A method and device for determining the brand represented by an agent

Abstract

The present invention provides a method and apparatus for determining brands represented by an agent. The method includes preprocessing the agent's corporate name and the brand to be determined by the agent to obtain multiple agent brand groups, vectorizing the agent brand groups to obtain feature vectors for the agent brand groups, and inputting the feature vectors into an enterprise-brand relationship prediction model to determine the represented brand. Because the brand represented by the agent is predicted using an enterprise-brand relationship prediction model obtained by training with training samples labeled with the corresponding relationship between the agent's corporate name and brand, the efficiency of identifying the brand represented by the agent can be improved. Because the brand represented by the agent is predicted by the enterprise-brand relationship prediction model, the accuracy of determining the brand represented by the agent can be improved compared to the prior art method of manually searching for the agent's company using a small amount of upstream and downstream data.

Description

Technical Field

[0001] The present invention relates to the field of financial technology (Fintech), and in particular to a method and device for determining a brand represented by an agent. Background Art

[0002] With the development of computer technology, more and more technologies are being applied in the financial sector. Traditional finance is gradually shifting towards FinTech. However, the security and real-time requirements of the financial industry also place higher demands on technology. In addition to general corporate loan products, supply chain-focused financial products are also developing in the financial sector. Some supply chain finance products have strict brand restrictions, requiring distributors to represent specific brands before applying for these products. This requires the ability to pre-assess the brands a distributor may represent. Currently, the primary method for identifying potential distributor brands is through manual internet searches. However, due to the limited availability of relevant data on distributors' upstream and downstream businesses, manual searches are difficult to accurately identify distributors and are also unable to predict the brands a distributor may represent. This results in low identification efficiency and accuracy when manually searching for distributor brands. Summary of the Invention

[0003] The present invention provides a method and device for determining the brand represented by an agent principal, which can quickly determine the brand represented by the agent principal and improve the recognition efficiency and accuracy of the brand represented by the agent principal.

[0004] In a first aspect, the present invention provides a method for determining a brand represented by an agent, comprising:

[0005] Obtain the company name of the agent and the brand of the agent to be determined;

[0006] Preprocessing the corporate name of the agent and the brand of the agent to be determined to obtain a plurality of agent brand groups; each agent brand group includes the corporate name of the agent and its corresponding brand to be determined;

[0007] For any agent principal brand group among the plurality of agent principal brand groups, vectorizing the enterprise name and the brand of the agent to be determined in the agent principal brand group to obtain a feature vector of the agent principal brand group;

[0008] The feature vectors of the plurality of agent subject brand groups are input into an enterprise brand relationship prediction model to determine the brand represented by the agent subject, wherein the enterprise brand relationship prediction model is obtained by training and learning training samples marked with the correspondence between the agent subject's enterprise name and the brand.

[0009] In the above technical solution, the brand represented by the agent is predicted by using a corporate brand relationship prediction model obtained through training and learning using training samples marked with the corresponding relationship between the agent's corporate name and brand. This allows for rapid identification of the brand represented by the agent, improving the efficiency of identifying the agent's brand. Compared to the prior art method of manually searching for the agent and the agent's brand using a small amount of upstream and downstream data, this method can improve the accuracy of determining the brand represented by the agent. At the same time, the embodiment of the present invention uses a corporate brand relationship prediction model to replace manual searching for the brand represented by the agent, which can reduce labor costs and thus reduce the cost of expanding the agent's brand.

[0010] Optionally, vectorizing the enterprise names in the agent entity brand group and the brands of the agents to be determined to obtain the feature vectors of the agent entity brand group includes:

[0011] Using a pre-trained word vector model, vectorize the company names in the agent entity brand group to obtain vectors of the company names;

[0012] Performing vector processing on the brand of the agent to be determined in the agent main brand group using a one-hot model to obtain a vector of the brand of the agent to be determined;

[0013] The vector of the enterprise name and the vector of the brand of the to-be-determined agent are determined as the feature vector of the agent-entity brand group.

