Business recommendation model training method, device, equipment and storage medium

Abstract

The present invention relates to intelligent decision-making technology and discloses a business recommendation model training method, comprising: extracting data of each target field in each historical inquiry record according to a preset target field set to obtain an inquiry keyword set and a business processing keyword set; importing the inquiry keyword set into a pre-trained inquiry recommendation model to obtain a predicted recommendation keyword set; calculating the similarity value between the business processing keyword set and the predicted recommendation keyword set; obtaining the hit rate of the inquiry recommendation model according to the similarity value and a preset standard threshold; training the inquiry recommendation model according to the hit rate and historical inquiry records to obtain a standard inquiry recommendation model. The present invention also proposes a business recommendation model training device, electronic device and storage medium. The present invention can improve the accuracy of the business recommendation model.

Description

Technical Field

[0001] The present invention relates to the field of intelligent decision-making technology, and in particular to a business recommendation model training method, device, electronic device and computer-readable storage medium. Background Art

[0002] To ensure that they receive cost-effective products or services, customers usually go through a price inquiry process before ordering products or conducting business.

[0003] Take insurance sales as an example. Before purchasing insurance, customers typically present their needs to sales representatives, asking for pricing and recommendations. Sales representatives typically assess customer needs and recommend insurance plans using pre-built recommendation models. However, as market trends evolve over time, these models gradually fail to keep pace with evolving customer needs. This makes it difficult for these models to quickly identify satisfactory plans, reducing customer interest and significantly reducing business efficiency. Therefore, there is an urgent need for methods to improve the accuracy of recommendation models. Summary of the Invention

[0004] The present invention provides a business recommendation model training method, device and computer-readable storage medium, the main purpose of which is to improve the accuracy of the business recommendation model.

[0005] To achieve the above objectives, the present invention provides a service recommendation model training method, comprising:

[0006] According to a preset time frequency, each historical inquiry record within a preset time period is obtained, and according to a preset target field set, data of each target field in each historical inquiry record is extracted to obtain an inquiry keyword set and a business processing keyword set corresponding to each historical inquiry record;

[0007] Importing the inquiry keyword set corresponding to each historical inquiry record into a pre-trained inquiry recommendation model to obtain a predicted recommended keyword set corresponding to each historical inquiry record;

[0008] Based on the pre-built similarity matching rules, similarity calculation is performed on the business processing keyword set corresponding to each historical inquiry record and the predicted recommendation keyword set to obtain a similarity value set;

[0009] Obtaining a hit rate of the price inquiry recommendation model based on the similarity value set and a preset standard threshold;

[0010] The price inquiry recommendation model is trained according to the hit rate and the various historical price inquiry records to obtain a standard price inquiry recommendation model.

[0011] Optionally, extracting data of each target field in each historical inquiry record according to a preset target field set to obtain an inquiry keyword set and a business processing keyword set corresponding to each historical inquiry record includes:

[0012] According to each target field of the preset target field set, data corresponding to each target field is obtained from each historical inquiry record to obtain an initial keyword set corresponding to each historical inquiry record;

[0013] According to the keyword category of each initial keyword in the initial keyword set, the initial keywords corresponding to each historical inquiry record are grouped to obtain the inquiry keyword set and the business processing keyword set corresponding to each historical inquiry record.

[0014] Optionally, the step of importing the inquiry keyword sets corresponding to the respective historical inquiry records into a pre-trained inquiry recommendation model to obtain the predicted recommendation keyword sets corresponding to the respective historical inquiry records includes:

[0015] Formatting the inquiry keyword set corresponding to each historical inquiry record using a preset program interactive interface to obtain input parameter data corresponding to each historical inquiry record;

[0016] Importing the input data corresponding to each historical inquiry record into the inquiry recommendation model, using the inquiry recommendation model to identify the main inquiry business in each input data, and querying the recommended solution corresponding to each main inquiry business;

[0017] Identify the keywords of each of the recommended solutions and obtain a set of predicted recommended keywords corresponding to each of the historical inquiry records.

[0018] Optionally, the similarity calculation is performed on the business processing keyword set corresponding to each historical inquiry record and the predicted recommendation keyword set according to the pre-built similarity matching rule to obtain a similarity value set, including:

[0019] Using a cosine similarity algorithm, the relative distance between the business processing keyword set corresponding to each historical inquiry record and the predicted recommendation keyword set is calculated to obtain a relative distance set corresponding to each historical inquiry record;

[0020] According to a preset weight configuration rule, a weighted sum calculation is performed on each relative distance in the relative distance set corresponding to each historical inquiry record to obtain a similarity value set.

