response to a type query request for the business entity, obtaining a mapping relationship set and the second business data, the mapping relationship set including a mapping relationship between the first business data and the business entity; obtaining the main business data of the business entity from the first business data and the second business data includes: storing the mapping relationship between the second business data and the business entity in the mapping relationship set, and obtaining the main business data of the business entity.

[0009] In some embodiments, the classification unit is used to: cross-process the behavior data and the location data corresponding to the behavior data to obtain transaction cross-features; and determine the target behavior data corresponding to a specified transaction method and the transaction location distribution based on the transaction cross-features.

[0010] In some embodiments, the cross-processing of the behavior data and the position data corresponding to the behavior data to obtain transaction cross-features includes: encoding the behavior data and the position data corresponding to the behavior data respectively to obtain multiple encoding features; performing linear transformation on the encoding features to obtain linear features; performing nonlinear transformation on the encoding features to obtain nonlinear features; cross-learning any two of the multiple encoding features to obtain cross-features; and performing feature fusion on the linear features, the nonlinear features, and the cross-features to obtain the transaction cross-features.

[0011] In some embodiments, the type determination unit is used to: combine the transaction distribution pattern and attribute characteristics to obtain features to be processed, wherein the attribute characteristics are features extracted from the subject attribute data; classify and transform the features to be processed to obtain business type probability; and determine the business type of the business entity based on the business type probability.

[0012] In some embodiments, the business type determination device also includes a verification unit, which is used to: obtain a pending business relationship from the main business data of the business entity, the pending business relationship including at least one of the business relationship between the business entity and the purchaser, the business relationship between the business entity and the person in charge of business, the business relationship between the business entity and the business equipment, the business relationship between the purchaser and the business equipment, the business relationship between the business entities, and the business relationship between the purchasers; determine the related transaction behavior between the business entity and the purchaser based on the pending business relationship; determine the reference business type of the business entity by the proportion of specified related transaction behaviors in the related transaction behaviors; determine the target business type of the business entity based on the business type of the business entity and the reference business type.

[0013] In some embodiments, determining the related transaction behavior between the business entity and the purchaser based on the pending business relationship includes: taking entities in the pending business relationship as nodes, the entities including the business entity, the purchaser, the business manager, and one or more of the business equipment; grouping the nodes into communities based on the pending business relationship to obtain a grouped community relationship network; and determining the related transaction behavior between the business entity and the purchaser from the community relationship network.

[0014] In some embodiments, determining the target business type of the business entity based on the business type of the business entity and the reference business type includes: if the business type of the business entity does not match the reference business type, obtaining a first probability value corresponding to the business type and a second probability value corresponding to the reference business type; taking a weighted sum of the first probability value and the second probability value to determine a target probability value; and determining the target business type of the business entity based on the target probability value.

[0015] In some embodiments, the business type determination device also includes a training unit, which is used to: obtain a data classification model to be trained, a type classification model to be trained, and a training sample set, wherein the training sample set includes training subject business data of at least one business subject sample, the training subject business data includes training subject attribute data and training transaction data, and the training transaction data includes training behavior data and training position data; through the data classification model to be trained, according to the association relationship between the training behavior data and the training position data, the training behavior data and the training position data are classified to obtain the training target behavior data corresponding to the specified transaction method and the training transaction position distribution; the training target behavior data and the training transaction position distribution are used to classify the training behavior data and the training position data. , combined to obtain a training transaction distribution law; through the type classification model to be trained, according to the training transaction distribution law and the training subject attribute data, determine the training business type of the business subject sample; obtain the training reference business type of the business subject sample; according to the difference between the training reference business type and the training business type, train the model to be trained to obtain a trained model, the model to be trained includes at least one of the data classification model to be trained and the type classification model to be trained, the trained model includes at least one of the trained data classification model and the trained type classification model, the trained data classification model is used to obtain the target behavior data and the transaction location distribution, and the trained type classification model is used to determine the business type of the business subject.

[0016] In some embodiments, the method of training the model to be trained based on the difference between the training reference business type and the training business type to obtain the trained model includes: adjusting the training sample set based on the difference between the training reference business type and the training business type to obtain an adjusted training sample set; and training the model to be trained using the adjusted training sample set to obtain the trained model.

[0017] An embodiment of the present application also provides an electronic device, including a processor and a memory, wherein the memory stores multiple instructions; the processor loads instructions from the memory to execute the steps in any one of the business type determination methods provided in the embodiments of the present application.

[0018] An embodiment of the present application also provides a computer-readable storage medium, which stores a plurality of instructions, and the instructions are suitable for a processor to load to execute the steps in any one of the business type determination methods provided in the embodiments of the present application.

[0019] The embodiment of the present application also provides a computer program product, including a computer program / instruction, which, when executed by a processor, implements the steps in any one of the business type determination methods provided in the embodiment of the present application.

[0020] The embodiment of the present application can obtain the main business data of the operating entity, the main business data includes main attribute data and transaction data, and the transaction data includes behavior data and location data; according to the association relationship between the behavior data and the location data, the behavior data and the location data are classified to obtain the target behavior data corresponding to the specified transaction method and the transaction location distribution; the transaction distribution pattern is determined from the target behavior data and the transaction location distribution; the business type of the operating entity is determined according to the transaction distribution pattern and the main attribute data.

[0021] In this application, based on the correlation between behavior data and location data in the transaction data of the business entity, the behavior data and location data are classified to obtain target behavior data and transaction location distribution that can more accurately reflect the transaction characteristics of the business entity based on the associated, multi-dimensional transaction data. At the same time, the target behavior data and transaction location distribution are used to determine the transaction distribution law, and the transaction distribution law is used to reflect the regular transaction characteristics of the business entity. In this way, this application combines the transaction distribution law and the subject attribute data, and can determine the business type of the business entity through these multi-dimensional data. The business type matches the transaction law and subject attribute of the business entity, and can more accurately characterize the real business type of the business entity, so as to provide a basis for the management and supervision of the business entity, and facilitate the business platform to manage and supervise the business entity. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For those skilled in the art, other drawings can be obtained based on these drawings without creative work.

[0023] Figure 1a is a scenario diagram of a method for determining a business type provided in an embodiment of the present application;

[0024] Figure 1b is a flowchart of a method for determining a business type provided in an embodiment of the present application;

[0025] Figure 1c It is a schematic diagram of the processing flow of the type classification model provided in the embodiment of the present application;

[0026] Figure 2ais a flowchart of a method for determining a business type provided by another embodiment of the present application;

[0027] Figure 2b is a flowchart of another method for determining a business type provided by another embodiment of the present application;

[0028] Figure 2c This is a schematic diagram of a process for obtaining business data of a subject provided in an embodiment of the present application;

[0029] Figure 2d It is a schematic diagram of the transaction data classification processing flow provided by the embodiment of the present application;

[0030] Figure 2e It is a schematic diagram of the combined processing flow provided by the embodiment of the present application;

[0031] Figure 2f is a schematic diagram of a processing flow of a type classification model provided by another embodiment of the present application;

[0032] Figure 2g It is a schematic diagram of a process for verifying a business type provided in an embodiment of the present application;

[0033] Figure 3 is a schematic diagram of the structure of a business type determination device provided in an embodiment of the present application;

[0034] Figure 4 It is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0035] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of this application.

[0036] Embodiments of the present application provide a method, device, electronic device, medium, and program product for determining a business type.

[0037] The business type determination device may be integrated in an electronic device, which may be a terminal, a server, or other equipment. The terminal may be a mobile phone, a tablet computer, a smart Bluetooth device, a laptop, or a personal computer (PC); the server may be a single server or a server cluster consisting of multiple servers.

[0038] In some embodiments, the business type determination device may also be integrated into multiple electronic devices. For example, the business type determination device may be integrated into multiple servers, and the business type determination method of the present application may be implemented by multiple servers.

[0039] In some embodiments, the server may also be implemented in the form of a terminal.

[0040] For example, refer to Figure 1a The business type determination method can be integrated in a server, which can obtain a type query request for a business entity from a client, and in response to the type query request, obtain the main business data of the business entity, the main business data includes main attribute data and transaction data, the transaction data includes behavior data and location data; according to the association relationship between the behavior data and the location data, the behavior data and the location data are classified to obtain the target behavior data corresponding to the specified transaction method and the transaction location distribution; the transaction distribution law is determined from the target behavior data and the transaction location distribution; according to the transaction distribution law and the main attribute data, the business type of the business entity is determined, and the determined business type is returned to the client.

[0041] The following are detailed descriptions. It should be noted that the order of the following embodiments is not intended to limit the preferred order of the embodiments. It is understood that in the specific implementation of the present application, data related to business data, transaction data, behavior data, location data, subject attribute data, business relationships, etc., which are related to users such as business entities or purchasers, are involved. When the embodiments of the present application are applied to specific products or technologies, user permission or consent is required, and the collection, use and processing of relevant data must comply with relevant laws, regulations and standards of relevant countries and regions.

[0042] Artificial Intelligence (AI) is a technology that uses digital computers to simulate human perception of the environment, acquire knowledge, and use knowledge. This technology can enable machines to have functions similar to human perception, reasoning, and decision-making. The basic technologies of artificial intelligence generally include sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, mechatronics, and other technologies. Artificial intelligence software technology mainly includes computer vision technology, speech processing technology, natural language processing technology, as well as machine learning / deep learning, autonomous driving, smart transportation, and other major directions.

[0043] Computer Vision (CV) is a technology that uses computers to replace human eyes to identify, measure and further process target images. Computer vision technology usually includes image processing, image recognition, image semantic understanding, image retrieval, virtual reality, augmented reality, simultaneous positioning and map construction, autonomous driving, smart transportation and other technologies, as well as common biometric recognition technologies such as face recognition and fingerprint recognition. For example, image processing technologies such as image coloring and image stroke extraction.

[0044] Natural language processing (NLP) is an important direction in the fields of computer science and artificial intelligence. It studies various theories and methods that can achieve effective communication between people and computers using natural language. Natural language processing is a science that integrates linguistics, computer science, and mathematics. Therefore, research in this field will involve natural language, that is, the language people use in daily life, so it is closely related to the study of linguistics. Natural language processing technology usually includes text processing, semantic understanding, machine translation, robot question answering, knowledge graph and other technologies.

[0045] Machine Learning (ML) is a multi-disciplinary subject that involves probability theory, statistics, approximation theory, convex analysis, algorithm complexity theory and other disciplines. It specializes in studying how computers simulate or implement human learning behavior to acquire new knowledge or skills and reorganize existing knowledge structures to continuously improve their performance. Machine learning is the core of artificial intelligence and the fundamental way to make computers intelligent. Its applications are spread across all areas of artificial intelligence. Machine learning and deep learning usually include artificial neural networks, belief networks, reinforcement learning, transfer learning, inductive learning, and teaching learning.

[0046] With the research and advancement of artificial intelligence technology, artificial intelligence technology has been studied and applied in many fields, such as common smart homes, smart wearable devices, virtual assistants, smart speakers, smart marketing, unmanned driving, automatic driving, drones, robots, smart medical care, smart customer service, Internet of Vehicles, smart transportation, etc. I believe that with the development of technology, artificial intelligence technology will be applied in more fields and play an increasingly important role.

[0047] In this embodiment, a business type determination method involving artificial intelligence is provided, such as Figure 1b As shown, the specific process of the business type determination method can be as follows:

[0048] 110. Obtain the main business data of the operating entity. The main business data includes the main attribute data and transaction data. The transaction data includes behavior data and location data.

[0049] Among them, the operating entity refers to an entity or organization that conducts business activities, which can be an individual, enterprise, institution or other entity, and the business activity can be called a transaction activity. The entity or organization that conducts transactions with the operating entity can be called the purchaser, that is, the operating entity and the purchaser are the seller and buyer in the transaction respectively. In some embodiments, the operating entity can be a merchant or merchant stationed on the business platform, which can display the products or services they provide through the business platform and use the platform resources and user base to conduct online or offline transactions.

[0050] Among them, business data refers to data related to the business activities of the business entity, and subject business data refers to the business data of the business entity. Business data may include but is not limited to subject attribute data and transaction data, and business data may be in the form of text, images or numerical values.