[0014] In the above technical solution, by vectorizing the enterprise name and the brand of the agent to be determined, the prediction efficiency of the enterprise-brand relationship prediction model can be improved.

[0015] Optionally, the pre-processing of the corporate name of the agent and the brand of the agent to be determined to obtain a plurality of agent brand groups includes:

[0016] Combining the corporate name of the agent and the brand of the agent to be determined to obtain the plurality of agent brand groups; and for any of the plurality of agent brand groups, performing word segmentation processing on the corporate name in the agent brand group and performing category coding processing on the brand of the agent to be determined in the agent brand group; or

[0017] The corporate name of the agency entity is segmented, and the brand of the agency to be determined is categorized and coded; and the corporate name after segmentation and the brand of the agency to be determined after categorization are combined to obtain the multiple agency entity brand groups.

[0018] Optionally, the word segmentation processing of the enterprise names in the agent entity brand group includes:

[0019] Segment the company names in the agent entity brand group to obtain the company name, region, business scope and business scale of the agent entity;

[0020] The company names in the agent entity brand group are vectorized using a pre-trained word vector model to obtain vectors of the company names, including:

[0021] The corporate name, region, business scope, and business scale of the agent are vectorized using a pre-trained word vector model to obtain a vector of the corporate name.

[0022] Optionally, the step of training samples marked with the correspondence between the company name and the brand of the agent to obtain the company-brand relationship prediction model includes:

[0023] Obtaining training samples of the corresponding relationship between the company name and brand marked with the agent entity;

[0024] Performing word segmentation processing on the corporate names of the agent entities in the training samples in the corresponding relationship with the brands, and performing category coding processing on the brands represented by the agent entities;

[0025] Vectorizing the company name after word segmentation processing and the brand after category coding processing in the correspondence between the company name of the agent in the training sample and the brand, to obtain a feature vector of the correspondence between the company name of the agent in the training sample and the brand;

[0026] The feature vectors of the correspondence between the corporate name and brand of the agent in the training sample are sequentially input into a preset machine learning model for training until the model converges, thereby obtaining the corporate-brand relationship prediction model.

[0027] Optionally, inputting the feature vectors of the plurality of agent-entity brand groups into an enterprise-brand relationship prediction model to determine the brand represented by the agent entity includes:

[0028] Inputting the feature vectors of the plurality of agent-entity brand groups into the enterprise-brand relationship prediction model in sequence to determine the prediction probability corresponding to each agent-entity brand group;

[0029] The brand of the to-be-determined agent with a predicted probability greater than a threshold is determined as the brand of the agent subject agent.

[0030] In the above technical solution, the brand represented by the agent entity is determined by judging that the predicted probability is greater than a threshold, which can improve the accuracy of determining the brand represented by the agent entity.

[0031] Optionally, after determining the brand represented by the agent, the method further includes:

[0032] The enterprise name of the agent entity and the brand of its corresponding agent are stored in the training sample, and the enterprise-brand relationship prediction model is retrained.

[0033] In the above technical solution, by storing the corporate name of the agent entity and its corresponding agent brand in the training sample, the data of the training sample can be improved, and the accuracy of the corporate-brand relationship prediction model can be further improved.

[0034] In a second aspect, an embodiment of the present invention provides a device for determining a brand represented by an agent, including:

[0035] An acquisition unit, configured to acquire the corporate name of an agent and the brand of the agent to be determined;

[0036] A processing unit is used to pre-process the corporate name of the agent entity and the brand of the agent to be determined by the agent entity to obtain multiple agent entity brand groups; each agent entity brand group includes the corporate name of the agent entity and its corresponding brand of the agent to be determined; for any agent entity brand group among the multiple agent entity brand groups, the corporate name and the brand of the agent to be determined in the agent entity brand group are vectorized to obtain a feature vector of the agent entity brand group; the feature vectors of the multiple agent entity brand groups are input into a corporate brand relationship prediction model to determine the brand of the agent entity, wherein the corporate brand relationship prediction model is obtained by training and learning training samples marked with the correspondence between the corporate name of the agent entity and the brand.