[0021] Optionally, obtaining the hit rate of the price inquiry recommendation model according to the similarity value set and a preset standard threshold includes:

[0022] Determine whether the similarity value corresponding to each historical inquiry record in the similarity value set is greater than a preset standard threshold;

[0023] When the similarity value is greater than a preset standard threshold, it is determined that the inquiry recommendation model predicts and hits the historical inquiry record corresponding to the similarity value;

[0024] When the similarity value is less than or equal to the standard threshold, it is determined that the price inquiry recommendation model does not predict the historical price inquiry record corresponding to the similarity value;

[0025] The hit rate of the price inquiry recommendation model is obtained according to the hit results of each of the historical price inquiry records.

[0026] Optionally, the training of the price inquiry recommendation model according to the hit rate and the historical price inquiry records to obtain a standard price inquiry recommendation model includes:

[0027] Determining whether the hit rate is less than a preset training threshold;

[0028] When the hit rate is less than a preset training threshold, the inquiry recommendation model is trained using the historical inquiry records according to a K-fold cross-validation method to obtain a standard inquiry recommendation model;

[0029] When the hit rate is greater than or equal to the training threshold, the price inquiry recommendation model is determined to be a standard price inquiry recommendation model.

[0030] Optionally, obtaining each historical inquiry record within a preset time period according to a preset time frequency includes:

[0031] Use preset tracking points to record a set of business records within a preset time period in a pre-built log space;

[0032] Using pre-configured data retrieval permissions, automatically uploading the business record set to a pre-built training sample database according to a preset time frequency;

[0033] The business record set is parsed using a pre-built parsing function in the training sample database to obtain various historical inquiry records.

[0034] In order to solve the above problems, the present invention further provides a service recommendation model training device, the device comprising:

[0035] A historical data acquisition module is used to acquire historical inquiry records within a preset time period according to a preset time frequency, and extract data of each target field in each historical inquiry record according to a preset target field set to obtain the inquiry keyword set and business processing keyword set corresponding to the historical inquiry record;

[0036] A model prediction module, configured to import the inquiry keyword set into a pre-trained inquiry recommendation model to obtain a predicted recommendation keyword set;

[0037] A hit judgment module is used to calculate the similarity value between the business processing keyword set and the predicted recommendation keyword set according to a pre-established similarity matching rule, and obtain the hit rate of the inquiry recommendation model based on the similarity value of the historical inquiry records within the time period and a preset standard threshold;

[0038] The training judgment module is used to train the price inquiry recommendation model according to the hit rate and the historical price inquiry records in the time period to obtain a standard price inquiry recommendation model.

[0039] In order to solve the above problem, the present invention further provides an electronic device, comprising:

[0040] at least one processor; and,

[0041] a memory communicatively connected to the at least one processor; wherein,

[0042] The memory stores a computer program that can be executed by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the business recommendation model training method described above.

[0043] In order to solve the above problems, the present invention also provides a computer-readable storage medium, in which at least one computer program is stored. The at least one computer program is executed by a processor in an electronic device to implement the above-mentioned business recommendation model training method.

[0044] The embodiment of the present application obtains historical inquiry records in a regular and quantitative manner according to the time frequency and the time period, and ensures the accuracy of the inquiry recommendation model by checking the hit rate of the historical inquiry records with the pre-trained inquiry recommendation model, wherein the inquiry recommendation model can quickly generate a predicted recommendation keyword set based on the inquiry keyword set, thereby improving the efficiency of solution recommendation; in addition, since the historical inquiry records are generated by users through negotiation on the basis of the predicted recommendation keyword set, the embodiment of the present application continuously trains the inquiry recommendation model through the historical inquiry records, which can make the accuracy of the inquiry recommendation model gradually approach user expectations and improve user satisfaction. Therefore, the business recommendation model training method, device, equipment and storage medium proposed in the embodiment of the present application can improve the accuracy of the business recommendation model. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 A flowchart of a business recommendation model training method provided by one embodiment of the present invention;

[0046] Figure 2 A detailed flowchart of a step in a business recommendation model training method provided by one embodiment of the present invention;

[0047] Figure 3 A detailed flowchart of a step in a business recommendation model training method provided by one embodiment of the present invention;

[0048] Figure 4 A functional module diagram of a business recommendation model training device provided by one embodiment of the present invention;

[0049] Figure 5 A schematic structural diagram of an electronic device for implementing the business recommendation model training method provided by one embodiment of the present invention.