[0051] Among them, subject attribute data refers to data that describes the basic characteristics of the business entity. For example, subject attribute data may include but is not limited to identity information such as name, person in charge, business address, establishment date, and one or more other attribute-related data.

[0052] Among them, transaction data refers to data that records and describes transaction activities. For example, transaction data may include but is not limited to behavioral data and location data. Among them, behavioral data refers to the details of transaction behaviors between business entities and buyers and other relevant information. For example, behavioral data may include but is not limited to one or more of behavioral data such as transaction time, transaction services or products, transaction amount, number of transactions, and transaction channels. Among them, location data refers to information that records the location where the transaction behavior occurs. Location data can be expressed in the form of geographic coordinates, address information, or other forms of information that can characterize the location. For example, location data may include but is not limited to one or more of business entity transaction location, business entity transaction location distribution, buyer transaction location, buyer transaction location distribution, etc.

[0053] The embodiments of the present application can be applied to various types of business platforms, such as catering platforms, e-commerce platforms, education platforms, travel platforms, health and fitness platforms, life information and transaction platforms, payment platforms, live broadcast platforms, etc., or mixed types of platforms such as social e-commerce platforms, health technology platforms, education and entertainment platforms, and life service platforms. For different types of business platforms, the embodiments of the present application can be applied in different scenarios, such as operation entity supervision, platform marketing, and operation entity differentiated management.

[0054] It should be noted that when the embodiments of the present application are applied to specific products or technologies, obtaining the main business data of the operating entity and the subsequent processing of the main business data require the permission or consent of the operating entity and its related purchasers, and the collection, use and processing of relevant data need to comply with the relevant laws, regulations and standards of relevant countries and regions. For example, when obtaining the permission or consent of the settled operating entity and its related purchasers, the business platform can obtain the main business data of the operating entity.

[0055] In some implementations, current business data can be counted based on a shorter statistical period to provide real-time and fine-grained data features, and historical business data can be counted based on a longer statistical period to provide long-term and overall data features. In this way, a variety of business data with multi-angle data features can be obtained based on different statistical periods to obtain more accurate and comprehensive business data. Specifically, obtaining the main business data of the business entity includes:

[0056] Acquire first business data and second business data of the business entity, where the first business data is obtained by counting current business data in a first statistical period, and the second business data is obtained by counting historical business data in a second statistical period, and the first statistical period is shorter than the second statistical period;

[0057] The main business data of the operating entity is obtained from the first business data and the second business data.

[0058] The statistical period refers to the time span (i.e., time interval) for statistical business data, which can be one day, one week, one month, or any other time period. The first statistical period and the second statistical period are statistical periods corresponding to different time intervals. Specifically, the time interval corresponding to the first statistical period is shorter than that of the second statistical period. For example, the first statistical period can be a period for statistical analysis according to one hour, and the second statistical period can be a period for statistical analysis according to one week.

[0059] Among them, current business data refers to business data at the current time point or in the recent period of time. The recent period of time may vary according to needs and application scenarios. For example, current business data may be business data for the most recent day or week, or business data within a specific time range such as the most recent activity period. Historical business data refers to business data for a period of time in the past. The period of time in the past may vary according to needs and application scenarios. For example, historical business data may be business data for the past year, the past quarter, or before the current time point.

[0060] For example, one hour can be used as the first statistical period to count the business data (current business data) within the most recent day (i.e., 24 hours) to obtain 24 groups of business data (i.e., first business data), where each group of business data is a first statistical period, such as business data updated within one hour. For example, within any hour, the subject attribute data such as the name of the business entity is changed, and n transaction activities occur. Then, the changed business entity name and the transaction data of the n transaction activities can be used as a group of business data obtained by statistics, such as {"business entity name":"changed name","transaction data 1","transaction data 1",…,"transaction data n"}. The transaction data of any transaction activity can be "merchant 1 and buyer 2 have a transaction at time t, the transaction amount is n, the merchant transaction location is A, the buyer transaction location is B, the transaction channel is C, the transaction product is D, and the number of transaction products is l". And take one week as the second statistical period, and collect statistics on all historical data to obtain N groups of business data (i.e., second business data), where N is a positive integer greater than one, and each group of business data is business data updated within a second statistical period, such as one week. The 24 groups of business data are combined with the N groups of business data to obtain 24+N groups of business data, which are the business data of the acquired business entity.

[0061] In some implementations, statistical methods such as Max (maximum value) and / or Min (minimum value) may also be used to collect statistics on the business data updated in the first statistical period and / or the second statistical period, and the statistical data may be used as the first business data and / or the second business data in the statistical period. For example, the first business data corresponding to any first statistical period may be expressed as "merchant 1 from time t to time t +1 There are n transactions, m transaction amounts, and x buyers.

[0062] In some implementations, the first business data may be obtained by sliding a time window to perform real-time statistics on the business data, so as to quickly provide instant first business data. Specifically, the first statistical period includes an event window, and the first business data is obtained by the following steps:

[0063] Through the time window, the current business data is subjected to streaming statistics to obtain the first business data.

[0064] The time window refers to the time range or time period used when counting and analyzing data.

[0065] Among them, streaming statistics is a real-time calculation method based on data streams. It can perform continuous and real-time statistical calculations on data and is suitable for situations with shorter statistical cycles.

[0066] For example, in actual applications, since the business data of the business entity is constantly changing, that is, new business data is constantly generated, the current business data can be business data that continuously arrives in the form of a stream within a specified time period, and the specified time period can be a period of time starting from the current time point, such as 24 hours. At this time, a fixed-size time window can be defined as the first statistical period, and the business data flow back can be divided according to the size of the time window, and statistics are performed in each time window to obtain a set of business data within the time window. Within the specified time period, as new business data arrives, the time window will move forward, discard past data, and count the business data newly entering the window until the end of the specified time period.

[0067] In some implementations, the first business data may be stored in a mapping relationship set, which includes a mapping relationship between the first business data and the business entity, so that it can be called at any time.

[0068] The mapping relationship set refers to a set of elements with a mapping relationship, and the mapping relationship consists of key-value pairs, where each key corresponds to a unique value.

[0069] For example, the identification information of the business entity can be used as a key, and the first business data can be used as a value to construct a corresponding mapping relationship and store it in a mapping relationship set. For example, the mapping relationship set can be expressed as {"identification information": first business data 1, "identification information": first business data 2, "identification information": first business data 3, ...}. The identification information can include, but is not limited to, name, ID, registration number or other information that can be used to uniquely identify and recognize a business entity. Each group of business data in the first business data can also be used as a value to construct a corresponding mapping relationship and store it in a mapping relationship set, so that the mapping relationship set includes multiple mapping relationships, and the key of each mapping relationship is the identification information of the business entity, and the value is a group of business data within a first statistical period, so as to facilitate reading and modifying business data in different statistical periods.

[0070] In some implementations, the first business data may be stored in a real-time database to quickly provide instant first business data, and the second business data may be stored in an offline database to ensure that the data are stored securely and reliably.

[0071] Among them, the real-time database refers to a database system used to store and manage real-time data, and the offline database refers to a database system used to store and manage offline data, such as a static database. Compared with the real-time database, the offline database is mainly used to process and store data that does not need to be accessed or updated in real time.

[0072] For example, in actual applications, a set of business data within a time window can be obtained by counting the real-time data stream of the current business data, a mapping relationship can be established between the set of business data and the identification information of the business entity, and the mapping relationship can be stored in a mapping relationship set stored in a real-time database. As the time window moves, a set of business data obtained by each movement statistics is stored in real time in the form of a mapping relationship in a mapping relationship set stored in the real-time database. At the same time, after the business data is obtained, the business data can be stored in an offline database, and when necessary, the historical business data stored in the offline database can be counted according to a second statistical period to obtain second business data.

[0073] In some implementations, the business data of the business entity may be stored in the form of a mapping relationship set, so that various related business data of the business entity can be called from the mapping relationship set in the subsequent processing process to avoid data errors. Specifically, obtaining the first business data and the second business data of the business entity includes:

[0074] In response to a type query request for a business entity, a mapping relationship set and second business data are acquired, where the mapping relationship set includes a mapping relationship between the first business data and the business entity;

[0075] The main business data of the operating entity is obtained from the first business data and the second business data, including:

[0076] The mapping relationship between the second business data and the business entity is stored in a mapping relationship set to obtain the main business data of the business entity.

[0077] Among them, the type query request refers to a request for querying the business entity, which can be implemented through different methods or parameters, such as GET request, POST request, query parameters, etc. In actual applications, in different scenarios, with the consent or permission of the querying party, the business entity and the purchaser, the type query request can be generated based on different needs. For example, the type query request can be generated based on the supervision needs of the business platform regulator, the query needs of the user, or the query needs of the business platform operator based on marketing activities.

[0078] For example, when the operator of a business platform organizes a marketing campaign, the operator can, after obtaining the consent or permission of the business entity and the buyer, generate a type query request for one or more business entities through the query entry provided by the business platform operation background, so that the server of the business platform can obtain the main business data of the one or more business entities according to the type query request, determine the business type of each business entity based on the main business data of the business entity, and return the business type of the one or more business entities to the business platform operation background, so that the operator can formulate a corresponding marketing plan based on the business type.

[0079] Specifically, taking the type query request for business entity A as an example, the server of the business platform can obtain the mapping relationship set A corresponding to the identification information A from the real-time database according to the identification information A of the business entity A carried in the type query request, and obtain the second business data corresponding to the identification information A from the offline database, and use the identification information A as the key and the second business data as the value to construct the corresponding mapping relationship and store it in the mapping relationship set. It is also possible to use each group of business data in the second business data as a value, respectively, to construct a corresponding mapping relationship and store it in the mapping relationship set, so that the mapping relationship set includes multiple mapping relationships, such as the mapping relationship set can be expressed as {"identification information": first business data 1, "identification information": first business data 2, "identification information": first business data 3,..., "identification information": second business data 1, "identification information": second business data 2, "identification information": second business data 3,...,}. The key of each mapping relationship is the identification information of the business entity, and the value is a group of business data in a first statistical period or a second statistical period, so as to facilitate reading and modifying business data in different statistical periods.

[0080] 120. According to the correlation between the behavior data and the location data, the behavior data and the location data are classified to obtain the target behavior data corresponding to the specified transaction method and the transaction location distribution.

[0081] Among them, the transaction method refers to the method by which the business entity conducts transactions, and the designated transaction method refers to the transaction method set according to actual needs or application scenarios. The target behavior data is the behavior data corresponding to the designated transaction method. In some embodiments, in order to better classify the behavior data corresponding to the business entity, the behavior data can be divided into two categories: online transactions and offline transactions, that is, the designated transaction method includes online transactions and offline transactions.

[0082] Among them, the transaction location distribution refers to the distribution of transaction activities in the geographical space. The transaction location distribution can describe the statistical data of the distribution of the place or region where the transaction occurs, such as regional distribution, city distribution, business district distribution, customer group distribution, etc. The transaction location distribution can characterize the geographical distribution characteristics of the transaction. For example, the regional distribution, city distribution, business district distribution, customer group location distribution, etc. can be expressed as the transaction location distribution of different regions, different cities, different business districts, and different customer groups. For example, the regional distribution can be expressed as {"region 1": location 1, location 2, ... location m, ..., "region k": location n+1, location n+2, ... location n+3, ...,}, and the customer group distribution can be expressed as {"customer group 1": location 1, location 2, ... location m, ..., "customer group k": location n+1, location n+2, ... location n+3, ...,}, where the location can be expressed as geographic coordinates and address information. A customer group refers to a group of consumers with similar needs, behavioral characteristics, and demographic characteristics. The customer group type divides buyers into types based on their needs, behavioral characteristics, and demographic characteristics. For example, with the buyer's consent or permission, they can be divided into loyal buyers, price-sensitive buyers, research buyers, and other customer groups based on their consumption habits.

[0083] In some embodiments, the location data includes the transaction location of the business entity and the transaction location of the buyer, and the behavior data includes the transaction channel.

