Employee rating method, device and equipment based on business data and storage medium

Abstract

The invention discloses an employee rating method and device based on business data, equipment and a storage medium, relates to the field of Internet technology and finance, and provides more detailed and comprehensive mining dimensions for employee rating so as to improve rating accuracy. The method comprises the steps of obtaining to-be-analyzed marketing data and to-be-analyzed communication data generated by a to-be-rated employee in a preset time interval, and taking the to-be-analyzed marketing data and the to-be-analyzed communication data as to-be-analyzed business data; vectorizing the to-be-analyzed business data to obtain an initial input vector, and splitting the initial input vector into a plurality of sub-vectors by adopting a plurality of preset rating models, each sub-vector in the plurality of sub-vectors corresponding to a rating dimension; and determining a target input vector based on the plurality of sub-vectors, and scoring the target input vector by using a comprehensive scoring model to obtain a target score corresponding to the to-be-evaluated employee.

Description

Technical Field

[0001] The present application relates to the fields of Internet technology and finance, and in particular to a method, device, equipment and storage medium for employee rating based on business data. Background Art

[0002] With the continuous development of Internet technology and the financial field, more and more financial enterprises have emerged. Each financial enterprise will hire a large number of marketing personnel to contact customers in order to increase the company's turnover. However, different marketers have different marketing capabilities, and their marketing performance will affect customers' purchasing choices to a certain extent, thereby affecting the marketing performance of financial enterprises.

[0003] In related technologies, structured data such as marketers' performance achievements and marketing times are usually used for analysis to evaluate marketers' marketing performance, and marketers with the same marketing capabilities are divided into the same marketing group, and then different levels of customers are divided into different marketing groups.

[0004] In the process of implementing this application, the applicant found that the related technology has at least the following problems:

[0005] The existing method of rating marketers is to use marketing data such as performance achievement and marketing times to score, and then divide them according to the rating results. However, marketing data can only evaluate the marketing capabilities of marketers based on marketing results. Its evaluation dimensions are not detailed and comprehensive, which leads to inaccurate ratings of marketers. Summary of the invention

[0006] In view of this, the present application provides an employee rating method, device, equipment and storage medium based on business data. The main purpose is to solve the problem that the current existing marketing personnel rating method is to use marketing data such as performance achievement and marketing times to score, and then divide them according to the rating results. The marketing data can only evaluate the marketing ability of the marketer from the marketing results, and its evaluation dimension is not detailed and comprehensive enough, which leads to the problem of inaccurate marketing personnel rating.

[0007] According to a first aspect of the present application, a method for rating employees based on business data is provided, the method comprising:

[0008] Acquire the marketing data to be analyzed and the communication data to be analyzed generated by the employee to be rated within a preset time period, and use the marketing data to be analyzed and the communication data to be analyzed as the business data to be analyzed;

[0009] Performing vectorization processing on the business data to be analyzed to obtain an initial input vector, and using multiple preset rating models to split the initial input vector into multiple sub-vectors, each of the multiple sub-vectors corresponding to a rating dimension;

[0010] A target input vector is determined based on the multiple sub-vectors, and the target input vector is scored using a comprehensive scoring model to obtain a target score corresponding to the employee to be evaluated.

[0011] Optionally, the obtaining of the marketing data to be analyzed and the communication data to be analyzed generated by the employee to be rated within a preset time interval includes:

[0012] querying the preset time interval, and querying multiple to-be-processed marketing data and multiple to-be-processed communication data associated with the employee to be rated in all marketing data and all communication data;

[0013] Traversing the plurality of marketing data to be processed and the plurality of communication data to be processed, and determining a recording time of each marketing data to be processed in the plurality of marketing data to be processed and a recording time of each communication data to be processed in the plurality of communication data to be processed;

[0014] Comparing the recording time of each of the to-be-processed marketing data with the preset time interval one by one, and taking the to-be-processed marketing data whose recording time hits the preset time interval as the to-be-processed marketing data to be analyzed, wherein the to-be-analyzed marketing data includes marketing behavior data and marketing attribute data;

[0015] The recording time of each communication data to be processed is compared with the preset time interval one by one, and the communication data to be processed whose recording time hits the preset time interval is used as the communication data to be analyzed, and the communication data to be analyzed includes communication text data and communication audio data.

[0016] Optionally, the vectorizing the service data to be analyzed to obtain an initial input vector includes:

[0017] Discretizing the numerical feature data in the marketing data to be analyzed, and performing vector encoding on the discretized marketing data to be analyzed using a one-hot encoding technique to obtain a marketing data vector;

[0018] Obtaining a context-related representation vector of the communication data to be analyzed and performing average pooling to obtain the communication data vector;

[0019] The marketing data vector and the communication data vector are concatenated to obtain the initial input vector.