[0050] The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings and in conjunction with the embodiments. DETAILED DESCRIPTION

[0051] It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0052] An embodiment of the present application provides a business recommendation model training method. In the embodiment of the present application, the execution subject of the business recommendation model training method includes but is not limited to at least one of the electronic devices such as the server, the terminal, etc. that can be configured to execute the method provided by the embodiment of the present application. In other words, the business recommendation model training method can be executed by software or hardware installed on the terminal device or the server device, and the software can be a blockchain platform. The server includes but is not limited to: a single server, a server cluster, a cloud server or a cloud server cluster, etc. The server can be an independent server, or it can be a cloud server that provides cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (CDN), and basic cloud computing services such as big data and artificial intelligence platforms.

[0053] Reference Figure 1 FIG. 1 is a flow chart of a method for training a business recommendation model according to an embodiment of the present invention. In this embodiment, the method for training a business recommendation model includes:

[0054] S1. According to a preset time frequency, obtain each historical inquiry record within a preset time period, and according to a preset target field set, extract the data of each target field in each historical inquiry record to obtain the inquiry keyword set and business processing keyword set corresponding to each historical inquiry record.

[0055] In an embodiment of the present application, the time frequency is set to once a day or once a week, and the time period is one day or one week. For example, if the time frequency is daily, the time frequency in an embodiment of the present application can be set to: at N o'clock in the morning every day (such as 3 o'clock in the morning), various business data registered in the enterprise business system of the previous day are obtained as historical inquiry records.

[0056] For more details, please refer to Figure 2 As shown, in the embodiment of the present application, obtaining each historical inquiry record within a preset time period according to a preset time frequency includes:

[0057] S11. Using preset tracking points, record a set of business records within a preset time period in a pre-built log space;

[0058] S12. Using pre-configured data retrieval permissions, automatically uploading the business record set to a pre-built training sample database according to a preset time frequency;

[0059] S13: Analyze the business record set using a pre-built analytical function in the training sample database to obtain various historical inquiry records.

[0060] Among them, the embodiment of the present application can use the pre-built training sample database as the parent database of the log space through pre-configured data retrieval permissions, and does not require repeated authentication during data transmission.

[0061] Specifically, the embodiment of the present application first records the updated business records in the log space of the enterprise business system through preset tracking points to obtain a business record set; then, when the time point preset by the user is reached, such as three o'clock in the morning every day, the database tool, such as sqoop, is automatically enabled to access the log space and retrieve data, and transmit it to the training sample database; then, the business record set is parsed using the pre-built json parsing function in the training sample database to obtain historical inquiry records.

[0062] In detail, in an embodiment of the present application, according to a preset target field set, data of each target field in each historical inquiry record is extracted to obtain an inquiry keyword set and a business processing keyword set corresponding to each historical inquiry record, including:

[0063] According to each target field of the preset target field set, data corresponding to each target field is obtained from each historical inquiry record to obtain an initial keyword set corresponding to each historical inquiry record;

[0064] According to the keyword category of each initial keyword in the initial keyword set, the initial keywords corresponding to each historical inquiry record are grouped to obtain the inquiry keyword set and the business processing keyword set corresponding to each historical inquiry record.

[0065] In an embodiment of the present application, the target field set is a set of various fields obtained by screening business records according to the principal component analysis method, wherein the principal component analysis method is a statistical method for obtaining the method with the highest correlation between each feature and the analysis result. For example, in the insurance sales industry, after the business policy involves [issues such as [target to be protected, region, industry type, scale, term, premium, amount], according to the principal component analysis method, it can be known that fields such as [target to be protected, region, industry type, premium, amount] have the highest correlation in the insurance recommendation process. Therefore, the target field set in the embodiment of the present application can be defined as [inquiry form product field (target to be protected), secondary institution field (region) and tertiary industry field (industry type), premium, amount...].

[0066] Furthermore, in order to build a model, it is necessary to select input and output parameters for the model. Therefore, the embodiment of the present application isolates the user's data from the enterprise's data to obtain the inquiry keyword set [inquiry form product field, secondary institution field, tertiary industry field] and the business processing keyword set [premium, insured amount, limit].

[0067] S2. Importing the inquiry keyword set corresponding to each historical inquiry record into a pre-trained inquiry recommendation model to obtain a predicted recommendation keyword set corresponding to each historical inquiry record.

[0068] In an embodiment of the present application, the inquiry recommendation model is a classification model based on a decision tree constructed according to historical inquiry records, and can perform relevant numerical judgments based on each keyword in the inquiry keyword set to obtain a set of product recommendation solutions.

[0069] Furthermore, the predicted recommendation keyword set is each keyword in the product recommendation solution.