[0084] Among them, the transaction channel refers to the channel through which the business entity conducts transactions. In actual applications, the business entity can conduct transactions with the purchaser through a variety of transaction channels, such as transaction channels that may include but are not limited to one or more of transactions through storefronts, transactions through mini-programs, transactions through links, transactions through applications, etc. Among them, storefront transactions may include different transaction channels such as scanning code transactions and transactions through payment codes.

[0085] For example, the behavior data of transactions through store channels can be determined as the target behavior data corresponding to offline transactions, and the behavior data of transactions through mini-programs, through links, and through applications can be determined as the target behavior data corresponding to online transactions. In addition, online transactions and offline transactions can be distinguished based on the transaction location of the business entity and the transaction location of the buyer. If the transaction location of the business entity and the transaction location of the buyer corresponding to any behavior data are in the same city, the behavior data can be determined as the target behavior data corresponding to offline transactions. If the transaction location of the business entity and the transaction location of the buyer corresponding to any behavior data are in different cities, the behavior data can be determined as the target behavior data corresponding to online transactions. For another example, the location distribution of the transaction, i.e., the transaction location distribution, can be determined based on the transaction location of the business entity and the transaction location of the buyer. For example, the transaction location distribution may include but is not limited to the transaction buyer location distribution {"buyer A": location 1, location 2, ... location m, ...,}, the business owner location distribution {"business entity B": location 1, location 2, ... location n, ...,}, etc. The transaction locations of the buyers and the transaction locations of the business entities may also be further subdivided based on different regions, different cities, different business districts, and different customer groups to obtain the buyer location distribution or the business entity location distribution based on regional distribution, city distribution, and business district distribution, such as, or obtain the customer group location distribution.

[0086] In practical applications, since a business entity may have multiple branch business entities, such as a merchant may have multiple branches, in some implementations, for a business entity with multiple branch business entities, the total transaction location distribution of its business entities and the branch transaction location distribution may be determined such as {"branch business entity 1": location 1, location 2, ... location m, ..., "branch business entity k": location n+1, location n+2, ... location n+3, ...,}.

[0087] Among them, the total transaction location distribution of the operating entity refers to the distribution of all transaction locations of the operating entity, such as the total transaction location distribution can be expressed as {"operating entity": position 1, position 2,…position n,...,}; the branch transaction location distribution refers to the transaction location distribution determined based on the branch business type, such as the branch transaction location distribution can be expressed as {"branch operating entity 1": position 1, position 2,…position m,..., "branch operating entity k": position n+1, position n+2,…position n+3,...,}.

[0088] Since the transaction data of the business entity may not necessarily accurately reflect the transaction characteristics, for example, the transaction behavior of scanning the code to pay in the store is a characteristic of offline transactions, but the accuracy of this characteristic is not high, so it is necessary to combine the behavioral data of the transaction behavior with the location data to obtain a more accurate classification result. It can be understood that by cross-processing the behavior data and the location data and then classifying them, the transaction data can be analyzed and mined in multiple dimensions to identify the correlation between the transaction behavior and the transaction location, and obtain more accurate target behavior data and transaction location distribution, which can more accurately reflect the transaction characteristics of the business entity. In some embodiments, the behavior data and location data can be cross-analyzed to learn the interaction and importance between the characteristics of the data, and increase the accuracy of the determined target behavior data and transaction location distribution. Specifically, according to the correlation between the behavior data and the location data, the behavior data and the location data are classified to obtain the target behavior data and the transaction location distribution corresponding to the specified transaction method, including:

[0089] Cross-processing the behavior data and the location data corresponding to the behavior data to obtain transaction cross-features;

[0090] According to the transaction cross-features, the target behavior data corresponding to the specified transaction method and the transaction location distribution are determined.

[0091] Among them, cross processing refers to the process of processing cross data obtained by associating and comparing different types of data. The data result obtained through data cross processing is the transaction cross feature, which contains the cross information of behavior data and location data.

[0092] In some embodiments, the behavior data and the location data corresponding to the behavior data may be the behavior data and location data corresponding to the same transaction activity in the transaction data, that is, the behavior data and location data generated when the business entity conducts a transaction with the buyer.

[0093] For example, the behavioral data and location data of any transaction activity in the transaction data can be cross-processed by feature cross-models, multi-head attention models, etc., to obtain the transaction cross-features of the transaction activity. Then, the transaction cross-features are predicted by regression models or classification models to determine the prediction results corresponding to the transaction cross-features. The prediction results can be binary classifications, such as determining the specified transaction mode of any transaction activity, such as offline transactions and online transactions, or multi-classification data, such as determining the transaction location classification of the business entity or the transaction location classification of the buyer in any transaction activity. For example, the location classification results can be set according to actual needs or business scenarios, such as the transaction location can be classified according to different regions, different cities, different business districts, and different customer groups. According to the transaction location classification of the business entity or the transaction location classification of the buyer of all transaction activities in all transaction data, the transaction location distribution of the business entity or the transaction location distribution of the buyer is determined, such as the multi-classification results according to different regions, the regional distribution, {"region 1": position 1, position 2,... position m,..., "region k": position n+1, position n+2,... position n+3,...,} is determined.

[0094] In some embodiments, the behavior data and the location data corresponding to the behavior data may be the behavior data and location data obtained by statistics in the same statistical period (such as the first statistical period or the second statistical period) in the transaction data. That is, for any group of business data in the mapping relationship set, the behavior data and location data in the group of business data can be cross-processed through a feature cross-model, a multi-head attention model, etc., to obtain the transaction cross-features of the group of business data, so as to determine the target behavior data corresponding to the specified transaction method in the group of business data, as well as the transaction location classification of the business entity or the transaction location classification of the buyer. Then, according to the transaction location classification of the business entity or the transaction location classification of the buyer in all business data groups, the transaction location distribution of the business entity or the transaction location distribution of the buyer is determined.

[0095] In some embodiments, the behavior data and location data can be classified by traversing the mapping relationship set of the business entity. For example, by traversing all the mapping relationships in the mapping relationship set corresponding to the business entity, such as {"identification information": first business data 1, "identification information": first business data 2, "identification information": first business data 3,..., "identification information": second business data 1, "identification information": second business data 2, "identification information": second business data 3,...,}, the transaction data in the value of each mapping relationship can be accessed to avoid data omissions. When accessing transaction data, the target behavior data corresponding to the specified transaction method in the group of business data, as well as the business entity transaction location classification or the buyer transaction location classification, etc., are determined, and then all data are merged to obtain the target behavior data and transaction location distribution in all transaction data.

[0096] In some implementations, the behavior data and location data may be classified by multiple parallel processing nodes to obtain target behavior data and transaction location distribution, so as to improve the efficiency of transaction data classification. Specifically, the transaction data in the mapping relationship set may be divided into multiple data blocks, where one data block includes one or more groups of business data, and multiple processing nodes may process multiple data blocks in parallel to determine the target behavior data corresponding to the specified transaction method in the group of data blocks, as well as the transaction location classification of the operating entity or the transaction location classification of the buyer, and then all the data may be merged to obtain the target behavior data and transaction location distribution in all transaction data.

[0097] In some implementations, first-order, second-order, and higher-order feature representations can be obtained through linear transformation, cross-learning, and nonlinear transformation, respectively, so as to comprehensively consider the interaction between low-order and high-order features during the cross-processing process to increase the accuracy of the determined data feature representation. Specifically, cross-processing is performed on the behavior data and the location data corresponding to the behavior data to obtain transaction cross-features, including:

[0098] Encoding the behavior data and the position data corresponding to the behavior data to obtain multiple encoding features;

[0099] Perform linear transformation on the encoded features to obtain linear features;

[0100] Perform nonlinear transformation on the encoded features to obtain nonlinear features;

[0101] Cross-learn any two coding features among the multiple coding features to obtain cross-features;

[0102] Linear features, nonlinear features and cross features are fused to obtain transaction cross features.

[0103] Among them, linear transformation refers to transforming data in a linear relationship, for example, linear transformation can be performed by addition, multiplication, etc. Nonlinear transformation refers to transforming data in a nonlinear relationship, for example, activation functions (such as ReLU function, sigmoid function, etc.) can be used to introduce nonlinear features to the encoding features to learn more complex features.

[0104] Among them, cross learning refers to the calculation of cross terms between features, which captures the interaction between features by calculating the cross terms between features. For example, n encoding features can be combined in pairs to obtain (n×(n-1)) / 2 cross features, each of which is calculated by the dot product of two different encoding features.

[0105] Feature fusion refers to combining multiple features. For example, multiple features can be fused by concatenation, weighting, intersection, transformation or mapping to obtain a feature representation with comprehensive expression capabilities.

[0106] For example, one or more groups of behavior data and location data in each transaction activity can be obtained from the transaction data of the business entity, such as M groups of behavior data and N groups of location data. One group of data corresponds to one type of data. The M groups of behavior data may include transaction channels, transaction services or products, transaction amounts, number of transactions and other types of data. The N groups of location data include transaction locations of the business entity, transaction locations of the buyer and other types of data. Each group of data can be encoded separately to obtain a coding feature. For example, M groups of behavior data and N groups of position data can be encoded to obtain M+N coding features. Each coding feature can be linearly transformed by weighted summation to obtain M+N linear features. Each coding feature can be nonlinearly transformed by an activation function to obtain M+N nonlinear features. Click calculations can be performed on each of the M+N coding features to obtain ((M+N)×((M+N)-1)) / 2 cross features. The M+N linear features, the M+N nonlinear features and the ((M+N)×((M+N)-1)) / 2 cross features can be accumulated, such as summed or weighted summed, to obtain transaction cross features.

[0107] In some embodiments, the behavior data and location data can be divided into discrete data (such as location data, transaction method, transaction channel, transaction service or product, etc.) and continuous data (such as transaction time, transaction amount, etc.) according to their data features, and encoded separately. The encoding feature includes a first encoding feature corresponding to the discrete data and a second encoding feature corresponding to the continuous data. The total encoding feature can be linearly transformed and cross-learned to obtain linear features and cross-features, wherein the total encoding feature is obtained by combining the first encoding feature and the second encoding feature, and the first encoding feature is nonlinearly transformed to obtain a nonlinear feature.

[0108] In some embodiments, a feature cross model can be used as a data classification model to perform data cross processing on behavior data and location data to obtain transaction cross features. The feature cross model can include a deep neural network and a cross layer, which are responsible for the extraction of low-order features and high-order features respectively. Specifically, the feature cross model includes a coding layer, a cross layer, a deep neural network and an output layer. Among them, the coding layer includes an embedding layer and a sparse feature layer. For example, the behavior data and the location data can be preprocessed to extract discrete features (such as location data, transaction methods, transaction channels, transaction services or products, etc.) and continuous features (such as transaction time, transaction amount, etc.) from them, and the discrete features, etc. are encoded by the sparse feature layer to convert them into multiple columns, so that the feature matrix of the obtained feature vector becomes sparse, and the embedding layer can receive the sparse features output by the sparse feature layer, and embed the high-dimensional sparse binary features to obtain low-dimensional dense features. Then, the low-dimensional dense features corresponding to any behavior data or any location data and the corresponding continuous data are spliced ​​to obtain the coding features corresponding to the behavior data or location data. The coded features are then input into the cross layer, and weights are assigned to the coded features through the linear part of the cross layer for weighted summation to obtain first-order linear features; the coded features are multiplied two by two through the cross part of the cross layer, and then weights are assigned for weighted summation to obtain second-order cross features; the cross layer accumulates the linear features and the cross features to obtain the output features of the cross layer. The low-dimensional dense features obtained by the embedding layer can be input into a deep neural network, which includes multiple fully connected layers. The multiple fully connected layers perform multiple nonlinear transformations on the low-dimensional dense features through activation functions to obtain nonlinear features. The output features of the cross layer and the nonlinear features are then accumulated to obtain transaction cross features. The output layer then predicts the transaction cross features to determine the prediction results corresponding to the transaction cross features.

[0109] In some embodiments, the feature cross model can be a DeepFM model (deep factorization machine model), which includes a deep neural network and a factorization machine, which are responsible for extracting low-order features and high-order features respectively. Specifically, the DeepFM model includes a coding layer, an FM layer (factorization machine layer), a DNN layer (deep neural network) and an output layer, wherein the FM layer is a cross layer and the DNN layer is a deep neural network.