[0020] Optionally, the adopting of multiple preset rating models to split the initial input vector into multiple sub-vectors includes:

[0021] Inputting the initial input vectors into the plurality of preset rating models for deep learning respectively, wherein the plurality of preset rating models are respectively a professional rating module, a guidance rating module, an objection handling rating module, and an emotion management rating module;

[0022] Obtain a sub-vector extracted from the initial input vector by each preset rating model in the multiple preset rating models to obtain the multiple sub-vectors.

[0023] Optionally, after respectively inputting the initial input vectors into the plurality of preset rating models for deep learning, the method further comprises:

[0024] For each preset rating model among the multiple preset rating models, obtaining a dimension rating output by the preset rating model, and comparing the dimension rating with a preset display level;

[0025] When the comparison result indicates that the dimension rating is greater than or equal to the preset display level, the display text extracted by the preset rating model is obtained for display.

[0026] Optionally, determining a target input vector based on the multiple sub-vectors comprises:

[0027] Determining an attention weight of each of the multiple sub-vectors based on the attention network, and calculating a vector product of each of the sub-vectors and the corresponding attention weight to obtain a plurality of vector products;

[0028] The sum of the products of the multiple vectors is used as the target input vector.

[0029] Optionally, the adopting a comprehensive scoring model to score the target input vector to obtain a target score corresponding to the employee to be evaluated includes:

[0030] Acquire multiple historical target input vectors corresponding to multiple historical business data and multiple historical scores corresponding to the multiple historical business data, set the multiple historical model target data vectors as input data and the multiple historical scores as output data for model training, and obtain the comprehensive scoring model;

[0031] The target input vector is input into the comprehensive scoring model, and the comprehensive scoring model is used to predict the score of the target input vector to obtain the target score.

[0032] According to a second aspect of the present application, a device for rating employees based on business data is provided, the device comprising:

[0033] An acquisition module, used to acquire the marketing data to be analyzed and the communication data to be analyzed generated by the employee to be rated within a preset time interval, and use the marketing data to be analyzed and the communication data to be analyzed as the business data to be analyzed;

[0034] A splitting module, used to perform vectorization processing on the business data to be analyzed to obtain an initial input vector, and use multiple preset rating models to split the initial input vector into multiple sub-vectors, each of the multiple sub-vectors corresponding to a rating dimension;

[0035] The rating module is used to determine a target input vector based on the multiple sub-vectors, and to score the target input vector using a comprehensive scoring model to obtain a target score corresponding to the employee to be evaluated.

[0036] Optionally, the acquisition module is used to query the preset time interval, and query multiple unprocessed marketing data and multiple unprocessed communication data associated with the employee to be rated in all marketing data and all communication data; traverse the multiple unprocessed marketing data and the multiple unprocessed communication data to determine the recording time of each unprocessed marketing data in the multiple unprocessed marketing data and the recording time of each unprocessed communication data in the multiple unprocessed communication data; compare the recording time of each unprocessed marketing data with the preset time interval one by one, and take the unprocessed marketing data whose recording time hits the preset time interval as the marketing data to be analyzed, and the marketing data to be analyzed includes marketing behavior data and marketing attribute data; compare the recording time of each unprocessed communication data with the preset time interval one by one, and take the unprocessed communication data whose recording time hits the preset time interval as the communication data to be analyzed, and the communication data to be analyzed includes communication text data and communication audio data.

[0037] Optionally, the splitting module is used to discretize the numerical feature data in the marketing data to be analyzed, and use the one-hot encoding technology to vector encode the discretized marketing data to be analyzed to obtain a marketing data vector; obtain the context-related representation vector of the communication data to be analyzed and perform average pooling to obtain the communication data vector; splice the marketing data vector and the communication data vector to obtain the initial input vector.

[0038] Optionally, the splitting module is used to input the initial input vector into the multiple preset rating models for deep learning, respectively, and the multiple preset rating models are a professional rating module, a guidance rating module, an objection handling rating module, and an emotion management rating module; obtain the sub-vector extracted from the initial input vector by each preset rating model in the multiple preset rating models to obtain the multiple sub-vectors.

[0039] Optionally, the splitting module is also used to obtain, for each preset rating model among the multiple preset rating models, a dimension rating output by the preset rating model, and compare the dimension rating with a preset display level; when the comparison result indicates that the dimension rating is greater than or equal to the preset display level, obtain the display text extracted by the preset rating model for display.

[0040] Optionally, the rating module is used to determine the attention weight of each sub-vector in the multiple sub-vectors based on the attention network, and calculate the vector product of each sub-vector and the corresponding attention weight to obtain multiple vector products; and use the sum of the multiple vector products as the target input vector.

[0041] Optionally, the rating module is used to obtain multiple historical target input vectors corresponding to multiple historical business data and multiple historical scores corresponding to the multiple historical business data, set the multiple historical model target data vectors as input data and the multiple historical scores as output data for model training to obtain the comprehensive scoring model; input the target input vector into the comprehensive scoring model, use the comprehensive scoring model to predict the score of the target input vector, and obtain the target score.