[0070] For more details, please refer to Figure 3 As shown, in the embodiment of the present application, the inquiry keyword set corresponding to each historical inquiry record is imported into the pre-trained inquiry recommendation model to obtain the predicted recommendation keyword set corresponding to each historical inquiry record, including:

[0071] S21. Formatting the inquiry keyword set corresponding to each historical inquiry record using a preset program interactive interface to obtain input parameter data corresponding to each historical inquiry record;

[0072] S22: Importing the input data corresponding to each historical inquiry record into the inquiry recommendation model, using the inquiry recommendation model to identify the main inquiry business in each input data, and querying the recommended solution corresponding to each main inquiry business;

[0073] S23: Identify the keywords of each of the recommended solutions, and obtain a set of predicted recommended keywords corresponding to each of the historical inquiry records.

[0074] Specifically, in order to unify the data format of the input of the inquiry recommendation model and ensure the normal execution of the inquiry recommendation model, the embodiment of the present application uses a preset API to format the inquiry keyword set through operations such as data sequence arrangement, missing value query and filling, and obtain the input parameter data corresponding to each historical inquiry record; then use the inquiry recommendation model to quantify, extract features, identify features and make decision tree judgments on each of the input parameter data to obtain a recommended solution.

[0075] Among them, in order to determine whether the recommended solution is similar to the keywords of the business example that has been implemented in the historical inquiry record, the embodiment of the present application needs to extract the keywords in the recommended solution to obtain a set of predicted recommended keywords.

[0076] Furthermore, in one embodiment of the present invention, the pre-trained price inquiry recommendation model can be obtained by the following method:

[0077] Step 1: obtaining a pre-built feature sample training set, and using a pre-built inquiry recommendation model to identify the feature sample training set to obtain a prediction result set;

[0078] Step II: determining whether the prediction result set is the same as the pre-built sample label set, and obtaining the accuracy of the inquiry recommendation model;

[0079] Step III: determining the convergence of the accuracy rate; if the accuracy rate has not converged, modifying the function parameters of the inquiry recommendation model, and returning to the above step I;

[0080] Step IV: When the accurate value converges, exit the training process of the price inquiry recommendation model to obtain a trained price inquiry recommendation model.

[0081] Among them, the feature sample training set contains various condition sets proposed by customers, and the sample label set is a set of business information that customers finally successfully handle. In one embodiment of the present invention, through the above steps I to IV, a trained inquiry recommendation model is obtained.

[0082] S3. Based on a pre-built similarity matching rule, similarity is calculated between the business processing keyword set corresponding to each historical inquiry record and the predicted recommendation keyword set to obtain a similarity value set.

[0083] In the embodiment of the present application, the similarity matching rule may be an algorithm capable of determining the similarity between words, such as a cosine similarity algorithm, wherein the cosine similarity algorithm is a method for clustering similar words by calculating the distance between words using a cosine algorithm.

[0084] In detail, in the embodiment of the present application, the similarity calculation is performed on the business processing keyword set corresponding to each historical inquiry record and the predicted recommendation keyword set according to the pre-built similarity matching rule to obtain a similarity value set, including:

[0085] Using a cosine similarity algorithm, the relative distance between the business processing keyword set corresponding to each historical inquiry record and the predicted recommendation keyword set is calculated to obtain a relative distance set corresponding to each historical inquiry record;

[0086] According to a preset weight configuration rule, a weighted sum calculation is performed on each relative distance in the relative distance set corresponding to each historical inquiry record to obtain a similarity value set.

[0087] Specifically, the embodiment of the present application uses the cosine similarity algorithm to calculate the relative distance between the business processing keyword set and the predicted recommendation keyword set to obtain a relative distance set, for example, [“D 办理保额-预测保额 ”, “D 办理保费-预测保费 ”, …], wherein, the D 办理保额-预测保额 It refers to the relative distance between the insurance amount in the business handling keyword set and the predicted recommended keyword set, and the D 办理保费-预测保费 It is the relative distance between the premiums in the business handling keyword set and the predicted recommended keyword set.

[0088] Among them, the preset weight configuration rules are based on the importance of each type of data pre-defined by the enterprise, such as [σ 保费 , σ 保额 , ...], the σ 保费 is the premium weight coefficient, such as 32%, and the σ 保额 The weight coefficient of the insured amount, such as 29%.

[0089] In one embodiment of the present application, the similarity value S between the business handling keyword set and the predicted recommendation keyword set is obtained by the following formula: 相似度值 :

[0090] S 相似度值 =α*[σ 保额 D 办理保额 - 预测保额 +σ 保费 D 办理保费-预测保费 +…]

[0091] Wherein, α is the relationship function between the relative distance and the similarity value.

[0092] S4. Obtaining a hit rate of the price inquiry recommendation model based on the similarity value set and a preset standard threshold.