[0110] 130. Determine the transaction distribution pattern based on the target behavior data and transaction location distribution.

[0111] The transaction distribution law refers to data used to describe the distribution law of transaction activities. For example, the transaction distribution law may include transaction frequency, transaction amount distribution law, transaction location distribution law, transaction time distribution law, transaction method distribution law, etc. The transaction location distribution law or transaction time distribution law may be expressed in the form of position series and time series.

[0112] For example, all target behavior data mined from transaction data and all transaction location distributions can be jointly counted to conduct multi-dimensional analysis and mining of transaction data, thereby obtaining transaction distribution patterns that can more accurately reflect the transaction patterns of business entities.

[0113] In some embodiments, basic transaction patterns such as transaction frequency distribution patterns, transaction amount distribution patterns, and transaction area distribution patterns can be determined by statistically analyzing target behavior data and transaction location distribution. For example, the transaction frequency distribution patterns and transaction amount distribution patterns can be respectively expressed as transaction frequencies and transaction amounts in different regions, cities, business districts, and customer groups, such as {"region 1": transaction frequency m,..., "region k": transaction frequency n,...,} and {"business district 1": transaction amount m,..., "business district k": transaction amount n,...,}.

[0114] In some implementations, the distribution of transaction locations corresponding to the target behavior data of different transaction methods can be statistically analyzed to determine the distribution rules of transaction methods, such as the location distribution rules of online transactions, the location distribution rules of offline transactions, and the location distribution rules of different transaction channels. For example, the location distribution rules of online transactions can be expressed as {"region 1": online transaction frequency m,..., "region k": online transaction frequency n,...,}, and the location distribution rules of different transaction channels can be expressed as {"transaction channel 1": transaction frequency m,..., "transaction channel k": transaction frequency n,...,}, and each transaction channel corresponds to a specified transaction method.

[0115] In some embodiments, the location distribution law of the business entity and the location distribution law of the buyer can be statistically analyzed to determine a more accurate location distribution feature of the business entity or the buyer. For example, the location distribution law may include the transaction location distribution law of the business entity, such as {"region 1": transaction frequency m of the business entity, ..., "region k": transaction frequency n of the business entity, ...,}, and the location distribution law may also be the joint location distribution law of the transaction location of the buyer, such as {"region 1": transaction frequency m1 of the business entity, transaction frequency m2 of the buyer, "region 2": transaction frequency n1 of the business entity, transaction frequency n2 of the buyer, ...,}

[0116] In some implementations, sequence features such as transaction timing sequence and transaction position sequence can be obtained based on the regularity of transaction time statistical behavior data and the regularity of position data. For example, the transaction timing sequence {transaction time 1, transaction time 2, transaction time 3...,}, the transaction position sequence {position 1, position 2, position 3...,}, the transaction time and position in the sequence are arranged in the order of occurrence of the corresponding transaction activities. The transaction timing sequence and transaction position sequence can also be obtained by arranging the transaction frequency or transaction amount at different transaction times or positions, such as the transaction timing sequence {transaction frequency 1, transaction frequency 2, transaction frequency 3...,}, which represents a sequence composed of transaction frequencies in different time periods, and the transaction position sequence {transaction frequency 1, transaction frequency 2, transaction frequency 3...,}, which represents a sequence composed of transaction frequencies at different transaction positions.

[0117] 140. Determine the business type of the operating entity based on transaction distribution patterns and entity attribute data.

[0118] The business type refers to the specific business type that the business entity is engaged in. For example, for merchants or businesses that have settled in the business platform, the business type may include but is not limited to online business type, offline business type, and offline and online mixed business type.

[0119] For example, based on the judgment conditions such as keywords, specific industry standards, and subject attributes, the transaction distribution patterns corresponding to the business entity can be matched with the subject attribute data to classify the business entity into different business types. Alternatively, a machine learning model can be used to input the transaction distribution patterns and subject attribute data corresponding to the business entity as input features, and input into a pre-trained type classification model to predict the business type of the business entity. Alternatively, cluster analysis can be performed on the transaction distribution patterns and subject attribute data corresponding to the business entity to determine the group of business entities, which can represent the business type of the business entity.

[0120] In some embodiments, the business type can be subdivided into multiple dimensions according to a variety of different type elements. For example, the type elements can include but are not limited to one or more of online business type, offline business type, online and offline mixed business type, single-branch business type, multi-branch business type, same-location transaction business type, different-location transaction business type, and multi-location transaction business type. These type elements can be combined to obtain different business types according to actual needs or application scenarios.

[0121] Among them, online means that the core transaction method of the business entity is online transaction, offline means that the core transaction method of the business entity is offline transaction, single branch means that the business entity has only one branch business type, multi-branch means that the business entity has multiple branch business types, online and offline mixed means that the core transaction method of the business entity includes online transactions and offline transactions, same-location transaction means that the transaction location of the business entity and the transaction location of its buyer are in the same place such as the same city or the same business district, different-location transaction means that the transaction location of the business entity and the transaction location of its buyer are in different places, and multi-location transaction means that the transaction location of the buyer of the business entity is located in multiple places.

[0122] For example, the business types may be combined to obtain online-single-branch business type, online-multi-branch business type, offline-single-branch business type, offline-multi-branch business type, etc., or further subdivided into online-single-branch-same-place transaction business type, online-single-branch-different-place transaction business type, online-multi-branch-same-place transaction business type, online-multi-branch-different-place transaction business type, offline-single-branch-same-place transaction business type, offline-single-branch-different-place transaction business type, offline-multi-branch-same-place transaction business type, offline-multi-branch-different-place transaction business type, online-and-offline mixed-single-branch-same-place transaction business type, online-and-offline mixed-single-branch-different-place transaction business type, online-and-offline mixed-multi-branch-same-place transaction business type, online-and-offline mixed-multi-branch-different-place transaction business type, etc.

[0123] In actual business operations, the subject attribute data of the business entity can be in the form of images, texts, etc. For example, the subject attribute data can include identity information in the form of image data such as storefront photos, business scene photos, business licenses, etc. The subject attribute data can also include identity information in the form of text such as the subject name and registration number. The subject attribute data can also include information in the form of text such as service or product names, or other transaction notes. In this way, in some embodiments, attribute features can be extracted from subject attribute data in different forms to obtain consistent and quantified data, which is convenient for subsequent processing and improves the efficiency and accuracy of determining the business type. Specifically, before determining the business type of the business entity based on the transaction distribution law and the subject attribute data, it also includes:

[0124] According to the data form of the subject attribute data, attribute features are extracted from the subject attribute data.

[0125] Among them, the attribute feature refers to the feature representation of the subject attribute extracted from the subject attribute data.

[0126] For example, computer vision technologies such as CNN network (convolutional neural network) and ViT algorithm (visual Transformer algorithm) can be used to identify and analyze image data in subject attribute data, such as storefront photos or business scene photos to identify the business location type, store name and other business entity identity information, and convert it into a first feature in vector form. Text processing methods such as Wrod2Vec (word vector algorithm) and BERT model (Bidirectional encoder representation based on Transformer) can be used to extract text data in subject attribute data such as company name, product name and transaction notes, and convert it into a second feature in vector form. The first feature and the second feature are concatenated to obtain the attribute feature.

[0127] In some implementations, the transaction distribution pattern and the subject attribute data can be combined to obtain comprehensive and multi-dimensional features to be processed for the business subject, and then the business type of the business subject can be automatically and accurately identified by classifying and transforming the features to be processed. Specifically, the business type of the business subject is determined based on the transaction distribution pattern and the subject attribute data, including:

[0128] The transaction distribution law and the attribute features are combined to obtain the features to be processed, where the attribute features are features extracted from the subject attribute data;

[0129] Perform classification transformation on the features to be processed to obtain the business type probability;

[0130] Determine the business type of the operating entity based on the business type probability.

[0131] Among them, classification transformation refers to the process of converting features into classification prediction results. For example, a combination of one or more types of classification models such as fully connected layer networks, convolutional neural networks, recurrent neural networks, residual networks, etc. can be used to classify and transform the features to be processed to obtain the probabilities corresponding to different specified business types, that is, the business type probabilities.

[0132] The business type probability refers to the probability that the business type of the business entity is any specified business type. The specified business type refers to the business type set according to actual needs or application scenarios. For example, the specified business type may include but is not limited to online-single branch business type, online-multi-branch business type, offline-single branch business type, offline-multi-branch business type, and online and offline mixed business type.

[0133] For example, the transaction distribution law attribute features can be spliced ​​to obtain the feature to be processed {transaction distribution law, attribute features}, and then the feature to be processed is input into the fully connected layer network. The feature to be processed multiplies the input object features with the weight matrix, and is linearly transformed through the activation function to obtain the probability corresponding to different specified business types, that is, the business type probability, and the specified business type with the highest business type probability is used as the business type of the operating entity. It should be noted that if the transaction distribution law is not in the form of a feature vector, it needs to be encoded to obtain the transaction distribution feature in the form of a feature vector, and then spliced ​​with the attribute feature to obtain the feature to be processed.

[0134] In some embodiments, the features to be processed may be classified and transformed by a normalized exponential function such as a Softmax function to obtain the business type probability. The normalized exponential function may convert the input real value into an output representing the probability distribution, namely the business type probability, to provide an intuitive and standardized expression of the business type probability.

[0135] For example, the normalized exponential function can be used Perform an exponential operation on each element of the feature to be processed, and divide each result by the sum of the exponentials of all elements to obtain a probability distribution that includes the probability of each specified business type. In this function, K represents the number of specified business types predicted, y represents the specified business type, and y i Indicates the specified business type i (if the business type is i, then y i is equal to 1, otherwise equal to 0), p represents the probability of business type, p i Represents the probability of business type i.

[0136] Specifically, Figure 1c The processing flow of the type classification model shown in the figure can input the element z1 such as 3, the element z2 such as 1, and the element z3 such as -3 in the to-be-processed features of the business entity into the Softmax layer (i.e., the type classification model), and perform exponential processing on the elements z1 to z3 respectively (i.e., take the exponent of e for each element) through the normalized exponential function in the Softmax layer to obtain their exponential values ​​respectively. For example, As in 2.7, For example, 0.05, then sum it up to get the total value Then divide the indexed value of each element by the total value to obtain the corresponding normalized business type probability distribution y1 such as 0.88, y2 such as 0.12, y3 such as 0, where 1>y i >0,∑ i y i = 1. According to the business type probability output by the Softmax layer, the designated business type corresponding to 0.88 is taken as the business type of the business entity.

[0137] In actual applications, since different business entities may have similar business data such as name, address, location data, and behavior data, these similar data may lead to confusion of the business data and cause errors in the business type of the determined business entity. Moreover, different business entities and different purchasers may have potential relationships, such as belonging to the same business manager. These potential relationships may also affect the business type of the business entity. Therefore, in some embodiments, potential relationships can be mined through subject attribute data, behavior data, and location data to obtain more accurate related transaction behaviors between the business entity and the purchaser, so as to obtain a reference business type of the business entity based on the potential relationship. The business type of the business entity is then compared and analyzed with the reference business type to determine the true business type of the business entity. Specifically, the business type determination method also includes:

[0138] Obtaining the pending business relationship from the main business data of the business entity, the pending business relationship including at least one of the business relationship between the business entity and the purchaser, the business relationship between the business entity and the person in charge of the business, the business relationship between the business entity and the business equipment, the business relationship between the purchaser and the business equipment, the business relationship between the business entities, and the business relationship between the purchasers;

[0139] Determine the related-party transaction between the operating entity and the purchaser based on the business relationship to be processed;

[0140] Determine the reference business type of the operating entity by the proportion of designated related-party transactions in related-party transactions;

[0141] Determine the target business type of the operating entity based on the business type of the operating entity and the reference business type.