[0037] Optionally, the processing unit is specifically configured to:

[0038] Using a pre-trained word vector model, vectorize the company names in the agent entity brand group to obtain vectors of the company names;

[0039] Using a one-hot model to vectorize the brand of the agent to be determined in the agent main brand group, to obtain a vector of the brand of the agent to be determined;

[0040] The vector of the enterprise name and the vector of the brand of the to-be-determined agent are determined as the feature vector of the agent-entity brand group.

[0041] Optionally, the processing unit is specifically configured to:

[0042] Combining the corporate name of the agent and the brand of the agent to be determined to obtain the plurality of agent brand groups; and for any of the plurality of agent brand groups, performing word segmentation processing on the corporate name in the agent brand group and performing category coding processing on the brand of the agent to be determined in the agent brand group; or

[0043] The corporate name of the agency entity is segmented, and the brand of the agency to be determined is categorized and coded; and the corporate name after segmentation and the brand of the agency to be determined after categorization are combined to obtain the multiple agency entity brand groups.

[0044] Optionally, the processing unit is specifically configured to:

[0045] Segment the company names in the agent entity brand group to obtain the company name, region, business scope and business scale of the agent entity;

[0046] The corporate name, region, business scope, and business scale of the agent are vectorized using a pre-trained word vector model to obtain a vector of the corporate name.

[0047] Optionally, the processing unit is specifically configured to:

[0048] Obtaining training samples of the corresponding relationship between the company name and brand marked with the agent entity;

[0049] Performing word segmentation processing on the corporate names of the agent entities in the training samples in the corresponding relationship with the brands, and performing category coding processing on the brands represented by the agent entities;

[0050] Vectorizing the company name after word segmentation processing and the brand after category coding processing in the correspondence between the company name of the agent in the training sample and the brand, to obtain a feature vector of the correspondence between the company name of the agent in the training sample and the brand;

[0051] The feature vectors of the correspondence between the corporate name and brand of the agent in the training sample are sequentially input into a preset machine learning model for training until the model converges, thereby obtaining the corporate-brand relationship prediction model.

[0052] Optionally, the processing unit is specifically configured to:

[0053] Inputting the feature vectors of the plurality of agent-entity brand groups into the enterprise-brand relationship prediction model in sequence to determine the prediction probability corresponding to each agent-entity brand group;

[0054] The brand of the to-be-determined agent with a predicted probability greater than a threshold is determined as the brand of the agent subject agent.

[0055] Optionally, the processing unit is further configured to:

[0056] After the brand represented by the agent is determined, the enterprise name of the agent and the corresponding brand represented by the agent are stored in the training sample, and the enterprise-brand relationship prediction model is retrained.

[0057] In a third aspect, the present invention provides a computing device, comprising:

[0058] memory for storing computer programs;

[0059] The processor is configured to call the computer program stored in the memory and execute the method described in the first aspect according to the obtained program.

[0060] In a fourth aspect, the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer-executable program, and the computer-executable program is used to enable a computer to execute the method described in the first aspect. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0062] Figure 1 A schematic diagram of a system architecture provided by an embodiment of the present invention;

[0063] Figure 2 A flowchart of a method for determining a brand represented by an agent provided by an embodiment of the present invention;

[0064] Figure 3 A schematic diagram of the structure of a device for determining a brand represented by an agent provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0065] To make the objectives, technical solutions, and advantages of the present invention more apparent, the present invention will be further described in detail below with reference to the accompanying drawings. It is apparent that the embodiments described are only some of the embodiments of the present invention, rather than all of them. All other embodiments derived by persons of ordinary skill in the art based on the embodiments of the present invention without creative effort are intended to fall within the scope of protection of the present invention.