[0093] In detail, in the embodiment of the present application, obtaining the hit rate of the inquiry recommendation model based on the similarity value set and the preset standard threshold includes:

[0094] Determine whether the similarity value corresponding to each historical inquiry record in the similarity value set is greater than a preset standard threshold;

[0095] When the similarity value is greater than a preset standard threshold, it is determined that the inquiry recommendation model predicts and hits the historical inquiry record corresponding to the similarity value;

[0096] When the similarity value is less than or equal to the standard threshold, it is determined that the price inquiry recommendation model does not predict the historical price inquiry record corresponding to the similarity value;

[0097] The hit rate of the price inquiry recommendation model is obtained according to the hit results of each of the historical price inquiry records.

[0098] Specifically, in this embodiment of the present application, the standard threshold is set to 0.85. If the overall similarity value of various parameters such as premium, insured amount, and limit corresponding to a historical inquiry record is 0.9, which is greater than 0.85, then it is recorded as a hit of the inquiry recommendation model. After all historical inquiry records within the time period are compared, the ratio of the number of historical inquiry records that hit to the total number of historical inquiry records is calculated to obtain the hit rate of the inquiry recommendation model.

[0099] S5. The price inquiry recommendation model is trained according to the hit rate and the historical price inquiry records to obtain a standard price inquiry recommendation model.

[0100] It should be noted that the product recommendations obtained by the inquiry recommendation model may not be accepted by the user, and the final business processing results will often fluctuate. To ensure that the inquiry recommendation model meets user expectations as quickly as possible and improve business processing efficiency, the embodiment of this application requires regular updates to the inquiry recommendation model.

[0101] Specifically, in the embodiment of the present application, the training of the inquiry recommendation model based on the hit rate and the historical inquiry records within the time period includes:

[0102] Determining whether the hit rate is less than a preset training threshold;

[0103] When the hit rate is less than a preset training threshold, the inquiry recommendation model is trained using the historical inquiry records according to a K-fold cross-validation method to obtain a standard inquiry recommendation model;

[0104] When the hit rate is greater than or equal to the training threshold, the price inquiry recommendation model is determined to be a standard price inquiry recommendation model.

[0105] Specifically, the embodiment of the present application can set a training threshold to control the hit rate. When the various historical inquiry records of a certain day make the hit rate of the product recommendation plan predicted by the inquiry recommendation model too high, it indicates that the inquiry recommendation model cannot adapt to the changing trend of the sales market and needs to be updated. The embodiment of the present application trains the inquiry recommendation model in a K-fold cross-validation manner through the various historical inquiry records of the certain day to obtain an updated inquiry recommendation model. Among them, the K-fold cross-validation method uses a training method without repeated sampling technology to enhance the model training effect. Among them, the specific model training process is not described here.

[0106] In addition, in an embodiment of the present application, after obtaining the standard inquiry recommendation model, the salesperson can obtain a business recommendation plan of the updated inquiry recommendation model based on the inquiry keyword set such as the target information to be protected and the regional information provided by the user, thereby increasing business processing efficiency.

[0107] The embodiment of the present application obtains historical inquiry records in a regular and quantitative manner according to the time frequency and the time period, and ensures the accuracy of the inquiry recommendation model by checking the hit rate of the historical inquiry records with the pre-trained inquiry recommendation model, wherein the inquiry recommendation model can quickly generate a predicted recommendation keyword set based on the inquiry keyword set, thereby improving the efficiency of solution recommendation; in addition, since the historical inquiry records are generated by users through negotiation on the basis of the predicted recommendation keyword set, the embodiment of the present application continuously trains the inquiry recommendation model through the historical inquiry records, which can make the accuracy of the inquiry recommendation model gradually approach user expectations and improve user satisfaction. Therefore, the business recommendation model training method proposed in the embodiment of the present application can improve the accuracy of the business recommendation model.

[0108] like Figure 4 , which is a functional module diagram of a business recommendation model training device provided by an embodiment of the present invention.

[0109] The business recommendation model training device 100 of the present invention can be installed in an electronic device. Depending on the functions to be implemented, the business recommendation model training device 100 may include a historical data acquisition module 101, a model prediction module 102, a hit determination module 103, and a training determination module 104. The modules of the present invention, also referred to as units, refer to a series of computer program segments that can be executed by an electronic device processor and can perform fixed functions, and are stored in the memory of the electronic device.