[0142] Among them, the business relationship refers to the relationship between entities related to business activities. The pending business relationship refers to the business relationship used to determine the related transaction behavior. For example, the business relationship between the business entity and the purchaser may include the transaction relationship between the business entity and the purchaser, such as it can be represented by transaction data; the business relationship between the business entity and the person in charge of the business may include the employment relationship or management relationship between the business entity and the person in charge of the business, such as the person in charge of the business may be the legal person of the business entity; the business relationship between the business entity and the business equipment may include the use relationship between the business entity and the business equipment, and the business equipment refers to various tools, machines, equipment and facilities used for production, operation or provision of services, such as production equipment, office equipment, communication equipment, etc.; the business relationship between the purchaser and the business equipment may include the transaction support relationship between the purchaser and the business equipment, that is, the purchaser purchases products or services from the business entity through the business equipment; the business relationship between the business entities may include the relationship between the transaction data of the business entities, such as the transaction locations of multiple business entities; the business relationship between the purchasers may include the relationship between the transaction data of the purchasers, such as the transaction locations of multiple purchasers.

[0143] In practical applications, the acquired pending business relationships can be stored in a variety of data forms. For example, the pending business relationships can be stored in relational data forms such as tables and graph data, so as to facilitate the mining of potential relationships between different business entities and different buyers. Taking tables as an example, one or more tables can be created using relational databases such as MySQL, Oracle, and SQLServer to store pending business relationships. Each table represents an entity and establishes associations through primary keys and foreign keys, so as to provide powerful query and association functions through structured storage.

[0144] Among them, the related transaction behavior refers to the transaction behavior between the business entity and the purchaser that is associated with the business relationship. The designated related transaction behavior refers to the related transaction behavior corresponding to the designated business type. Different designated related transaction behaviors can be set for different designated business types, and the setting of designated related transaction behaviors can be different according to different application scenarios. For example, the related transaction behavior corresponding to the online-single branch business type can be an offline transaction between the business entity and the purchaser, and the business entity and the purchaser are in the same branch business type transaction, the related transaction behavior corresponding to the online-multi-branch business type can be an online transaction between the business entity and the purchaser, and the business entity and the purchaser are in multiple branches of business type transactions, the related transaction behavior corresponding to the offline-single branch business type can be an offline transaction between the business entity and the purchaser, and the business entity and the purchaser are in the same branch of business type transactions, the related transaction behavior corresponding to the offline-multi-branch business type can be an offline transaction between the business entity and the purchaser, and the business entity and the purchaser are in multiple branches of business type transactions, and the related transaction behavior corresponding to the online and offline mixed business entity can be an offline transaction and an online transaction between the business entity and the purchaser.

[0145] The reference business type refers to a business type determined based on the proportion of the designated related-party transaction behavior in the related-party transaction behavior. For example, the reference business type may be the business type corresponding to the designated related-party transaction behavior with the highest proportion in the related-party transaction behavior, and the business type is one of the designated business types.

[0146] For example, the main business data of multiple business entities can be obtained, and the transaction relationship between the business entity and the purchaser, the employment relationship or management relationship between the business entity and the person in charge of the business, the use relationship between the business entity and the business equipment, the transaction support relationship between the purchaser and the business equipment, the transaction location of multiple business entities, the transaction location of multiple purchasers and other business relationships can be obtained from the main business data to determine the reference business type of the business entity based on these business relationships. The reference business type can be used to verify the accuracy of the determined business type of the business entity. If the reference business type of business entity A is consistent with the business type of business entity A, it is considered that the business type of business entity A is accurate, and the business type can be used as the final business type of business entity A, that is, the target business type. If they are inconsistent, it is considered that the business type of business entity A is inaccurate, and the business type of business entity A needs to be re-determined. For example, the main business data of business entity A can be cleaned and the cleaned main business data can be used to determine the business type of business entity A, or the model parameters of the feature cross model can be optimized to obtain a new transaction distribution law, so as to determine the business type of the business entity using the new transaction distribution law.

[0147] For example, in some implementations, these business relationships can be analyzed for association through association rule mining, cluster analysis, decision trees, etc., such as matching through common key fields such as business entity identity information, business manager identity information, business equipment number, transaction location, and buyer identity information, in order to determine the associated transaction behavior between the business entity and the buyer.

[0148] Specifically, the employment relationship or management relationship between the business entity and the person in charge of the business, the transaction support relationship between the purchaser and the business equipment, and the use relationship between the business entity and the business equipment can be used to determine whether the business entity includes multiple branch business types. If multiple business entities are managed by the same person in charge of the business or use the same business equipment, and the corresponding business entity identity information is similar, then the multiple business entities belong to the branch business type of the same business entity. The transaction relationship and transaction location between the business entity and the purchaser can be used to determine the transaction method of the business entity. If the transaction location of the business entity in any transaction relationship is located in a different place from the transaction location of the purchaser, then the transaction method of the transaction is an online transaction. If the business entity includes multiple branch business types or is a branch business type of any business entity, then the related transaction behavior 1 corresponding to the transaction is an online transaction between the business entity and the purchaser, and the transaction between the business entity and the purchaser is in multiple branch business types. If the business entity has no branch business type and is not a branch business type of any business entity, then the related transaction behavior 2 corresponding to the transaction is an online transaction between the business entity and the purchaser, and the transaction between the business entity and the purchaser is in the same branch business type. Similarly, if the transaction location in any transaction relationship of the business entity is in the same place as the transaction location of the purchaser, the transaction method of the transaction is an offline transaction. If the business entity includes multiple branch business types or is a branch business type of any business entity, then the related transaction behavior 3 corresponding to the transaction is an offline transaction between the business entity and the purchaser, and the business entity and the purchaser are in multiple branch business types. If the business entity has no branch business type and is not a branch business type of any business entity, then the related transaction behavior 4 corresponding to the transaction is an offline transaction between the business entity and the purchaser, and the business entity and the purchaser are in the same branch business type. The total number of all related transaction behaviors 1 to 4 can be counted, and the number of related transaction behaviors 1, 2, 3, and 4 can be counted respectively, and then the proportion of the number of each related transaction behavior to the total number is calculated. If related transaction behavior 1 accounts for the highest proportion, the business type corresponding to related transaction behavior 1, online-multi-branch business type, can be used as the reference business type.

[0149] In some implementations, the business relationships to be processed can be stored in the form of graph data, and the entity nodes in the graph data can be grouped into communities to distinguish business entities with different related transaction behaviors through communities, and potential related relationships can be automatically and deeply mined to efficiently and accurately determine related transaction behaviors. Specifically, based on the business relationships to be processed, the related transaction behaviors between the business entity and the purchaser are determined, including:

[0150] The entities in the business relationship to be processed are taken as nodes, and the entities include one or more of the business entity, the purchaser, the business person in charge, and the business equipment;

[0151] The nodes are grouped into communities according to the business relationships to be processed, and the target communities corresponding to the business entities are determined;

[0152] From the target community, identify the related transaction behaviors between the operating entity and the buyer.

[0153] A community refers to a set of closely connected and internally similar nodes in a community relationship network. A community can represent the relationship between related relationship entities, which are entities in a business relationship. In some implementations, the graph data to be processed for business relationship construction can be grouped into communities using community discovery algorithms such as Louvain (Lohan algorithm) and LPA (label propagation algorithm).

[0154] For example, the entities in the pending business relationship can be used as nodes, and the pending business relationship can be used as an edit to construct the pending business relationship in the form of graph data, so as to facilitate the query and classification of replicated relationships. Then, based on the node characteristics and edge characteristics, entities with similar characteristics or mutual relations are classified into the same community to group the nodes in the graph data into communities. For any community, the proportion of various related transaction behaviors in all its related transaction behaviors can be counted, and the designated business type corresponding to the related transaction behavior with the highest proportion can be used as the reference business type corresponding to the business entity corresponding to the business entity type node (i.e., the node corresponding to the business entity) in the community. For example, for business entity A, we can determine whether it is a related transaction behavior and make statistics based on the edge features in the community where the node corresponding to business entity A is located. For example, community A contains entity nodes 1 to 5, where nodes 1 and 2 are business entity type nodes. Based on the edge features, we can determine that there are 3 offline non-local transactions between nodes 1 and 2, 1 offline co-local transaction between nodes 1 and 3, 1 online non-local transaction between nodes 1 and 4, 5 offline non-local transactions between nodes 2 and 3, and 5 offline non-local transactions between nodes 2 and 4. Obviously, offline co-local transactions account for more than 90% of all transaction behaviors or offline non-local transactions account for more than 90%. Therefore, the business type of the business entity represented by nodes 1 and 2 can be determined as an offline-co-local transaction business type or an offline-multi-branch-non-local business type.

[0155] In some implementations, the final real business type can be obtained by weighted summing the probability value of the business type of the business entity and the probability value of the reference business type. Specifically, according to the business type of the business entity and the reference business type, determining the target business type of the business entity includes:

[0156] If the business type of the business entity does not match the reference business type, obtain a first probability value corresponding to the business type and a second probability value corresponding to the reference business type;

[0157] Performing a weighted summation on the first probability value and the second probability value to determine a target probability value;

[0158] Determine the target business type of the operating entity based on the target probability value.

[0159] The first probability value and the second probability value are probability values ​​corresponding to the business type and the reference business type, respectively. For example, the second probability value may be the proportion of the designated related transaction behavior corresponding to the reference business type in the related transaction relationship.

[0160] In some embodiments, the first probability value is the business type probability of the business type obtained by classification transformation. For example, the business type probability corresponding to business type A determined by the type classification model of business entity A is 0.6, and the probability of the reference business type determined by the business relationship is 0.9. The weight values ​​(a, b) are used to perform weighted summation to obtain a target probability value = 0.6 × a + 0.9 + b, and the designated business type corresponding to the target probability value is used as the target business type of business entity A. The correspondence between the target probability value and the designated business type can be set according to actual needs or application scenarios.

[0161] In some embodiments, the first probability value is the business type probability of all designated business types obtained by classification transformation. For example, the designated business types include business types 1 to 3, and the business type probabilities corresponding to business types 1 to 3 determined by the type classification model of the business entity A are 0.1, 0.4, and 0.5, respectively. The probability of the reference business type determined by the business relationship is 0.9, and the weight values ​​(A1, B1), (A2, B2), and (A3, B3) are used for weighted summation to obtain the target probability value 1 = 0.1 × A1 + 0.9 + B1, the target probability value 2 = 0.4 × A2 + 0.9 + B2, and the target probability value 3 = 0.5 × A3 + 0.9 + B3. The highest probability value among the target probability values ​​1 to 3, such as the designated business type 3 corresponding to the target probability value 3, can be used as the target business type of the business entity A.

[0162] In some embodiments, the weights for weighted summation of probability values ​​can be determined by a parameter search method to automatically find the optimal solution and improve the efficiency and accuracy of determining the business type of the business entity. For example, multiple groups of parameter combinations can be set for weight values, such as (a1, b1), (a2, b2), ..., (an, bn). For any business type, multiple groups of parameter combinations can be used to weighted sum the corresponding first probability value and the second probability value, respectively, to obtain the target probability value corresponding to each group of parameter combinations, and then by comparing the target probability values ​​calculated by different parameter combinations, the optimal parameter combination corresponding to the business type and the target probability value corresponding to the optimal parameter combination are determined according to preset evaluation indicators such as accuracy, recall rate, F1 score, etc., and the target probability value corresponding to the optimal parameter combination is used to determine the target business type of the business entity.

[0163] In some embodiments, the model parameters of at least one of the feature cross-model and the type classification model can be adjusted using reference business data to optimize the expressive power of the model, the adjusted feature cross-model can be used to perform data cross-processing, and / or the adjusted type classification model can be used to predict the business type of the business entity to obtain a more accurate business type of the business entity.