[0066] Figure 1 A system architecture is provided for an embodiment of the present invention. Figure 1 As shown, the system architecture may be a server 100 , including a processor 110 , a communication interface 120 and a memory 130 .

[0067] The communication interface 120 is used to communicate with the terminal device, send and receive information transmitted by the terminal device, and realize communication.

[0068] The processor 110 is the control center of the server 100. It connects various parts of the server 100 using various interfaces and lines. It executes various functions of the server 100 and processes data by running or executing software programs and / or modules stored in the memory 130 and calling data stored in the memory 130. Optionally, the processor 110 may include one or more processing units.

[0069] Memory 130 can be used to store software programs and modules. Processor 110 executes various functional applications and data processing by running the software programs and modules stored in memory 130. Memory 130 may primarily include a program storage area and a data storage area. The program storage area may store an operating system, at least one application required for a function, and the data storage area may store data generated based on business processing. Memory 130 may also include high-speed random access memory and non-volatile memory, such as at least one disk storage device, flash memory device, or other volatile solid-state memory device.

[0070] It should be noted that the above Figure 1 The structure shown is only an example and is not limited in the embodiment of the present invention.

[0071] Based on the above description, Figure 2 The flow of a method for determining a brand represented by an agent subject provided by an embodiment of the present invention is exemplarily shown. The flow can be executed by a device for determining a brand represented by an agent subject.

[0072] like Figure 2 As shown, the specific steps of the process include:

[0073] Step 201: Obtain the company name of the agent and the brand of the agent to be determined.

[0074] In an embodiment of the present invention, the company name of the agent and the brand to be determined by the agent can be obtained. The company name is the overall name of the company, for example, Shanghai XXX Daily Chemical Trading Co., Ltd. The brand to be determined may include A, B, and C.

[0075] It should be noted that the agent in the embodiment of the present invention may be a dealer or other enterprise operating an agent brand.

[0076] Step 202 : pre-processing the corporate name of the agent and the brand of the agent to be determined to obtain a plurality of agent brand groups.

[0077] Specifically, the pre-processing of the corporate name of the agency entity and the brand to be determined by the agency entity is mainly carried out in the following two ways:

[0078] Method 1:

[0079] First, the company name of the agent and the brand to be determined as an agent are combined to obtain multiple agent brand groups. Then, for any of the multiple agent brand groups, the company name in the agent brand group is segmented and the brand to be determined as an agent in the agent brand group is categorized and coded.

[0080] In other words, once the agency's corporate name and the brand it's seeking to represent are obtained, they can be grouped. Each agency brand group includes the agency's corporate name and the corresponding brand it's seeking to represent. For example, the first agency brand group includes Shanghai XXX Daily Chemical Trading Co., Ltd. - A; the second agency brand group includes Shanghai XXX Daily Chemical Trading Co., Ltd. - B; and the third agency brand group includes Shanghai XXX Daily Chemical Trading Co., Ltd. - C. This results in three agency brand groups.

[0081] Then, the company names in the three agent brand groups can be segmented and the brands to be determined as agents can be categorized and coded. By segmenting the company names, the company name, region, business scope, and scale of the agent can be obtained.

[0082] For example, by performing word segmentation on Shanghai XXX Daily Chemical Trading Co., Ltd., we can obtain the company name after word segmentation = "Shanghai XXX Daily Chemical Trading Co., Ltd.". The company's location, business scope, and scale can be determined through the company name after word segmentation. That is, the company's location is Shanghai, its business scope is daily chemical products, and its scale is a limited liability company.

[0083] In addition, each brand has a preset category code. By searching the preset category code, the category code corresponding to each brand can be determined, that is, the brand to be determined as the agent can be subjected to category coding processing.

[0084] Method 2:

[0085] First, the agency entity's corporate name is segmented, and the agency entity's brand to be determined is categorized and coded. Then, the segmented corporate name and categorized brand to be determined are combined to obtain multiple agency entity brand groups.

[0086] This method is to first perform word segmentation and category coding, and then perform grouping to obtain multiple agent brand groups. The word segmentation and category coding process has been described in the above method 1 and will not be repeated here.