[0110] In this embodiment, the functions of each module / unit are as follows:

[0111] The historical data acquisition module 101 is configured to acquire each historical inquiry record within a preset time period according to a preset time frequency, and extract data of each target field in each historical inquiry record according to a preset target field set to obtain an inquiry keyword set and a business processing keyword set corresponding to each historical inquiry record;

[0112] The model prediction module 102 is used to import the inquiry keyword set corresponding to each historical inquiry record into the pre-trained inquiry recommendation model to obtain the predicted recommendation keyword set corresponding to each historical inquiry record;

[0113] The hit judgment module 103 is used to calculate the similarity between the business processing keyword set corresponding to each historical inquiry record and the predicted recommendation keyword set according to the pre-established similarity matching rule to obtain a similarity value set, and obtain the hit rate of the inquiry recommendation model based on the similarity value set and a preset standard threshold;

[0114] The training judgment module 104 is configured to train the price inquiry recommendation model according to the hit rate and the various historical price inquiry records to obtain a standard price inquiry recommendation model.

[0115] In detail, each module described in the business recommendation model training device 100 in the embodiment of the present application adopts the same Figures 1 to 3 The same technical means as the business recommendation model training method described in and can produce the same technical effects are used, and will not be repeated here.

[0116] like Figure 5 , which is a structural diagram of an electronic device for implementing a service recommendation model training method provided by an embodiment of the present invention.

[0117] The electronic device 1 may include a processor 10, a memory 11, a communication bus 12, and a communication interface 13. It may also include a computer program stored in the memory 11 and executable on the processor 10, such as a business recommendation model training program.

[0118] In some embodiments, the processor 10 may be composed of an integrated circuit, for example, a single packaged integrated circuit, or a plurality of packaged integrated circuits with the same or different functions, including one or more central processing units (CPUs), microprocessors, digital processing chips, graphics processors, and a combination of various control chips. The processor 10 is the control core (Control Unit) of the electronic device, and uses various interfaces and lines to connect the various components of the entire electronic device. It runs or executes programs or modules stored in the memory 11 (for example, executing a business recommendation model training program, etc.), and calls data stored in the memory 11 to perform various functions of the electronic device and process data.

[0119] The memory 11 includes at least one type of readable storage medium, and the readable storage medium includes a flash memory, a mobile hard disk, a multimedia card, a card-type memory (for example, an SD or DX memory, etc.), a magnetic memory, a disk, an optical disk, etc. In some embodiments, the memory 11 may be an internal storage unit of an electronic device, such as a mobile hard disk of the electronic device. In other embodiments, the memory 11 may also be an external storage device of an electronic device, such as a plug-in mobile hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. equipped on the electronic device. Furthermore, the memory 11 may also include both an internal storage unit and an external storage device of the electronic device. The memory 11 can be used not only to store application software and various types of data installed in the electronic device, such as the code of the business recommendation model training program, but also to temporarily store data that has been output or is to be output.

[0120] The communication bus 12 may be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus. The bus may be divided into an address bus, a data bus, a control bus, etc. The bus is configured to enable communication between the memory 11 and at least one processor 10, etc.

[0121] The communication interface 13 is used for communication between the above-mentioned electronic device and other devices, including a network interface and a user interface. Optionally, the network interface may include a wired interface and / or a wireless interface (such as a WI-FI interface, a Bluetooth interface, etc.), which is generally used to establish a communication connection between the electronic device and other electronic devices. The user interface may be a display (Display), an input unit (such as a keyboard (Keyboard)), optionally, the user interface may also be a standard wired interface, a wireless interface. Optionally, in some embodiments, the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, and an OLED (Organic Light-Emitting Diode, organic light-emitting diode) touch device, etc. Among them, the display may also be appropriately referred to as a display screen or a display unit, which is used to display information processed in the electronic device and to display a visual user interface.

[0122] Figure 5 Only the electronic device with components is shown, and it can be understood by those skilled in the art that Figure 5The structure shown does not constitute a limitation on the electronic device 1 , and may include fewer or more components than shown in the figure, or combine certain components, or arrange the components differently.

[0123] For example, although not shown, the electronic device may further include a power source (such as a battery) for powering various components. Preferably, the power source may be logically connected to the at least one processor 10 via a power management device, thereby implementing functions such as charge management, discharge management, and power consumption management through the power management device. The power source may further include any components such as one or more DC or AC power sources, a recharging device, a power failure detection circuit, a power converter or inverter, a power status indicator, etc. The electronic device may further include various sensors, Bluetooth modules, Wi-Fi modules, etc., which will not be described in detail here.

[0124] It should be understood that the embodiment is for illustration only and the scope of the patent application is not limited to this structure.