[0164] In some embodiments, the business type determined by the feature cross model and the type classification model can be compared with the reference business type during the model training process, and the model parameters can be adjusted according to the difference to optimize the performance of the model and obtain a more accurate business type of the business entity. Specifically, the business type determination method also includes:

[0165] Obtaining a data classification model to be trained, a type classification model to be trained, and a training sample set, wherein the training sample set includes training subject business data of at least one business subject sample, the training subject business data includes training subject attribute data and training transaction data, and the training transaction data includes training behavior data and training location data;

[0166] Using the data classification model to be trained, the training behavior data and the training position data are classified according to the association between the training behavior data and the training position data, so as to obtain the training target behavior data corresponding to the specified transaction mode and the training transaction position distribution;

[0167] The training transaction distribution law is obtained by combining the training target behavior data and the training transaction position distribution;

[0168] Determine the training business type of the business entity sample through the type classification model to be trained, according to the training transaction distribution law and the training entity attribute data;

[0169] Obtain reference business types for training of business entity samples;

[0170] According to the difference between the training reference business type and the training business type, the model to be trained is trained to obtain the trained model, the model to be trained includes at least one of the data classification model to be trained and the type classification model to be trained, the trained model includes at least one of the trained data classification model and the trained type classification model, the trained data classification model is used to obtain the target behavior data and the transaction location distribution, and the trained type classification model is used to determine the business type of the business entity.

[0171] Among them, the business entity sample refers to the business entity used for training. The training entity business data, training entity attribute data, training transaction data, training behavior data, training location data, training target behavior data, training transaction location distribution, training business type, and training reference business type refer to the entity business data, entity attribute data, transaction data, behavior data, location data, target behavior data, transaction location distribution, business type, and reference business type in the model training process, respectively.

[0172] For example, a training sample set of multiple business entity samples can be used to train a feature crossover model and a type classification model to determine the reference business type for training of each business entity sample. And according to the aforementioned method, the business relationship to be processed can be obtained from the business data of the training entity of the business entity sample, and the related transaction behavior between the business entity sample and the purchaser can be determined based on the business relationship to be processed. The reference business type for training of the business entity sample is determined by the proportion of the specified related transaction behavior in the related transaction behavior. And the training reference business type of each business entity sample is compared with its training business type. If they are different, the model to be trained is trained until the training reference business type of all business entity samples is the same as their training business type, and the training is terminated, and the obtained model is used as the trained model.

[0173] In some embodiments, the training sample set is obtained semi-automatically to improve the efficiency and accuracy of constructing the training sample set. For example, samples related to the target task can be screened by feature rules, and then manually confirmed and labeled with business types. Feature rules refer to rules that describe a certain pattern or association in the data, and these rules can be extracted based on domain knowledge, empirical rules, or statistical analysis.

[0174] In some implementations, during the model training process, sample data may be adjusted based on the difference between the training reference business type and the training business type to improve the performance of the model. Specifically, based on the difference between the training reference business type and the training business type, the model to be trained is trained to obtain a trained model, including:

[0175] According to the difference between the training reference business type and the training business type, the training sample set is adjusted to obtain an adjusted training sample set;

[0176] The model to be trained is trained using the adjusted training sample set to obtain a trained model.

[0177] For example, when the reference business type used for training of the business entity sample is different from its training business type, the training sample set can be adjusted by adding or deleting samples, modifying sample features or category labels, etc., and the adjusted training sample set can be used to train the model to be trained to provide more representative training data and improve the performance of the model.

[0178] The business type determination scheme provided in the embodiment of the present application can be applied in the scenarios of various business platforms. For example, taking the catering platform as an example, the main business data of the business entity settled in the catering platform is obtained, and the main business data includes the main attribute data and transaction data, and the transaction data includes behavior data and location data; according to the association relationship between the behavior data and the location data, the behavior data and the location data are classified to obtain the target behavior data corresponding to the specified transaction method and the transaction location distribution; the transaction distribution law is determined from the target behavior data and the transaction location distribution; according to the transaction distribution law and the main attribute data, the business type of the business entity is determined.

[0179] As can be seen from the above, based on the correlation between the behavior data and the location data in the transaction data of the business entity, the behavior data and the location data are classified to obtain the target behavior data and transaction location distribution that can more accurately reflect the transaction characteristics of the business entity based on the associated, multi-dimensional transaction data. At the same time, the target behavior data and the transaction location distribution are used to determine the transaction distribution law, and the transaction distribution law is used to reflect the regular transaction characteristics of the business entity. In this way, the embodiment of the present application combines the transaction distribution law and the subject attribute data, and can determine the business type of the business entity through these multi-dimensional data. The business type matches the transaction law and the subject attribute of the business entity, and can more accurately characterize the real business type of the business entity, so as to provide a basis for the management and supervision of the business entity, and facilitate the business platform to manage and supervise the business entity.

[0180] The method described in the above embodiment will be further described in detail below.

[0181] In this embodiment, the method of the embodiment of the present application will be described in detail by taking the platform marketing scenario applied to the Internet service platform as an example.

[0182] In existing Internet service platforms such as payment platforms, e-commerce platforms, and anchor platforms, merchants (i.e., business entities) can easily enter these platforms through online applications. However, the information provided by these merchants when entering the platform may be inaccurate or incomplete. This makes it difficult for the platform to verify the merchant's real business type, increasing the difficulty of the business platform to manage and supervise the merchant. For example, in the offline restaurant newly settled in the platform, most of the merchant's transaction objects are online users. These users may be user groups in the same place or user groups distributed in various places. The merchant may be an online merchant with a single store or a merchant with multiple stores. At this time, if the type of the merchant is determined only by the information filled in by the merchant when applying for entry, when the platform recommends services to the merchant and identifies risky transactions, it will lead to promotion errors or poor supervision of its risky transactions. For example, when the merchant fills in its business scope as local offline operation when applying for entry, but it actually has user groups distributed in various places, if it is only promoted locally based on the business scope filled in, it will not reach its real user group, resulting in poor promotion effect.

[0183] In this way, for the platform marketing scenario of the Internet service platform, the embodiment of the present application provides a business type determination method, which can identify the real business type of merchants on the platform to provide them with accurate marketing services.

[0184] like Figure 2a and Figure 2b As shown, the specific process of a method for determining a business type is as follows:

[0185] 210. In response to a type query request for a business entity, obtain a mapping relationship set and second business data, where the mapping relationship set includes a mapping relationship between the first business data and the business entity.

[0186] For example, when the operator of an Internet service platform conducts promotion activities for one or more merchants settled in the platform, after obtaining the consent or permission of the one or more merchants and the users (i.e., the purchasers) associated with these merchants, the operator can generate a type query request by clicking the query control in the platform operation backend system interface. Based on the type query request, the mapping relationship set containing the real-time update characteristics (i.e., the first business data) of the one or more merchants is obtained, and the offline characteristics (i.e., the second business data) are obtained from the offline library.

[0187] Specifically, Figure 2cThe process of obtaining the subject's business data is shown in the figure. Business data may include business data entered when the merchant settles in, transaction environment, transaction goods and other subject attribute data. These data may be in the form of text, images and other forms. For example, subject attribute data may include images such as storefront photos, business scene pictures, business licenses, and texts such as company names, product names, and transaction notes. Business data may also include transaction data such as transaction time, transaction location, and transaction channels. For example, transaction data may include transaction flow, credit risk, and behavior sequence data. Therefore, Figure 2b In the business type determination method shown, the mapping relationship set can store subject attribute data and transaction data. With the consent or permission of the one or more merchants and the users (i.e., buyers) associated with these merchants, these data can be collected through the multimodal data collection platform and sent to the server of the Internet service platform for statistical processing.

[0188] like Figure 2c As shown in the process of obtaining the main business data, the server of the Internet service platform can perform streaming statistical short-cycle calculations on the transaction data of each merchant through streaming real-time computing such as Flink processing framework or spark processing framework, and obtain the first business data after cleaning, filtering and other pre-processing of data in the form of images, texts, etc. in the business data, and store the first business data in the mapping relationship set to update the mapping relationship set in real time. The first business data is mainly calculated based on the short-cycle data of the first statistical period, and is more focused on the mining of fine-grained behavioral cross-data. Among them, the mapping relationship set is stored in the KV (Key-Value) storage system, which is a high-performance system that supports millisecond-level reading and writing, so that the reading and writing operations of the data can be completed in a very short time, which is suitable for scenarios that require real-time response and high concurrent access.

[0189] At the same time, the server of the Internet service platform will store the merchant's transaction data and subject attribute data for a long time, and make statistics according to the second statistical period such as 1 week / 1 month / 1 year, and then enter the statistical data into the offline database. In this way, the subject business data can be obtained from the first business data and the second business data.

[0190] 220. Store the mapping relationship between the second business data and the business entity in a mapping relationship set to obtain the main business data of the business entity.

[0191] For example, the server of the Internet service platform can store the second business data of any merchant A obtained from the offline library in the mapping relationship set of merchant A, so as to obtain a variety of business data with multi-angle data characteristics based on different statistical periods as the main business data of merchant A.

[0192] 230. According to the association between the behavior data and the location data, the behavior data and the location data are classified to obtain the target behavior data corresponding to the specified transaction method and the transaction location distribution.

[0193] For example, the MapReduce framework or HDFS system can be used to classify the transaction data of merchant A. For example, the transaction data in the mapping relationship set can be divided into multiple data blocks, one data block includes one or more groups of business data, and multiple processing nodes process multiple data blocks in parallel to determine the target behavior data corresponding to the specified transaction method in the group of data blocks, as well as the transaction location classification of the business entity or the transaction location classification of the buyer, and then merge all the data to obtain the target behavior data and transaction location distribution in all transaction data.

[0194] Specifically, Figure 2d The transaction data classification processing flow shown in the figure can include the following types of behavior data and location data: user transaction channels: active code scanning (i.e. code scanning transaction), payment code (i.e. transaction through payment code), mini program (i.e. transaction through mini program), photo album code scanning (i.e. transaction through photo album code scanning), URL connection (i.e. transaction through link), APP initiation (i.e. transaction through application); merchant behavior: merchant location distribution (i.e. transaction location of business entity), customer location distribution (i.e. transaction location of buyer), commodity content; buyer behavior: common location of behavior, common time of behavior, transaction frequency, amount, etc. The behavior data and location data of each transaction activity in the transaction data can be cross-processed through feature cross-processing, and the classification result of each transaction activity can be obtained, such as the target behavior data corresponding to the specified transaction method: offline transaction: active code scanning, payment code, same-city transaction, online transaction: photo album, mini program, and transaction location distribution such as transaction location: user location distribution, transaction point location distribution; store customer group distribution: merchant multi-location distribution, user location multi-location distribution, etc.

[0195] 240. Determine the transaction distribution pattern based on the target behavior data and transaction location distribution.

[0196] For example, Figure 2eThe combined processing flow shown can combine all target behavior data of merchant A and all transaction location distributions through the feature cross model to obtain the transaction distribution law of merchant A. For example, the merchant fine-grained cross-features (i.e., transaction distribution law) can be analyzed by combining multiple behavior data and location data. The transaction distribution law can include multiple types of distribution data such as basic transaction class (i.e., basic distribution data), online and offline channel distribution class (transaction channel distribution), merchant store distribution, transaction time series (i.e., transaction sequence characteristics). The transaction distribution law is a fine-grained feature data that can improve the model prediction effect and can have a stronger explanatory power for the model. Through fine-grained mining of merchant transaction data, mainly from transaction behavior, user habits, merchant habits and other features, through the patterns initiated by the nature of behavior and historical habits, the merchant's online, offline, multi-store and other highly credible and highly explanatory features are mined to more quickly identify the merchant's business type.

[0197] 250. Determine the business type of the operating entity based on transaction distribution patterns and entity attribute data.

[0198] For example, Figure 2b The business type determination method shown can convert different data in the main attribute data of merchant A into attribute features in vector form according to the data form, and obtain the transaction distribution law obtained by classifying and combining the transaction data. Then, the transaction distribution law and attribute features of merchant A are spliced ​​to obtain the features to be processed, and the type classification model is input to learn the business type to predict and classify the business type of merchant A. Obviously, the features to be processed are multimodal features composed of data in various forms, which can provide information of different dimensions and angles to more comprehensively represent the characteristics of the business entity.

[0199] Specifically, Figure 2f The processing flow of the type classification model shown in the figure is that the type classification model can predict the business type probability of merchant A corresponding to multiple specified business types based on the transaction distribution law and the characteristics to be processed obtained by combining the subject attribute data. The multiple specified business types include online-single store, online-multiple stores, offline-single store, offline-multiple stores, and online and offline mixed operations. Among them, the business type probability corresponding to online-multiple stores is the highest, so it can be determined that the business type of merchant A is online-multiple stores. The server of the Internet service platform can return "online-multiple stores" as the query result of merchant A to the backend system of the Internet service platform, and display it in the backend system interface of the Internet service platform, so that the operation personnel can formulate accurate promotion plans based on the query results.