[0042] According to a third aspect of the present application, a computer device is provided, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of any one of the methods described in the first aspect when executing the computer program.

[0043] According to a fourth aspect of the present application, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of any one of the methods described in the first aspect are implemented.

[0044] By means of the above technical scheme, the present application provides an employee rating method, device, equipment and storage medium based on business data. The present application first obtains the marketing data to be analyzed and the communication data to be analyzed generated by the employee to be rated within a preset time interval, and uses the marketing data to be analyzed and the communication data to be analyzed as the business data to be analyzed. Further, the business data to be analyzed is vectorized to obtain an initial input vector, and multiple preset rating models are used to split the initial input vector into multiple sub-vectors. The target input vector is then determined based on multiple sub-vectors. Finally, a comprehensive scoring model is used to score the target input vector to obtain the target score corresponding to the employee to be evaluated. The embodiment of the present application adds unstructured data such as communication data to be analyzed that can reflect marketing performance, which provides a more detailed and comprehensive mining dimension for employee ratings. It combines structured data and unstructured data that can reflect marketing methods, and can comprehensively evaluate the level of employees, thereby improving the rating accuracy.

[0045] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Various other advantages and benefits will become apparent to those of ordinary skill in the art by reading the detailed description of the preferred embodiments below. The accompanying drawings are only for the purpose of illustrating the preferred embodiments and are not to be considered as limiting the present application. Also, the same reference symbols are used throughout the accompanying drawings to represent the same components. In the accompanying drawings:

[0047] Figure 1 A schematic diagram of a method for rating employees based on business data provided in an embodiment of the present application is shown;

[0048] Figure 2A A schematic diagram of a method for rating employees based on business data provided in an embodiment of the present application is shown;

[0049] Figure 2B A schematic diagram of a method for rating employees based on business data provided in an embodiment of the present application is shown;

[0050] Figure 3 A schematic diagram of the structure of an employee rating device based on business data provided in an embodiment of the present application is shown;

[0051] Figure 4 A schematic diagram of the device structure of a computer device provided in an embodiment of the present application is shown. DETAILED DESCRIPTION

[0052] The exemplary embodiments of the present application will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present application are shown in the accompanying drawings, it should be understood that the present application can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided in order to enable a more thorough understanding of the present application and to fully convey the scope of the present application to those skilled in the art.

[0053] The present application embodiment provides an employee rating method based on business data, such as Figure 1 As shown, the method includes:

[0054] 101. Obtain the marketing data to be analyzed and the communication data to be analyzed generated by the employee to be rated within a preset time period, and use the marketing data to be analyzed and the communication data to be analyzed as the business data to be analyzed.

[0055] With the continuous development of Internet technology and the financial field, more and more financial enterprises have emerged, and each financial enterprise will hire a large number of marketing personnel to contact customers in order to increase the company's turnover. However, different marketers have different marketing capabilities, and their marketing performance will affect the customer's purchase choice to a certain extent, thereby affecting the marketing performance of financial enterprises. At present, it is usually necessary to use structured data such as the performance achievement of marketers and the number of marketing times to analyze and evaluate the marketing performance of marketers, and divide marketers with the same marketing capabilities into the same marketing group, and then divide different levels of customers into different marketing groups. However, the applicant recognizes that the existing marketing personnel rating method is to use marketing data such as performance achievement and the number of marketing times for scoring, and then divide them according to the rating results, while marketing data can only evaluate the marketing ability of marketers from the marketing results, and its evaluation dimension is not detailed and comprehensive enough, which leads to inaccurate rating of marketers. Therefore, the present application provides an employee rating method based on business data, first obtaining the marketing data to be analyzed and the communication data to be analyzed generated by the employee to be rated within a preset time interval, and using the marketing data to be analyzed and the communication data to be analyzed as the business data to be analyzed. Furthermore, the business data to be analyzed is vectorized to obtain an initial input vector, and multiple preset rating models are used to split the initial input vector into multiple sub-vectors. The target input vector is then determined based on the multiple sub-vectors. Finally, the target input vector is scored using a comprehensive scoring model to obtain a target score corresponding to the employee to be evaluated. The embodiment of the present application adds unstructured data such as communication data to be analyzed that can reflect marketing performance, which provides a more detailed and comprehensive mining dimension for employee ratings. It combines structured data that can reflect marketing methods with unstructured data, and can comprehensively evaluate the level of employees, thereby improving the accuracy of ratings.

[0056] The embodiment of the present application is applicable to a rating system. The rating system can obtain the marketing data to be analyzed and the communication data to be analyzed generated by the employee to be rated within a preset time interval in the database in response to a rating request, and use the marketing data to be analyzed and the communication data to be analyzed as the business data to be analyzed. Among them, the marketing data to be analyzed is structured data, which can reflect the marketing methods of employees, specifically including but not limited to marketing behavior data (such as the number of target customers, marketing frequency, and total duration), basic attribute data (such as seat age, years of service), and team attributes (such as team size and historical average performance). The communication data to be analyzed is unstructured data, which can reflect the marketing performance of employees in the marketing process, specifically including communication text data and communication audio data. Then, the marketing data to be analyzed and the communication data to be analyzed are used as business data to be analyzed. By combining structured data that can reflect marketing methods and unstructured data that can reflect marketing performance, the level of employees can be comprehensively evaluated, thereby improving the accuracy of rating.