[0087] Step 203 : for any agent principal brand group among the plurality of agent principal brand groups, vectorize the enterprise name and the brand of the agent to be determined in the agent principal brand group to obtain a feature vector of the agent principal brand group.

[0088] Specifically, a pre-trained word vector model can be used to vectorize the company names in the agent-based brand group to obtain the company name vector. A one-hot model can be used to vectorize the brand of the agent to be determined in the agent-based brand group to obtain the brand vector of the agent to be determined. Finally, the vectors of the company names and the brand of the agent to be determined are determined as the feature vector of the agent-based brand group. Specifically, the vectors of the company names and the brand of the agent to be determined are arranged to obtain the feature vector of the agent-based brand group.

[0089] It should be noted that when segmenting the company names within the agent brand group, the company name, region, business scope, and scale of the agent can also be obtained. In other words, when vectorizing the company name to obtain the vector for the company name, a pre-trained word vector model can be used to vectorize the company name, region, business scope, and scale of the agent to obtain the vector for the company name.

[0090] When the above-mentioned vectorization processing is performed on the company name and the brand of the agent to be determined, the word vector model used is not limited to the above-mentioned models. Other word vector models can also be used, such as the word vector model Word2vec (Word to vector), GloVe (Global Vectors), ELMo (Embeddings from Language Models), etc.

[0091] Step 204: Input the feature vectors of the plurality of agent-entity brand groups into an enterprise-brand relationship prediction model to determine the brand represented by the agent-entity.

[0092] Before using the enterprise-brand relationship prediction model, training and learning are required. Specifically, training samples can be obtained that are marked with the correspondence between the agent's enterprise name and brand. Then, the enterprise names in the correspondence between the agent's enterprise name and brand in the training sample are segmented, and the brands represented by the agent are categorized. The segmented enterprise names and categorized brands in the correspondence between the agent's enterprise name and brand in the training sample are vectorized to obtain a feature vector for the correspondence between the agent's enterprise name and brand in the training sample. Finally, the feature vectors of the correspondence between the agent's enterprise name and brand in the training sample are sequentially input into a preset machine learning model for training until the model converges, thereby obtaining an enterprise-brand relationship prediction model.

[0093] In an embodiment of the present invention, when training the enterprise brand relationship prediction model, the preset machine learning model used may be a factorization machine (FM) model, but is not limited to the machine learning model of the factorization machine, and may also be other machine learning models, such as decision trees, naive Bayes classifiers, least squares methods, support vector machines, etc.

[0094] The training sample includes both positive and negative samples. Positive samples are assigned a value of 1, while negative samples are assigned a value of 0. First, we need to collect a small amount of data on the agents and their corresponding brands, as well as some company names without any brand agency information. We segment all company names and classify all agent brands. Based on the segmentation results, we determine the agent's corporate name, region, business scope, and scale. We combine the company name and the brand it represents to create new data as positive samples. We randomly assemble a list of company names and brands for agents whose brands are unknown as negative samples.

[0095] For example, known agent-brand relationships include:

[0096] Agent A - Brand A / B;

[0097] Agent B-Brand B;

[0098] Agent C-Brand B / C.

[0099] And the agent DEF of unknown brand agency relationship, then the positive samples that can be combined are:

[0100] Agent A, Brand A, 1;

[0101] Agent A, Brand B, 1;

[0102] Agent B, Brand B, 1;

[0103] Agent C, Brand B, 1;

[0104] Agent C, Brand C, 1.

[0105] The corresponding combined negative samples are:

[0106] Agent D, Brand A, 0;

[0107] Agent D, Brand B, 0;

[0108] Agent D, Brand C, 0;

[0109] Agent E, Brand A, 0;

[0110] Agent E, Brand B, 0;

[0111] Agent E, Brand C, 0;

[0112] Agent E, Brand A, 0;

[0113] Agent E, Brand B, 0;

[0114] Agent E, brand C, 0.