[0200] In some implementations, in order to improve the accuracy of querying the business type of a merchant, the determined business type of the merchant may not be returned to the backend system of the online service platform, but the business type may be further verified based on the reference business type to determine the true business type of the business entity. Figure 2g As shown in the figure, the specific process for verifying the business type is as follows:

[0201] 261. If the business type of the business entity does not match the reference business type, obtain a first probability value corresponding to the business type and a second probability value corresponding to the reference business type.

[0202] 262. Perform a weighted summation on the first probability value and the second probability value to determine a target probability value.

[0203] 263. Determine the target business type of the operating entity based on the target probability value.

[0204] For example, since merchants naturally have multiple relationship data, and the merchant transaction behaviors and store names of offline transactions will have a certain degree of similarity, and merchants with multiple stores also have natural similarities in their names. For example: XX City XX Xue XX1 Road Store, XX City XX Xue XX2 Road Store, merchants with multiple stores will also be created by the same legal person. Therefore, Louvain (Lohan Algorithm), LPA (Label Propagation Algorithm) and other methods can be used to conduct community mining (i.e. community grouping) based on the merchant relationship graph (i.e., pending business relationships in the form of graph data), and classify merchants with similar business models into the same community. The graph data includes transaction behaviors such as the relationship between the two parties of the purchase, such as merchant 1-user 1, merchant 1-user 2, merchant settlement relationships, such as merchant 1-legal person 1, merchant 2-legal person 1, company name 1-merchant 1, company name 1-merchant 2, user and environment relationships, such as user 1-device 1, user 2-device 1, merchant transaction location, user transaction location and other relationship data (i.e., pending business relationships). In actual applications, a mapping relationship may be established between the identification information of the merchant and the relationship data of the merchant, such as the data in the diagram, and stored in a mapping relationship set for easy calling at any time.

[0205] The related merchants discovered in the community can be analyzed based on the characteristics of the merchants in the community. For example, if the proportion of offline transactions in the same location in the community is > 90% or the proportion of offline transactions in multiple locations is > 90%, they can be classified into offline-same-city business types and offline-different-location-multiple-store business types respectively, and used as reference business types for merchants in the community such as Merchant A.

[0206] like Figure 2bThe business type determination method shown is then combined and identified (i.e. verified) through the reference business type of merchant A determined by community mining and the business type predicted by the type classification model. The same auditable and confirmed classification results are accurate; for those that are different, the samples can be manually iterated and adjusted during the model training phase to improve the effect of the type classification model. The reference business type of merchant A determined by community mining and the business type predicted by the type classification model can also be combined to obtain the final merchant business type (i.e., target business type), and the target business type is returned to the Internet service platform backend system as the query result of merchant A, and displayed in the Internet service platform backend system interface, so that operators can formulate accurate promotion plans based on the query results.

[0207] As can be seen from the above, the embodiment of the present application obtains a variety of business data of the business entity, and processes the business data in two ways: offline and on-site, so as to obtain a variety of business data with multi-angle data features based on different statistical periods. At the same time, by cross-processing and combining the transaction data, more fine-grained and highly interpretable transaction features can be mined, and the actual business type of the business entity can be determined by integrating the transaction characteristics and attribute characteristics of the business entity and predicting and processing them by the type classification model. In addition, the accuracy of the business type determined by the reference business type determined by community mining is verified to determine the actual business type of the business entity.

[0208] In order to better implement the above method, the embodiment of the present application also provides a business type determination device, which can be integrated in an electronic device, and the electronic device can be a terminal, a server, etc. Among them, the terminal can be a mobile phone, a tablet computer, a smart Bluetooth device, a laptop, a personal computer, etc.; the server can be a single server or a server cluster composed of multiple servers.

[0209] For example, in this embodiment, the method of the embodiment of the present application will be described in detail by taking the business type determination device being specifically integrated in the server as an example.

[0210] For example, Figure 3 As shown, the business type determination device may include an acquisition unit 310, a classification unit 320, a data determination unit 330, and a type determination unit 340, as follows:

[0211] (I) Acquisition Unit 310

[0212] Used to obtain the main business data of the operating entity. The main business data includes the main attribute data and transaction data. The transaction data includes behavior data and location data.

[0213] In some implementations, the acquisition unit 310 may be specifically configured to:

[0214] Acquire first business data and second business data of the business entity, where the first business data is obtained by counting current business data in a first statistical period, and the second business data is obtained by counting historical business data in a second statistical period, and the first statistical period is shorter than the second statistical period;

[0215] The main business data of the operating entity is obtained from the first business data and the second business data.

[0216] In some implementations, obtaining first business data and second business data of an operating entity includes:

[0217] In response to a type query request for a business entity, a mapping relationship set and second business data are acquired, where the mapping relationship set includes a mapping relationship between the first business data and the business entity;

[0218] The main business data of the operating entity is obtained from the first business data and the second business data, including:

[0219] The mapping relationship between the second business data and the business entity is stored in a mapping relationship set to obtain the main business data of the business entity.

[0220] (II) Classification Unit 320

[0221] It is used to classify the behavior data and the location data according to the association relationship between the behavior data and the location data, and obtain the target behavior data corresponding to the specified transaction method and the transaction location distribution.

[0222] In some implementations, the classification unit 320 may be specifically configured to:

[0223] Cross-processing the behavior data and the location data corresponding to the behavior data to obtain transaction cross-features;

[0224] According to the transaction cross-features, the target behavior data corresponding to the specified transaction method and the transaction location distribution are determined.

[0225] In some implementations, cross-processing is performed on the behavior data and the location data corresponding to the behavior data to obtain transaction cross-features, including:

[0226] Encoding the behavior data and the position data corresponding to the behavior data to obtain multiple encoding features;

[0227] Perform linear transformation on the encoded features to obtain linear features;

[0228] Perform nonlinear transformation on the encoded features to obtain nonlinear features;

[0229] Cross-learn any two coding features among the multiple coding features to obtain cross-features;

[0230] Linear features, nonlinear features and cross features are fused to obtain transaction cross features.

[0231] (III) Data determination unit 330

[0232] Used to determine transaction distribution patterns based on target behavior data and transaction location distribution.

[0233] (IV) Type determination unit 340

[0234] Used to determine the business type of the operating entity based on transaction distribution patterns and entity attribute data.

[0235] In some implementations, the type determination unit 340 may be specifically configured to:

[0236] The transaction distribution law and the attribute features are combined to obtain the features to be processed, where the attribute features are features extracted from the subject attribute data;

[0237] Perform classification transformation on the features to be processed to obtain the business type probability;

[0238] Determine the business type of the operating entity based on the business type probability.

[0239] In some implementations, the business type determination device may further include a verification unit, which may be specifically used to:

[0240] Obtaining the pending business relationship from the main business data of the business entity, the pending business relationship including at least one of the business relationship between the business entity and the purchaser, the business relationship between the business entity and the person in charge of the business, the business relationship between the business entity and the business equipment, the business relationship between the purchaser and the business equipment, the business relationship between the business entities, and the business relationship between the purchasers;

[0241] Determine the related-party transaction between the operating entity and the purchaser based on the business relationship to be processed;

[0242] Determine the reference business type of the operating entity by the proportion of designated related-party transactions in related-party transactions;

[0243] Determine the target business type of the operating entity based on the business type of the operating entity and the reference business type.

[0244] In some implementations, determining the associated transaction behavior between the operating entity and the purchaser based on the business relationship to be processed includes:

[0245] The entities in the business relationship to be processed are taken as nodes, and the entities include one or more of the business entity, the purchaser, the business person in charge, and the business equipment;

[0246] The nodes are grouped into communities according to the business relationships to be processed, and a grouped community relationship network is obtained;

[0247] From the community relationship network, determine the related transaction behavior between the operating entity and the buyer.

[0248] In some implementations, determining the target business type of the business entity according to the business type of the business entity and the reference business type includes:

[0249] If the business type of the business entity does not match the reference business type, obtain a first probability value corresponding to the business type and a second probability value corresponding to the reference business type;

[0250] Performing a weighted summation on the first probability value and the second probability value to determine a target probability value;

[0251] Determine the target business type of the operating entity based on the target probability value.

[0252] In some implementations, the business type determination device may further include a training unit, which may be specifically used to:

[0253] Obtaining a data classification model to be trained, a type classification model to be trained, and a training sample set, wherein the training sample set includes training subject business data of at least one business subject sample, the training subject business data includes training subject attribute data and training transaction data, and the training transaction data includes training behavior data and training location data;

[0254] Using the data classification model to be trained, the training behavior data and the training position data are classified according to the association between the training behavior data and the training position data, so as to obtain the training target behavior data corresponding to the specified transaction mode and the training transaction position distribution;

[0255] The training transaction distribution law is obtained by combining the training target behavior data and the training transaction position distribution;

[0256] Determine the training business type of the business entity sample through the type classification model to be trained, according to the training transaction distribution law and the training entity attribute data;

[0257] Obtain reference business types for training of business entity samples;

[0258] According to the difference between the training reference business type and the training business type, the model to be trained is trained to obtain the trained model, the model to be trained includes at least one of the data classification model to be trained and the type classification model to be trained, the trained model includes at least one of the trained data classification model and the trained type classification model, the trained data classification model is used to obtain the target behavior data and the transaction location distribution, and the trained type classification model is used to determine the business type of the business entity.

[0259] In some embodiments, training the model to be trained according to the difference between the training reference business type and the training business type to obtain the trained model includes:

[0260] According to the difference between the training reference business type and the training business type, the training sample set is adjusted to obtain an adjusted training sample set;

[0261] The model to be trained is trained using the adjusted training sample set to obtain a trained model.

[0262] In specific implementation, the above units can be implemented as independent entities, or can be arbitrarily combined to be implemented as the same or several entities. The specific implementation of the above units can refer to the previous method embodiments, which will not be repeated here.

[0263] As can be seen from the above, the business type determination device of this embodiment includes an acquisition unit, a classification unit, a data determination unit and a type determination unit. Among them, the acquisition unit is used to acquire the main business data of the business entity, the main business data includes the main attribute data and transaction data, and the transaction data includes behavior data and location data; the classification unit is used to classify the behavior data and location data according to the association relationship between the behavior data and the location data, and obtain the target behavior data corresponding to the specified transaction method and the transaction location distribution; the data determination unit is used to determine the transaction distribution law from the target behavior data and the transaction location distribution; the type determination unit is used to determine the business type of the business entity according to the transaction distribution law and the main attribute data.

[0264] Therefore, the embodiment of the present application classifies the behavior data and location data based on the correlation between the behavior data and the location data in the transaction data of the business entity, so as to obtain the target behavior data and transaction location distribution that can more accurately reflect the transaction characteristics of the business entity based on the associated, multi-dimensional transaction data. At the same time, the target behavior data and the transaction location distribution are used to determine the transaction distribution law, and the transaction distribution law is used to reflect the regular transaction characteristics of the business entity. In this way, the embodiment of the present application combines the transaction distribution law and the subject attribute data, and can determine the business type of the business entity through these multi-dimensional data. The business type matches the transaction law and the subject attribute of the business entity, and can more accurately characterize the real business type of the business entity, so as to provide a basis for the management and supervision of the business entity, and facilitate the business platform to manage and supervise the business entity.

[0265] The embodiment of the present application also provides an electronic device, which can be a terminal, a server, etc. The terminal can be a mobile phone, a tablet computer, a smart Bluetooth device, a laptop, a personal computer, etc. The server can be a single server or a server cluster composed of multiple servers, etc.

[0266] In some embodiments, the business type determination device may also be integrated into multiple electronic devices. For example, the business type determination device may be integrated into multiple servers, and the business type determination method of the present application may be implemented by multiple servers.