[0057] 102. The business data to be analyzed is vectorized to obtain an initial input vector, and multiple preset rating models are used to split the initial input vector into multiple sub-vectors, each of the multiple sub-vectors corresponding to a rating dimension.

[0058] In an embodiment of the present application, the rating system vectorizes the acquired business data to be analyzed to obtain an initial input vector. The initial input vector needs to be input into multiple preset rating models for rating in different dimensions and extracting sub-vectors in different dimensions. The multiple preset rating models include a professional rating module, a guidance rating module, an objection handling rating module, and an emotion management rating module.

[0059] 103. Determine a target input vector based on multiple sub-vectors, and score the target input vector using a comprehensive scoring model to obtain a target score corresponding to the employee to be evaluated.

[0060] In this step, the rating system determines the vector weight of each sub-vector based on the attention network, and performs weighted summation of multiple sub-vectors according to the vector weight of each sub-vector to obtain the target input vector. Then, the pre-built comprehensive rating model is used to perform a comprehensive rating on the target input vector to obtain the target score of the employee to be evaluated. Finally, the employee to be evaluated can be assigned to a matching marketing group based on the target score, or employees with higher target scores can be rewarded.

[0061] The employee rating method based on business data provided in this embodiment first obtains the marketing data to be analyzed and the communication data to be analyzed generated by the employee to be rated within a preset time interval, and uses the marketing data to be analyzed and the communication data to be analyzed as the business data to be analyzed. Further, the business data to be analyzed is vectorized to obtain an initial input vector, and multiple preset rating models are used to split the initial input vector into multiple sub-vectors. The target input vector is then determined based on the multiple sub-vectors. Finally, a comprehensive scoring model is used to score the target input vector to obtain the target score corresponding to the employee to be evaluated. The embodiment of the present application adds unstructured data such as communication data to be analyzed that can reflect marketing performance, which provides a more detailed and comprehensive mining dimension for employee ratings. It combines structured data and unstructured data that can reflect marketing methods, and can comprehensively evaluate the level of employees, thereby improving the rating accuracy.

[0062] Further, as a refinement and extension of the specific implementation of the above embodiment, in order to fully illustrate the implementation process of this embodiment, the embodiment of the present application provides an employee rating method based on business data, such as Figure 2A As shown, the method includes:

[0063] 201. Query a preset time interval, and query multiple unprocessed marketing data and multiple unprocessed communication data associated with the employee to be rated in all marketing data and all communication data.

[0064] In the embodiment of the present application, the administrator can set one week, one month or three months as the preset time interval according to the actual rating requirements. When the rating system detects the rating request, it determines the employee to be rated and searches for multiple unprocessed marketing data and multiple unprocessed communication data associated with the employee to be rated in all marketing data and all communication data stored in the database.

[0065] 202. Extract the marketing data to be analyzed and the communication data to be analyzed from the plurality of marketing data to be processed and the plurality of communication data to be processed according to a preset time interval, and use the marketing data to be analyzed and the communication data to be analyzed as the business data to be analyzed.

[0066] Furthermore, the rating system traverses multiple pending marketing data to determine the recording time of each pending marketing data, and then compares the recording time of each pending marketing data with the preset time interval one by one, and takes the pending marketing data whose recording time hits the preset time interval as the marketing data to be analyzed. It should be noted that the marketing data is structured data that can reflect the marketing methods of employees, specifically including but not limited to marketing behavior data (such as the number of target customers, marketing frequency, and total duration), basic attribute data (such as seat age, years of service), and team attributes (such as team size and historical average performance). At the same time, the rating system also traverses multiple pending communication data to determine the recording time of each pending communication data in the multiple pending communication data. The recording time of each pending communication data is compared with the preset time interval one by one, and the pending communication data whose recording time hits the preset time interval is taken as the communication data to be analyzed. It should be noted that the communication data is unstructured data that can reflect the marketing performance of employees in the marketing process, specifically including communication text data and communication audio data.

[0067] It should be noted that when the preset time interval is set to be long, the amount of marketing data to be processed and communication data to be processed is large. In order to speed up the rating, the rating system can also randomly extract part of the marketing data to be analyzed and part of the communication data to be analyzed from multiple marketing data to be processed and multiple communication data to be processed as the business data to be analyzed according to the overall distribution.

[0068] 203. Perform vectorization processing on the business data to be analyzed to obtain an initial input vector.