[0125] The business recommendation model training program stored in the memory 11 of the electronic device 1 is a combination of multiple instructions. When running in the processor 10, it can achieve:

[0126] According to a preset time frequency, each historical inquiry record within a preset time period is obtained, and according to a preset target field set, data of each target field in each historical inquiry record is extracted to obtain an inquiry keyword set and a business processing keyword set corresponding to each historical inquiry record;

[0127] Importing the inquiry keyword set corresponding to each historical inquiry record into a pre-trained inquiry recommendation model to obtain a predicted recommended keyword set corresponding to each historical inquiry record;

[0128] Based on the pre-built similarity matching rules, similarity calculation is performed on the business processing keyword set corresponding to each historical inquiry record and the predicted recommendation keyword set to obtain a similarity value set;

[0129] Obtaining a hit rate of the price inquiry recommendation model based on the similarity value set and a preset standard threshold;

[0130] The price inquiry recommendation model is trained according to the hit rate and the various historical price inquiry records to obtain a standard price inquiry recommendation model.

[0131] Specifically, the specific implementation method of the processor 10 for the above instructions can refer to the description of the relevant steps in the corresponding embodiment of the accompanying drawings, which will not be repeated here.

[0132] Furthermore, if the modules / units integrated into the electronic device 1 are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. The computer-readable storage medium can be volatile or non-volatile. For example, the computer-readable medium can include: any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a mobile hard drive, a magnetic disk, an optical disk, a computer memory, or a read-only memory (ROM).

[0133] The present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program. When the computer program is executed by a processor of an electronic device, the computer program can implement:

[0134] According to a preset time frequency, each historical inquiry record within a preset time period is obtained, and according to a preset target field set, data of each target field in each historical inquiry record is extracted to obtain an inquiry keyword set and a business processing keyword set corresponding to each historical inquiry record;

[0135] Importing the inquiry keyword set corresponding to each historical inquiry record into a pre-trained inquiry recommendation model to obtain a predicted recommended keyword set corresponding to each historical inquiry record;

[0136] Based on the pre-built similarity matching rules, similarity calculation is performed on the business processing keyword set corresponding to each historical inquiry record and the predicted recommendation keyword set to obtain a similarity value set;

[0137] Obtaining a hit rate of the price inquiry recommendation model based on the similarity value set and a preset standard threshold;

[0138] The price inquiry recommendation model is trained according to the hit rate and the various historical price inquiry records to obtain a standard price inquiry recommendation model.

[0139] In the several embodiments provided by the present invention, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the module division is merely a logical function division, and other division methods may be used in actual implementation.

[0140] The modules described as separate components may or may not be physically separate, and the components shown as modules may or may not be physical units, that is, they may be located in one place or distributed across multiple network elements. Some or all of the modules may be selected to achieve the purpose of the solution of this embodiment according to actual needs.

[0141] In addition, the functional modules in various embodiments of the present invention may be integrated into a single processing unit, each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or hardware plus software functional modules.

[0142] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0143] Therefore, the embodiments should be considered in all respects as illustrative and non-restrictive, and the scope of the invention is defined by the appended claims rather than the foregoing description, and all changes that come within the meaning and range of equivalents of the claims are intended to be embraced therein. Any reference to a figure in a claim should not be construed as limiting the claim to which it relates.

[0144] Blockchain, as used in this article, refers to a novel application model for computer technologies such as distributed data storage, peer-to-peer transmission, consensus mechanisms, and encryption algorithms. Blockchain is essentially a decentralized database, a series of data blocks generated using cryptographic methods. Each block contains information about a batch of online transactions, used to verify the validity of this information (to prevent counterfeiting) and generate the next block. Blockchain can include the underlying blockchain platform, the platform product service layer, and the application service layer.

[0145] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Artificial Intelligence (AI) is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use knowledge to achieve optimal results.

[0146] Furthermore, it is clear that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or devices recited in a system claim may also be implemented by a single unit or device through software or hardware. Terms such as "first" and "second" are used to indicate names and do not imply any particular order.

[0147] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not limiting. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A business recommendation model training method, characterized in that: The method comprises: According to a preset time frequency, each historical inquiry record within a preset time period is obtained; according to each target field of a preset target field set, data corresponding to each target field is obtained from each historical inquiry record to obtain an initial keyword set corresponding to each historical inquiry record; according to the keyword category of each initial keyword in the initial keyword set, the initial keywords corresponding to each historical inquiry record are grouped to obtain an inquiry keyword set and a business processing keyword set corresponding to each historical inquiry record, wherein the target field set is a set of each field obtained by screening business records according to a principal component analysis method; Formatting the inquiry keyword set corresponding to each historical inquiry record using a preset program interactive interface to obtain input parameter data corresponding to each historical inquiry record, importing the input parameter data corresponding to each historical inquiry record into a pre-trained inquiry recommendation model, using the inquiry recommendation model to identify the main inquiry business in each input parameter data, querying the recommended solution corresponding to each main inquiry business, identifying the keyword of each recommended solution, and obtaining the predicted recommended keyword set corresponding to each historical inquiry record; Based on the pre-built similarity matching rules, similarity calculation is performed on the business processing keyword set corresponding to each historical inquiry record and the predicted recommendation keyword set to obtain a similarity value set; Obtaining a hit rate of the price inquiry recommendation model based on the similarity value set and a preset standard threshold; The price inquiry recommendation model is trained according to the hit rate and the various historical price inquiry records to obtain a standard price inquiry recommendation model.