[0267] In this embodiment, the electronic device of this embodiment is a server as an example for detailed description, for example, Figure 4 As shown, it shows a schematic diagram of the structure of the server involved in the embodiment of the present application, specifically:

[0268] The server may include one or more processing core processors 410, one or more computer-readable storage media memories 420, a power supply 430, an input module 440, and a communication module 450. Those skilled in the art will appreciate that Figure 4 The server structure shown in the figure does not constitute a limitation on the server, and may include more or fewer components than shown in the figure, or combine certain components, or arrange the components differently. Among them:

[0269] The processor 410 is the control center of the server, and uses various interfaces and lines to connect various parts of the entire server. It executes various functions of the server and processes data by running or executing software programs and / or modules stored in the memory 420, and calling data stored in the memory 420. In some embodiments, the processor 410 may include one or more processing cores; in some embodiments, the processor 410 may integrate an application processor and a modem processor, wherein the application processor mainly processes the operating system, user interface, and application programs, and the modem processor mainly processes wireless communications. It is understandable that the above-mentioned modem processor may not be integrated into the processor 410.

[0270] The memory 420 can be used to store software programs and modules. The processor 410 executes various functional applications and data processing by running the software programs and modules stored in the memory 420. The memory 420 may mainly include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area may store data created according to the use of the server, etc. In addition, the memory 420 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other volatile solid-state storage devices. Accordingly, the memory 420 may also include a memory controller to provide the processor 410 with access to the memory 420.

[0271] The server also includes a power supply 430 for supplying power to various components. In some embodiments, the power supply 430 can be logically connected to the processor 410 through a power management system, so that the power management system can manage charging, discharging, and power consumption. The power supply 430 can also include any components such as one or more DC or AC power supplies, recharging systems, power failure detection circuits, power converters or inverters, and power status indicators.

[0272] The server may further include an input module 440, which may be used to receive input digital or character information and generate keyboard, mouse, joystick, optical or trackball signal inputs related to user settings and function controls.

[0273] The server may further include a communication module 450. In some embodiments, the communication module 450 may include a wireless module. The server may perform short-range wireless transmission through the wireless module of the communication module 450, thereby providing wireless broadband Internet access to the user. For example, the communication module 450 may be used to help the user send and receive emails, browse web pages, and access streaming media.

[0274] Although not shown, the server may further include a display unit, etc., which will not be described in detail herein. Specifically in this embodiment, the processor 410 in the server will load the executable files corresponding to the processes of one or more application programs into the memory 420 according to the following instructions, and the processor 410 will run the application programs stored in the memory 420, thereby realizing various functions, as follows:

[0275] Obtain the main business data of the operating entity, the main business data includes main attribute data and transaction data, and the transaction data includes behavior data and location data; classify the behavior data and location data according to the correlation between the behavior data and the location data, and obtain the target behavior data corresponding to the specified transaction method and the transaction location distribution; determine the transaction distribution pattern from the target behavior data and the transaction location distribution; determine the business type of the operating entity according to the transaction distribution pattern and the main attribute data.

[0276] The specific implementation of the above operations can be found in the previous embodiments, which will not be described in detail here.

[0277] As can be seen from the above, the embodiment of the present application classifies the behavior data and location data based on the correlation between the behavior data and the location data in the transaction data of the business entity, so as to obtain the target behavior data and transaction location distribution that can more accurately reflect the transaction characteristics of the business entity based on the associated, multi-dimensional transaction data. At the same time, the target behavior data and the transaction location distribution are used to determine the transaction distribution law, and the transaction distribution law is used to reflect the regular transaction characteristics of the business entity. In this way, the embodiment of the present application combines the transaction distribution law and the subject attribute data, and can determine the business type of the business entity through these multi-dimensional data. The business type matches the transaction law and the subject attribute of the business entity, and can more accurately characterize the real business type of the business entity, so as to provide a basis for the management and supervision of the business entity, and facilitate the business platform to manage and supervise the business entity.

[0278] A person of ordinary skill in the art will appreciate that all or part of the steps in the various methods of the above embodiments may be completed by instructions, or by controlling related hardware through instructions. The instructions may be stored in a computer-readable storage medium and loaded and executed by a processor.

[0279] To this end, an embodiment of the present application provides a computer-readable storage medium, in which a plurality of instructions are stored, and the instructions can be loaded by a processor to execute the steps in any of the business type determination methods provided in the embodiments of the present application. For example, the instructions can execute the following steps:

[0280] Obtain the main business data of the operating entity, the main business data includes main attribute data and transaction data, and the transaction data includes behavior data and location data; classify the behavior data and location data according to the correlation between the behavior data and the location data, and obtain the target behavior data corresponding to the specified transaction method and the transaction location distribution; determine the transaction distribution pattern from the target behavior data and the transaction location distribution; determine the business type of the operating entity according to the transaction distribution pattern and the main attribute data.

[0281] The storage medium may include: a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0282] According to one aspect of the present application, a computer program product or computer program is provided, the computer program product or computer program including a computer program / instruction, the computer program / instruction being stored in a computer-readable storage medium. A processor of an electronic device reads the computer program / instruction from the computer-readable storage medium, and the processor executes the computer program / instruction, so that the electronic device executes the method provided in various optional implementations provided in the above embodiments.

[0283] Since the instructions stored in the storage medium can execute the steps in any one of the business type determination methods provided in the embodiments of the present application, the beneficial effects that can be achieved by any one of the business type determination methods provided in the embodiments of the present application can be achieved. Please refer to the previous embodiments for details and will not be repeated here.

[0284] The above is a detailed introduction to a business type determination provided in an embodiment of the present application. Specific examples are used in this article to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only used to help understand the method of the present application and its core idea; at the same time, for technical personnel in this field, according to the idea of ​​the present application, there will be changes in the specific implementation method and application scope. In summary, the content of this specification should not be understood as a limitation on the present application.

Claims

1. A method for determining a business type, characterized in that: include: Acquire the subject business data of the business subject, wherein the subject business data includes subject attribute data and transaction data, and the transaction data includes behavior data and location data; Classifying the behavior data and the location data according to the association relationship between the behavior data and the location data to obtain target behavior data corresponding to a specified transaction method and a transaction location distribution; Determining a transaction distribution pattern based on the target behavior data and the transaction location distribution; The business type of the operating entity is determined based on the transaction distribution pattern and the entity attribute data.

2. The method for determining the business type according to claim 1, wherein: The obtaining of the main business data of the operating entity includes: Acquire first business data and second business data of the business entity, where the first business data is obtained by counting current business data in a first statistical period, and the second business data is obtained by counting historical business data in a second statistical period, and the first statistical period is shorter than the second statistical period; The main business data of the operating entity is obtained from the first business data and the second business data.

3. The method for determining the business type according to claim 2, wherein: The obtaining of the first business data and the second business data of the business entity includes: In response to a type query request for the business entity, acquiring a mapping relationship set and the second business data, the mapping relationship set including a mapping relationship between the first business data and the business entity; The obtaining of the main business data of the operating entity from the first business data and the second business data includes: The mapping relationship between the second business data and the business entity is stored in the mapping relationship set to obtain the main business data of the business entity.

4. The method for determining the business type according to claim 1, wherein: The classifying the behavior data and the location data according to the association relationship between the behavior data and the location data to obtain the target behavior data corresponding to the specified transaction method and the transaction location distribution includes: Cross-processing the behavior data and the location data corresponding to the behavior data to obtain transaction cross-features; According to the transaction cross-features, target behavior data corresponding to the specified transaction method and transaction location distribution are determined.

5. The method for determining the business type according to claim 4, characterized in that: The cross-processing of the behavior data and the location data corresponding to the behavior data to obtain transaction cross-features includes: Encoding the behavior data and the position data corresponding to the behavior data respectively to obtain a plurality of encoding features; Performing a linear transformation on the coding feature to obtain a linear feature; Performing a nonlinear transformation on the coding feature to obtain a nonlinear feature; Cross-learning any two of the plurality of coding features to obtain a cross-feature; The linear feature, the nonlinear feature and the cross feature are fused to obtain the transaction cross feature.

6. The method for determining the business type according to claim 1, wherein: Determining the business type of the operating entity according to the transaction distribution pattern and the entity attribute data includes: Combining the transaction distribution law and the attribute features to obtain features to be processed, wherein the attribute features are features extracted from the subject attribute data; Classify and transform the features to be processed to obtain business type probability; The business type of the business entity is determined according to the business type probability.

7. The method for determining the business type according to claim 1, wherein: The method further comprises: Acquire a pending business relationship from the business data of the business entity, wherein the pending business relationship includes at least one of a business relationship between the business entity and a purchaser, a business relationship between the business entity and a person in charge of business, a business relationship between the business entity and business equipment, a business relationship between the purchaser and business equipment, a business relationship between the business entities, and a business relationship between the purchasers; Determine the related transaction between the business entity and the purchaser according to the business relationship to be processed; Determine the reference business type of the operating entity by the proportion of designated related-party transactions in the related-party transactions; The target business type of the business entity is determined according to the business type of the business entity and the reference business type.

8. The method for determining the business type according to claim 7, wherein: Determining the associated transaction between the business entity and the purchaser based on the business relationship to be processed includes: Taking entities in the business relationship to be processed as nodes, the entities include one or more of the business entity, the purchaser, the business person in charge, and the business equipment; According to the business relationship to be processed, the nodes are grouped into communities to obtain a grouped community relationship network; From the community relationship network, determine the related transaction behavior between the business entity and the purchaser.

9. The method for determining the business type according to claim 7, wherein: The step of determining the target business type of the operating entity according to the business type of the operating entity and the reference business type includes: If the business type of the business entity does not match the reference business type, obtaining a first probability value corresponding to the business type and a second probability value corresponding to the reference business type; Performing a weighted summation on the first probability value and the second probability value to determine a target probability value; The target business type of the business entity is determined according to the target probability value.

10. The method for determining the business type according to claim 7, wherein: The method further comprises: Acquire a data classification model to be trained, a type classification model to be trained, and a training sample set, wherein the training sample set includes training subject business data of at least one business subject sample, wherein the training subject business data includes training subject attribute data and training transaction data, wherein the training transaction data includes training behavior data and training location data; Using the data classification model to be trained, the training behavior data and the training position data are classified according to the association relationship between the training behavior data and the training position data to obtain the training target behavior data corresponding to the specified transaction method and the training transaction position distribution; Combining the training target behavior data and the training transaction location distribution to obtain a training transaction distribution law; Determine the training business type of the business entity sample by using the type classification model to be trained, according to the training transaction distribution law and the training entity attribute data; Obtaining reference business types for training of the business entity samples; According to the difference between the training reference business type and the training business type, the model to be trained is trained to obtain a trained model, the model to be trained includes at least one of the data classification model to be trained and the type classification model to be trained, the trained model includes at least one of the trained data classification model and the trained type classification model, the trained data classification model is used to obtain the target behavior data and the transaction location distribution, and the trained type classification model is used to determine the business type of the business entity.

11. The method for determining the business type according to claim 10, wherein: The step of training the model to be trained according to the difference between the training reference business type and the training business type to obtain the trained model comprises: According to the difference between the training reference business type and the training business type, adjusting the training sample set to obtain an adjusted training sample set; The model to be trained is trained using the adjusted training sample set to obtain the trained model.

12. A device for determining a business type, characterized in that: include: An acquisition unit, used to acquire the subject business data of the business subject, wherein the subject business data includes subject attribute data and transaction data, and the transaction data includes behavior data and location data; A classification unit, configured to classify the behavior data and the location data according to the association relationship between the behavior data and the location data, and obtain target behavior data corresponding to a specified transaction method and a transaction location distribution; A data determination unit, configured to determine a transaction distribution rule based on the target behavior data and the transaction location distribution; A type determination unit is used to determine the business type of the business entity according to the transaction distribution law and the entity attribute data.

13. An electronic device, characterized in that: It comprises a processor and a memory, wherein the memory stores a plurality of instructions; the processor loads instructions from the memory to execute the steps in the business type determination method according to any one of claims 1 to 11.

14. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a plurality of instructions, and the instructions are suitable for being loaded by a processor to execute the steps in the business type determination method according to any one of claims 1 to 11.

15. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instructions are executed by a processor, the steps in the business type determination method according to any one of claims 1 to 11 are implemented.

Images