[0069] In the embodiment of the present application, the rating system needs to process the business data to be analyzed, that is, the structured data and unstructured data, into an initial input vector in a standard format that can be directly used by subsequent preset rating models and comprehensive scoring models.

[0070] Specifically, for structured marketing data to be analyzed, the rating system first performs data cleaning on it to fill in the missing data values ​​in the marketing data to be analyzed and delete the data outliers in the marketing data to be analyzed. Further, the numerical feature data in the cleaned marketing data to be analyzed is first discretized, and then the one-hot encoding technology is used to vector encode the discretized marketing data to be analyzed to obtain the marketing data vector x_stru. For unstructured communication data to be analyzed, the rating system first performs data cleaning on it to perform operations such as role text distinction (marking customer text and agent text), text cleaning (removing modal particles, useless punctuation, expressions, etc.). It should be noted that since the text of the financial industry is different from the text of daily communication, it will contain professional terms such as major financial industry products and processes. Therefore, the embodiment of the present application is based on millions of financial industry dialogue texts to train a pre-trained model for a specific field, and then the context-related representation vector of the communication data to be analyzed is obtained through the pre-trained model for average pooling to obtain the communication data vector x_unstru. It should be noted that the pre-trained model can be a bag of words vector model or a BERT model. In addition, when the communication data to be analyzed is communication audio data, it is necessary to convert the audio into text to generate the corresponding communication text data, and then use the pre-trained model to extract features and vectorize the communication text data to obtain the communication data vector x_unstru. Finally, the marketing data vector x_stru and the communication data vector x_unstru are concatenated to obtain the initial input vector X_input. X_input = concate(x_stru, x_unstru), where X_input, x_stru, and x_unstru are all vector matrices.

[0071] 204. Using multiple preset rating models, the initial input vector is split into multiple sub-vectors.

[0072] In this step, in order to achieve multi-dimensional and objective comprehensive rating of employees, this application is equipped with a preset rating module with multiple dimensions, which can split the initial input vector into multiple sub-vectors, each sub-vector corresponding to a rating dimension. This provides a more detailed and comprehensive mining dimension for the subsequent comprehensive rating model, and improves the rating accuracy of the comprehensive rating model.

[0073] Specifically, the initial input vector X_input is input into multiple preset rating models for deep learning, and the multiple preset rating models are respectively a professional rating module, a guidance rating module, an objection handling rating module, and an emotion management rating module. The rating system can obtain the sub-scores of the corresponding dimensions output by each preset rating model and the sub-vectors of the corresponding dimensions. It should be noted that there are multiple options for these four types of model structures, which can be simple single machine learning (lightgbm, xgboost, random forest, etc.) models, multi-layer fusion machine learning (self-designed multi-layer factor fusion, etc.) models, or single deep learning (LSTM, GRU, bi-GRU, etc.) models. The final model result selection mainly selects the most suitable model structure based on the accuracy and efficiency required by actual production, so that it can be flexibly applied to the actual production link. In addition, for each preset rating model in the multiple preset rating models, the dimension rating output by the preset rating model is obtained, and the dimension rating is compared with the preset display level. When the comparison result indicates that the dimension rating is greater than or equal to the preset display level, the display text extracted by the preset rating model is obtained for display for other employees to learn.

[0074] 205. Determine a target input vector based on the multiple sub-vectors.

[0075] In the embodiment of the present application, the weight distribution of this step can be equal to 0.25, but considering the different situations of each seat, if the computing resources are sufficient, the attention network can be used to determine the attention weight a of each sub-vector in the multiple sub-vectors. 专业度 、a 指引度 、a 异议处理度 、a 情绪管理度 , where a 专业度 +a 指引度 +a 异议处理度 +a 情绪管理度 = 1. After weight allocation, the vector product of each sub-vector and the corresponding attention weight can be weighted according to the following formula 1 to obtain multiple vector products, and finally the sum of the multiple vector products is used as the target input vector.

[0076] Formula 1: x final =a 专业度 ·x 专业度 +a 指引度 ·x 指引度 +a 异议处理度 ·x 异议处理度 +a 情绪管理度 ·x 情绪管理度

[0077] Among them, x final is the target input vector; x专业度 、x 指引度 、x 异议处理度 、x 情绪管理度 They are the sub-vectors input to the professionalism rating module, guidance rating module, objection handling rating module, and emotion management rating module respectively.

[0078] 206. Use the comprehensive scoring model to score the target input vector and obtain the target score corresponding to the employee to be evaluated.

[0079] In actual operation, relevant staff will pre-build a comprehensive scoring model. Specifically, they will obtain multiple historical target input vectors corresponding to multiple historical business data and multiple historical scores corresponding to multiple historical business data, set multiple historical model target data vectors as input data of the multi-layer perceptual vector machine MLP network, and multiple historical scores as output data of the multi-layer perceptual vector machine MLP network, and train the multi-layer perceptual vector machine MLP network. When the model scoring accuracy meets the preset conditions, a comprehensive scoring model will be put into use. The scoring system substitutes the final target input vector into the comprehensive scoring model to obtain the final employee marketing ability score, which can range from 0 to 100 points.