[0115] The above training sample data is trained using a factorization machine model to obtain an enterprise-brand relationship prediction model that can estimate the relationship between enterprises and brands.

[0116] When using this enterprise-brand relationship prediction model, the feature vectors of multiple agent-based brand groups can be input into the enterprise-brand relationship prediction model to determine the brands represented by the agent. Specifically, the feature vectors of multiple agent-based brand groups can be sequentially input into the enterprise-brand relationship prediction model to determine the predicted probability corresponding to each agent-based brand group. The brands to be represented by the agent whose predicted probability exceeds a threshold are then determined as the brands represented by the agent. The threshold can be set based on experience.

[0117] In addition, after determining the brand represented by the agent, the agent's company name and its corresponding brand can be stored in the training sample to retrain the enterprise-brand relationship prediction model, further optimizing the model and continuously improving the results.

[0118] In this embodiment of the present invention, only a small number of agent names and corresponding brands are needed to discover brands that other agents may represent, rapidly expanding the customer base for supply chain finance products. Furthermore, the use of a factorization machine algorithm that automatically mines feature associations effectively reduces the workload of algorithm developers and rapidly develops a predictive model for enterprise-brand relationships.

[0119] The embodiment of the present invention demonstrates that by obtaining the corporate name of the agent and the brand of the agent to be determined, the corporate name of the agent and the brand of the agent to be determined are pre-processed to obtain multiple agent brand groups, wherein each agent brand group includes the corporate name of the agent and its corresponding brand to be determined. For any of the multiple agent brand groups, the corporate name and the brand of the agent to be determined in the agent brand group are vectorized to obtain a feature vector of the agent brand group. The feature vectors of the multiple agent brand groups are input into a corporate-brand relationship prediction model to determine the brand of the agent. The corporate-brand relationship prediction model is obtained by training and learning training samples marked with the corresponding relationship between the corporate name of the agent and the brand. Since the brand represented by the agent is predicted by the enterprise-brand relationship prediction model obtained by training and learning using training samples marked with the correspondence between the enterprise name and brand of the agent, the brand represented by the agent can be quickly identified, thereby improving the efficiency of identifying the agent's brand. Compared with the existing technology of manually searching for the agent and the agent's brand through a small amount of upstream and downstream data, the accuracy of determining the brand represented by the agent can be improved.

[0120] Based on the same technical concept, Figure 3 The schematic diagram of the structure of a device for determining a brand represented by an agent provided by an embodiment of the present invention is exemplarily shown. The device can execute the process of determining a brand represented by an agent.

[0121] like Figure 3 As shown, the device specifically includes:

[0122] An acquisition unit 301 is used to acquire the company name of the agent and the brand of the agent to be determined;

[0123] The processing unit 302 is used to pre-process the corporate name of the agent entity and the brand of the agent to be determined by the agent entity to obtain multiple agent entity brand groups; each agent entity brand group includes the corporate name of the agent entity and its corresponding brand of the agent to be determined; for any agent entity brand group among the multiple agent entity brand groups, the corporate name and the brand of the agent to be determined in the agent entity brand group are vectorized to obtain a feature vector of the agent entity brand group; the feature vectors of the multiple agent entity brand groups are input into a corporate brand relationship prediction model to determine the brand of the agent entity, wherein the corporate brand relationship prediction model is obtained by training and learning training samples marked with the correspondence between the corporate name of the agent entity and the brand.

[0124] Optionally, the processing unit 302 is specifically configured to:

[0125] Using a pre-trained word vector model, vectorize the company names in the agent entity brand group to obtain vectors of the company names;

[0126] Using a one-hot model to vectorize the brand of the agent to be determined in the agent main brand group, to obtain a vector of the brand of the agent to be determined;

[0127] The vector of the enterprise name and the vector of the brand of the to-be-determined agent are determined as the feature vector of the agent-entity brand group.