2. The business recommendation model training method according to claim 1, characterized in that: The similarity calculation is performed on the business processing keyword set corresponding to each historical inquiry record and the predicted recommendation keyword set according to the pre-built similarity matching rule to obtain a similarity value set, including: Using a cosine similarity algorithm, the relative distance between the business processing keyword set corresponding to each historical inquiry record and the predicted recommendation keyword set is calculated to obtain a relative distance set corresponding to each historical inquiry record; According to a preset weight configuration rule, a weighted sum calculation is performed on each relative distance in the relative distance set corresponding to each historical inquiry record to obtain a similarity value set.

3. The business recommendation model training method according to claim 2, characterized in that: Obtaining the hit rate of the inquiry recommendation model according to the similarity value set and a preset standard threshold includes: Determine whether the similarity value corresponding to each historical inquiry record in the similarity value set is greater than a preset standard threshold; When the similarity value is greater than a preset standard threshold, it is determined that the inquiry recommendation model predicts and hits the historical inquiry record corresponding to the similarity value; When the similarity value is less than or equal to the standard threshold, it is determined that the price inquiry recommendation model does not predict the historical price inquiry record corresponding to the similarity value; The hit rate of the price inquiry recommendation model is obtained according to the hit results of each of the historical price inquiry records.

4. The business recommendation model training method according to claim 1, wherein: The training of the price inquiry recommendation model according to the hit rate and the historical price inquiry records to obtain a standard price inquiry recommendation model includes: Determining whether the hit rate is less than a preset training threshold; When the hit rate is less than a preset training threshold, the inquiry recommendation model is trained using the historical inquiry records according to a K-fold cross-validation method to obtain a standard inquiry recommendation model; When the hit rate is greater than or equal to the training threshold, the price inquiry recommendation model is determined to be a standard price inquiry recommendation model.

5. The business recommendation model training method according to claim 1, characterized in that: The step of obtaining each historical inquiry record within a preset time period according to a preset time frequency includes: Use preset tracking points to record a set of business records within a preset time period in a pre-built log space; Using pre-configured data retrieval permissions, automatically uploading the business record set to a pre-built training sample database according to a preset time frequency; The business record set is parsed using a pre-built parsing function in the training sample database to obtain various historical inquiry records.

6. A business recommendation model training device, characterized in that: The device comprises: A historical data acquisition module is configured to acquire, based on a preset time frequency, each historical inquiry record within a preset time period; acquire, based on each target field of a preset target field set, data corresponding to each target field from each historical inquiry record; obtain an initial keyword set corresponding to each historical inquiry record; group the initial keywords corresponding to each historical inquiry record according to the keyword category of each initial keyword in the initial keyword set; and obtain an inquiry keyword set and a business processing keyword set corresponding to each historical inquiry record; wherein the target field set is a set of fields obtained by screening business records according to a principal component analysis method; A model prediction module is used to format the inquiry keyword set corresponding to each historical inquiry record using a preset program interaction interface, obtain input parameter data corresponding to each historical inquiry record, import the input parameter data corresponding to each historical inquiry record into a pre-trained inquiry recommendation model, use the inquiry recommendation model to identify the main inquiry business in each input parameter data, and query the recommended solution corresponding to each main inquiry business, identify the keyword of each recommended solution, and obtain the predicted recommendation keyword set corresponding to each historical inquiry record; A hit judgment module is used to calculate the similarity between the business processing keyword set corresponding to each historical inquiry record and the predicted recommendation keyword set according to a pre-established similarity matching rule to obtain a similarity value set, and obtain the hit rate of the inquiry recommendation model based on the similarity value set and a preset standard threshold; The training judgment module is used to train the price inquiry recommendation model according to the hit rate and the various historical price inquiry records to obtain a standard price inquiry recommendation model.

7. An electronic device, characterized in that: The electronic device comprises: at least one processor; and, a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the business recommendation model training method according to any one of claims 1 to 5.

8. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the business recommendation model training method according to any one of claims 1 to 5 is implemented.

Images