[0080] In summary, if Figure 2B As shown in the figure, the employee rating process is divided into five stages. The first stage is the data preparation stage. In this stage, the rating system will access the database to query the communication data (unstructured data) and marketing data (structured data) to be analyzed of the employees to be rated, and use the communication data and marketing data to be analyzed as the business data to be analyzed. The second stage is the data preprocessing stage. In this stage, the rating system vectorizes the business data to be analyzed to obtain the initial input vector. The third stage is the sub-module rating stage. In this stage, the initial input vector will be input into multiple preset rating models to split the initial input vector into multiple sub-vectors. The fourth stage is the weight allocation stage. Multiple sub-vectors are input into the weight learning module, and the vector weight of each sub-vector is determined based on the attention network. Then, the multiple sub-vectors are weighted and summed to obtain the target input vector. The fifth stage is the rating stage. In this stage, the comprehensive scoring model is used to score the target input vector, and finally the target score corresponding to the employee to be evaluated is obtained.

[0081] The method provided in the embodiment of the present application first obtains the marketing data to be analyzed and the communication data to be analyzed generated by the employee to be rated within a preset time interval, and uses the marketing data to be analyzed and the communication data to be analyzed as the business data to be analyzed. Furthermore, the business data to be analyzed is vectorized to obtain an initial input vector, and multiple preset rating models are used to split the initial input vector into multiple sub-vectors. The target input vector is then determined based on the multiple sub-vectors. Finally, the target input vector is scored using a comprehensive scoring model to obtain the target score corresponding to the employee to be evaluated. The embodiment of the present application adds unstructured data such as communication data to be analyzed that can reflect marketing performance, which provides a more detailed and comprehensive mining dimension for employee ratings. It combines structured data and unstructured data that can reflect marketing methods, and can comprehensively evaluate the level of employees, thereby improving the rating accuracy.

[0082] Further, as Figure 1 In a specific implementation of the method, the present application embodiment provides an employee rating device based on business data, such as Figure 3 As shown, the device includes: an acquisition module 301, a splitting module 302, and a rating module 303.

[0083] The acquisition module 301 is used to acquire the marketing data to be analyzed and the communication data to be analyzed generated by the employee to be rated within a preset time interval, and use the marketing data to be analyzed and the communication data to be analyzed as the business data to be analyzed;

[0084] The splitting module 302 is used to perform vectorization processing on the business data to be analyzed to obtain an initial input vector, and use multiple preset rating models to split the initial input vector into multiple sub-vectors, each of the multiple sub-vectors corresponding to a rating dimension;

[0085] The rating module 303 is used to determine a target input vector based on the multiple sub-vectors, and to score the target input vector using a comprehensive scoring model to obtain a target score corresponding to the employee to be evaluated.

[0086] In a specific application scenario, the acquisition module 301 is used to query the preset time interval, and to query multiple unprocessed marketing data and multiple unprocessed communication data associated with the employee to be rated in all marketing data and all communication data; traverse the multiple unprocessed marketing data and the multiple unprocessed communication data, and determine the recording time of each unprocessed marketing data in the multiple unprocessed marketing data and the recording time of each unprocessed communication data in the multiple unprocessed communication data; compare the recording time of each unprocessed marketing data with the preset time interval one by one, and use the unprocessed marketing data whose recording time hits the preset time interval as the marketing data to be analyzed, and the marketing data to be analyzed includes marketing behavior data and marketing attribute data; compare the recording time of each unprocessed communication data with the preset time interval one by one, and use the unprocessed communication data whose recording time hits the preset time interval as the communication data to be analyzed, and the communication data to be analyzed includes communication text data and communication audio data.

[0087] In a specific application scenario, the splitting module 302 is used to discretize the numerical feature data in the marketing data to be analyzed, and use the one-hot encoding technology to vector encode the discretized marketing data to be analyzed to obtain a marketing data vector; obtain the context-related representation vector of the communication data to be analyzed and perform average pooling to obtain the communication data vector; splice the marketing data vector and the communication data vector to obtain the initial input vector.

[0088] In a specific application scenario, the splitting module 302 is used to input the initial input vector into the multiple preset rating models for deep learning, and the multiple preset rating models are respectively a professional rating module, a guidance rating module, an objection handling rating module, and an emotion management rating module; obtain the sub-vector extracted by each preset rating model in the multiple preset rating models from the initial input vector to obtain the multiple sub-vectors.

[0089] In a specific application scenario, the splitting module 302 is also used to obtain the dimension rating output by each preset rating model among the multiple preset rating models, and compare the dimension rating with the preset display level; when the comparison result indicates that the dimension rating is greater than or equal to the preset display level, obtain the display text extracted by the preset rating model for display.