[0128] Optionally, the processing unit 302 is specifically configured to:

[0129] Combining the corporate name of the agent and the brand of the agent to be determined to obtain the plurality of agent brand groups; and for any of the plurality of agent brand groups, performing word segmentation processing on the corporate name in the agent brand group and performing category coding processing on the brand of the agent to be determined in the agent brand group; or

[0130] The corporate name of the agency entity is segmented, and the brand of the agency to be determined is categorized and coded; and the corporate name after segmentation and the brand of the agency to be determined after categorization are combined to obtain the multiple agency entity brand groups.

[0131] Optionally, the processing unit 302 is specifically configured to:

[0132] Segment the company names in the agent entity brand group to obtain the company name, region, business scope and business scale of the agent entity;

[0133] The corporate name, region, business scope, and business scale of the agent are vectorized using a pre-trained word vector model to obtain a vector of the corporate name.

[0134] Optionally, the processing unit 302 is specifically configured to:

[0135] Obtaining training samples of the corresponding relationship between the company name and brand marked with the agent entity;

[0136] Performing word segmentation processing on the corporate names of the agent entities in the training samples in the corresponding relationship with the brands, and performing category coding processing on the brands represented by the agent entities;

[0137] Vectorizing the company name after word segmentation processing and the brand after category coding processing in the correspondence between the company name of the agent in the training sample and the brand, to obtain a feature vector of the correspondence between the company name of the agent in the training sample and the brand;

[0138] The feature vectors of the correspondence between the corporate name and brand of the agent in the training sample are sequentially input into a preset machine learning model for training until the model converges, thereby obtaining the corporate-brand relationship prediction model.

[0139] Optionally, the processing unit 302 is specifically configured to:

[0140] Inputting the feature vectors of the plurality of agent-entity brand groups into the enterprise-brand relationship prediction model in sequence to determine the prediction probability corresponding to each agent-entity brand group;

[0141] The brand of the to-be-determined agent with a predicted probability greater than a threshold is determined as the brand of the agent subject agent.

[0142] Optionally, the processing unit 302 is further configured to:

[0143] After the brand represented by the agent is determined, the enterprise name of the agent and the corresponding brand represented by the agent are stored in the training sample, and the enterprise-brand relationship prediction model is retrained.

[0144] Based on the same technical concept, an embodiment of the present invention provides a computing device, including:

[0145] memory for storing computer programs;

[0146] The processor is configured to call the computer program stored in the memory and execute the method for determining the brand of the agent according to the obtained program.

[0147] Based on the same technical concept, an embodiment of the present invention provides a computer-readable storage medium storing a computer-executable program for causing a computer to execute the above-mentioned method for determining the brand of an agent subject.

[0148] It will be understood by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0149] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0150] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0151] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 The steps for the function specified in one or more boxes.

[0152] Although the preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0153] Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if such changes and modifications of the present invention fall within the scope of the claims and their equivalents, the present invention is intended to include such changes and modifications.

Claims

1. A method for determining the brand represented by an agent, characterized in that: include: Obtain the company name of the agent and the brand of the agent to be determined; Preprocessing the corporate name of the agent and the brand of the agent to be determined to obtain a plurality of agent brand groups; each agent brand group includes the corporate name of the agent and its corresponding brand to be determined; For any agent principal brand group among the plurality of agent principal brand groups, vectorizing the enterprise name and the brand of the agent to be determined in the agent principal brand group to obtain a feature vector of the agent principal brand group; The feature vectors of the plurality of agent subject brand groups are input into an enterprise brand relationship prediction model to determine the brand represented by the agent subject, wherein the enterprise brand relationship prediction model is obtained by training and learning training samples marked with the correspondence between the agent subject's enterprise name and the brand.

2. The method according to claim 1, wherein The vectorization processing of the enterprise names and the brands of the agents to be determined in the agent entity brand group to obtain the feature vectors of the agent entity brand group includes: Using a pre-trained word vector model, vectorize the company names in the agent entity brand group to obtain vectors of the company names; Using a one-hot model to vectorize the brand of the agent to be determined in the agent main brand group, to obtain a vector of the brand of the agent to be determined; The vector of the enterprise name and the vector of the brand of the to-be-determined agent are determined as the feature vector of the agent-entity brand group.