[0090] In a specific application scenario, the rating module 303 is used to determine the attention weight of each sub-vector in the multiple sub-vectors based on the attention network, and calculate the vector product of each sub-vector and the corresponding attention weight to obtain multiple vector products; and use the sum of the multiple vector products as the target input vector.

[0091] In a specific application scenario, the rating module 303 is used to obtain multiple historical target input vectors corresponding to multiple historical business data and multiple historical scores corresponding to the multiple historical business data, set the multiple historical model target data vectors as input data, and the multiple historical scores as output data for model training to obtain the comprehensive scoring model; input the target input vector into the comprehensive scoring model, use the comprehensive scoring model to predict the score of the target input vector, and obtain the target score.

[0092] The device provided in the embodiment of the present application first obtains the marketing data to be analyzed and the communication data to be analyzed generated by the employee to be rated within a preset time interval, and uses the marketing data to be analyzed and the communication data to be analyzed as the business data to be analyzed. Further, the business data to be analyzed is vectorized to obtain an initial input vector, and multiple preset rating models are used to split the initial input vector into multiple sub-vectors. The target input vector is then determined based on the multiple sub-vectors. Finally, the target input vector is scored using a comprehensive scoring model to obtain the target score corresponding to the employee to be evaluated. The embodiment of the present application adds unstructured data such as communication data to be analyzed that can reflect marketing performance, which provides a more detailed and comprehensive mining dimension for employee ratings. It combines structured data and unstructured data that can reflect marketing methods, and can comprehensively evaluate the level of employees, thereby improving the accuracy of ratings.

[0093] It should be noted that for other corresponding descriptions of the functional units involved in the employee rating device based on business data provided in the embodiment of the present application, reference can be made to Figure 1 and FIG. 2A to FIG. 2B The corresponding description in will not be repeated here.

[0094] Based on the above Figure 1 , FIG. 2A to FIG. 2B The method shown, accordingly, this embodiment also provides a storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the employee rating method based on business data are implemented.

[0095] Based on this understanding, the technical solution of the present application can be embodied in the form of a software product. The software product to be identified can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.), including a number of instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each implementation scenario of the present application.

[0096] Based on the above Figure 1 , FIG. 2A to FIG. 2B The method shown, and Figure 3In order to achieve the above-mentioned purpose, in the exemplary embodiment, see Figure 4 , and also provides a device, which includes a communication bus, a processor, a memory and a communication interface, and may also include an input and output interface and a display device, wherein each functional unit can communicate with each other through the bus. The memory stores a computer program, and the processor is used to execute the program stored in the memory and execute the employee rating method based on business data in the above embodiment.

[0097] Optionally, the physical device may also include a user interface, a network interface, a camera, a radio frequency (RF) circuit, a sensor, an audio circuit, a WI-FI module, etc. The user interface may include a display, an input unit such as a keyboard, etc., and the optional user interface may also include a USB interface, a card reader interface, etc. The network interface may optionally include a standard wired interface, a wireless interface (such as a WI-FI interface), etc.

[0098] Those skilled in the art will appreciate that the physical device structure for employee rating based on business data provided in this embodiment does not constitute a limitation on the physical device, and may include more or fewer components, or a combination of certain components, or different arrangements of components.

[0099] The storage medium may also include an operating system and a network communication module. The operating system is a program that manages the hardware of the physical device and the software resources to be identified, and supports the operation of the information processing program and other software and / or programs to be identified. The network communication module is used to realize the communication between the components inside the storage medium, and the communication with other hardware and software in the information processing physical device.

[0100] Through the description of the above implementation methods, those skilled in the art can clearly understand that the present application can be implemented by means of software plus necessary general hardware platforms, or by hardware. By applying the technical solution of the present application, the marketing data to be analyzed and the communication data to be analyzed generated by the employee to be rated within a preset time interval are first obtained, and the marketing data to be analyzed and the communication data to be analyzed are used as the business data to be analyzed. Further, the business data to be analyzed is vectorized to obtain an initial input vector, and multiple preset rating models are used to split the initial input vector into multiple sub-vectors. The target input vector is then determined based on multiple sub-vectors. Finally, the target input vector is scored using a comprehensive scoring model to obtain the target score corresponding to the employee to be evaluated. Compared with the prior art, the embodiment of the present application adds unstructured data such as communication data to be analyzed that can reflect marketing performance, which provides a more detailed and comprehensive mining dimension for the rating of employees, combines structured data and unstructured data that can reflect marketing methods, and can comprehensively evaluate the level of employees, thereby improving the rating accuracy.

[0101] Those skilled in the art will appreciate that the accompanying drawings are merely schematic diagrams of a preferred implementation scenario, and the modules or processes in the accompanying drawings are not necessarily required for implementing the present application.

[0102] Those skilled in the art will appreciate that the modules in the device in the implementation scenario can be distributed in the device in the implementation scenario according to the implementation scenario description, or can be changed accordingly and located in one or more devices different from the implementation scenario. The modules in the above implementation scenario can be combined into one module, or can be further split into multiple sub-modules.

[0103] The above application serial numbers are for description only and do not represent the advantages or disadvantages of the implementation scenarios.

[0104] The above disclosure only discloses several specific implementation scenarios of the present application. However, the present application is not limited thereto, and any changes that can be conceived by technicians in this field should fall within the scope of protection of the present application.

Claims

1. A method for employee rating of business data, characterized in that: include: Acquire the marketing data to be analyzed and the communication data to be analyzed generated by the employee to be rated within a preset time period, and use the marketing data to be analyzed and the communication data to be analyzed as the business data to be analyzed; Vectorizing the business data to be analyzed to obtain an initial input vector, and using multiple preset rating models to split the initial input vector into multiple sub-vectors, each of the multiple sub-vectors corresponding to a rating dimension; A target input vector is determined based on the multiple sub-vectors, and the target input vector is scored using a comprehensive scoring model to obtain a target score corresponding to the employee to be evaluated.

2. The method according to claim 1, characterized in that The obtaining of the marketing data to be analyzed and the communication data to be analyzed generated by the employee to be rated within a preset time period includes: querying the preset time interval, and querying multiple to-be-processed marketing data and multiple to-be-processed communication data associated with the employee to be rated in all marketing data and all communication data; Traversing the plurality of marketing data to be processed and the plurality of communication data to be processed, and determining a recording time of each marketing data to be processed in the plurality of marketing data to be processed and a recording time of each communication data to be processed in the plurality of communication data to be processed; Comparing the recording time of each of the to-be-processed marketing data with the preset time interval one by one, and taking the to-be-processed marketing data whose recording time hits the preset time interval as the to-be-processed marketing data to be analyzed, wherein the to-be-analyzed marketing data includes marketing behavior data and marketing attribute data; The recording time of each communication data to be processed is compared with the preset time interval one by one, and the communication data to be processed whose recording time hits the preset time interval is used as the communication data to be analyzed, and the communication data to be analyzed includes communication text data and communication audio data.

3. The method according to claim 1, characterized in that The vectorization processing of the service data to be analyzed to obtain an initial input vector includes: Discretizing the numerical feature data in the marketing data to be analyzed, and performing vector encoding on the discretized marketing data to be analyzed using a one-hot encoding technique to obtain a marketing data vector; Obtaining a context-related representation vector of the communication data to be analyzed and performing average pooling to obtain the communication data vector; The marketing data vector and the communication data vector are concatenated to obtain the initial input vector.

4. The method according to claim 1, characterized in that: The method uses multiple preset rating models to split the initial input vector into multiple sub-vectors, including: Inputting the initial input vectors into the plurality of preset rating models for deep learning respectively, wherein the plurality of preset rating models are respectively a professional rating module, a guidance rating module, an objection handling rating module, and an emotion management rating module; Obtain a sub-vector extracted from the initial input vector by each preset rating model in the multiple preset rating models to obtain the multiple sub-vectors.

5. The method according to claim 4, characterized in that After the initial input vectors are respectively input into the plurality of preset rating models for deep learning, the method further includes: For each preset rating model of the plurality of preset rating models, obtaining a dimension rating output by the preset rating model, and comparing the dimension rating with a preset display level; When the comparison result indicates that the dimension rating is greater than or equal to the preset display level, the display text extracted by the preset rating model is obtained for display.

6. The method according to claim 1, characterized in that The determining a target input vector based on the multiple sub-vectors comprises: Determining an attention weight of each of the multiple sub-vectors based on the attention network, and calculating a vector product of each of the sub-vectors and the corresponding attention weight to obtain a plurality of vector products; The sum of the products of the multiple vectors is used as the target input vector.

7. The method according to claim 1, characterized in that The step of using a comprehensive scoring model to score the target input vector to obtain a target score corresponding to the employee to be evaluated includes: Acquire multiple historical target input vectors corresponding to multiple historical business data and multiple historical scores corresponding to the multiple historical business data, set the multiple historical model target data vectors as input data and the multiple historical scores as output data for model training, and obtain the comprehensive scoring model; The target input vector is input into the comprehensive scoring model, and the comprehensive scoring model is used to predict the score of the target input vector to obtain the target score.

8. A method for employee rating of business data, characterized in that: include: An acquisition module, used to acquire the marketing data to be analyzed and the communication data to be analyzed generated by the employee to be rated within a preset time interval, and use the marketing data to be analyzed and the communication data to be analyzed as the business data to be analyzed; A splitting module, used to perform vectorization processing on the business data to be analyzed to obtain an initial input vector, and use multiple preset rating models to split the initial input vector into multiple sub-vectors, each of the multiple sub-vectors corresponding to a rating dimension; The rating module is used to determine a target input vector based on the multiple sub-vectors, and to score the target input vector using a comprehensive scoring model to obtain a target score corresponding to the employee to be evaluated.

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

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

Images