3. The method according to claim 2, wherein The pre-processing of the corporate name of the agent and the brand of the agent to be determined to obtain a plurality of agent brand groups includes: Combining the corporate name of the agent and the brand of the agent to be determined to obtain the plurality of agent brand groups; and for any of the plurality of agent brand groups, performing word segmentation processing on the corporate name in the agent brand group and performing category coding processing on the brand of the agent to be determined in the agent brand group; or The corporate name of the agency entity is segmented, and the brand of the agency to be determined is categorized and coded; and the corporate name after segmentation and the brand of the agency to be determined after categorization are combined to obtain the multiple agency entity brand groups.

4. The method according to claim 3, wherein The word segmentation processing of the enterprise names in the agent entity brand group includes: Segment the company names in the agent entity brand group to obtain the company name, region, business scope and business scale of the agent entity; The company names in the agent entity brand group are vectorized using a pre-trained word vector model to obtain vectors of the company names, including: The corporate name, region, business scope, and business scale of the agent are vectorized using a pre-trained word vector model to obtain a vector of the corporate name.

5. The method according to claim 1, wherein The step of training the training samples marked with the corresponding relationship between the enterprise name and the brand of the agent to obtain the enterprise-brand relationship prediction model includes: Obtaining training samples of the corresponding relationship between the company name and brand marked with the agent entity; Performing word segmentation processing on the corporate names of the agent entities in the training samples in the corresponding relationship with the brands, and performing category coding processing on the brands represented by the agent entities; Vectorizing the company name after word segmentation processing and the brand after category coding processing in the correspondence between the company name of the agent in the training sample and the brand, to obtain a feature vector of the correspondence between the company name of the agent in the training sample and the brand; The feature vectors of the correspondence between the corporate name and brand of the agent in the training sample are sequentially input into a preset machine learning model for training until the model converges, thereby obtaining the corporate-brand relationship prediction model.

6. The method according to claim 1, wherein Inputting the feature vectors of the plurality of agent brand groups into the enterprise brand relationship prediction model to determine the brand represented by the agent includes: Inputting the feature vectors of the plurality of agent-entity brand groups into the enterprise-brand relationship prediction model in sequence to determine the prediction probability corresponding to each agent-entity brand group; The brand of the to-be-determined agent with a predicted probability greater than a threshold is determined as the brand of the agent subject agent.

7. The method according to any one of claims 1 to 6, wherein: After determining the brand represented by the agent, the method further includes: The enterprise name of the agent entity and the brand of its corresponding agent are stored in the training sample, and the enterprise-brand relationship prediction model is retrained.

8. A device for determining the brand represented by an agent, characterized in that: include: An acquisition unit, configured to acquire the corporate name of an agent and the brand of the agent to be determined; A processing unit is used to pre-process the corporate name of the agent entity and the brand of the agent to be determined by the agent entity to obtain multiple agent entity brand groups; each agent entity brand group includes the corporate name of the agent entity and its corresponding brand of the agent to be determined; for any agent entity brand group among the multiple agent entity brand groups, the corporate name and the brand of the agent to be determined in the agent entity brand group are vectorized to obtain a feature vector of the agent entity brand group; the feature vectors of the multiple agent entity brand groups are input into a corporate brand relationship prediction model to determine the brand of the agent entity, wherein the corporate brand relationship prediction model is obtained by training and learning training samples marked with the correspondence between the corporate name of the agent entity and the brand.

9. A computing device, characterized in that include: memory for storing computer programs; A processor, configured to call a computer program stored in the memory, and execute the method according to any one of claims 1 to 7 according to the obtained program.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer-executable program, and the computer-executable program is used to enable a computer to execute the method according to any one of claims 1 to 7.

Images