CN113052508B

CN113052508B - Data processing method and device

Info

Publication number: CN113052508B
Application number: CN202110486343.3A
Authority: CN
Inventors: 宋雨; 程璐; 杨晓明; 赵辉
Original assignee: Bank of China Ltd
Current assignee: Bank of China Ltd
Priority date: 2021-04-30
Filing date: 2021-04-30
Publication date: 2024-02-02
Anticipated expiration: 2041-04-30
Also published as: CN113052508A

Abstract

The invention discloses a data processing method and a data processing device, wherein a proficiency evaluation model can be used for obtaining the proficiency evaluation score of a teller for a target to-be-evaluated service, a whole waiting time evaluation model can be used for obtaining the whole waiting time evaluation score required by a customer when the target to-be-evaluated service is handled, the sum of the reciprocal of the proficiency evaluation score and the whole waiting time evaluation score is determined to be the final effective score for evaluating the target to-be-evaluated service, and the final effective score can quantitatively evaluate the improvement amplitude of the net point effectiveness of the target to-be-evaluated service when the target to-be-evaluated service is used as a learning object of a net point robot, so that the evaluation of the target to-be-evaluated service is effectively realized.

Description

Data processing method and device

Technical Field

The present invention relates to the field of data processing technologies, and in particular, to a data processing method and apparatus.

Background

With the improvement of intelligent science and technology, the deployment and application range of banks to website robots are continuously expanded.

The website robot can be business processing equipment deployed in banking websites, can transact business requests proposed by clients on the basis of boring with the clients and helping to solve the problems of the clients, and solves the demands of the clients. The banking sites have a plurality of service types to be processed, and the site robots cannot cover all services due to limited processing resources.

Currently, the prior art can evaluate the improvement amplitude of the website efficiency that the banking website can obtain when each business is used as a learning object of the website robot, and can determine the business with the greatest improvement amplitude of the website efficiency of the banking website as the learning object of the website robot.

However, in the above-described process of evaluating a service, the prior art fails to effectively evaluate the service.

Disclosure of Invention

In view of the above problems, the present invention provides a data processing method and apparatus for overcoming the above problems or at least partially solving the above problems, and the technical solution is as follows:

a data processing method, comprising:

acquiring working information of each teller in a target network point, wherein the working information at least comprises age and time for entering the office;

acquiring historical handling information of the target network point on a target service to be evaluated, wherein the historical handling information comprises handling quantity and handling time information in the historical period;

inputting first input data into a trained proficiency assessment model to obtain a proficiency assessment score output by the proficiency assessment model, wherein the first input data comprises work information of each teller, history handling information and the target business to be assessed;

Obtaining time information of the target network point for handling each request service in a target period, wherein the time information comprises number calling time and waiting time;

inputting second input data into a trained overall waiting time length evaluation model to obtain an overall waiting time length evaluation score output by the overall waiting time length evaluation model, wherein the second input data is composed of at least one element data which is orderly arranged, and each element data comprises the target service to be evaluated, one request service processed by the target network point in the target time period and corresponding time information;

and determining the sum of the reciprocal of the proficiency evaluation score and the overall waiting time evaluation score as the final effective score corresponding to the target service to be evaluated.

Optionally, the target service to be evaluated is a service to be evaluated in a service set to be evaluated, where the service set to be evaluated includes at least one service to be evaluated; the method further comprises the steps of:

respectively obtaining final effective scores corresponding to all the to-be-evaluated services in the to-be-evaluated service set;

and sequencing the final effective scores according to the sequence from the high score to the low score, and determining the service to be evaluated corresponding to the final effective score with the sequencing sequence number not larger than a preset threshold value as the service to be learned of the net point robot.

Optionally, the duration of the target period is one day; the method further comprises the steps of:

collecting time information of the first network site handling the request service every day through a number calling machine;

determining a training sample of the overall waiting time length assessment model by utilizing each request service handled by the first website in the same day and corresponding time information, wherein the training sample consists of at least one ordered sub-data, and each sub-data comprises a first service to be assessed, one request service handled by the first website in the same day and corresponding time information;

a positive sample or a negative sample is obtained, the positive sample being the training sample determined manually based on the principle of optimizing the dot efficacy of the first dot, the negative sample being the training sample determined manually based on the principle of optimizing the dot efficacy of the first dot.

Optionally, the method further comprises:

when the training sample is a positive sample, marking the overall waiting time length evaluation score corresponding to the training sample as 1;

when the training sample is a negative sample, marking the overall waiting time length evaluation score corresponding to the training sample as 0;

And training the overall waiting duration assessment model by using the training sample.

Optionally, the method further comprises:

acquiring working information of each teller in a first network point;

acquiring historical handling information of the first website on a first service to be evaluated;

determining a first training sample, wherein the first training sample comprises work information of each teller in the first website, historical handling information of the first website on a first service to be evaluated and the first service to be evaluated;

obtaining a first proficiency assessment score of each teller and service expert for the first service to be assessed;

marking a proficiency assessment score corresponding to the first training sample as the first proficiency assessment score;

the proficiency assessment model is trained using a first training sample labeled with a first proficiency assessment score.

A data processing apparatus comprising: a first obtaining unit, a second obtaining unit, a first input unit, a third obtaining unit, a fourth obtaining unit, a second input unit, a fifth obtaining unit, and a first determining unit, wherein:

the first obtaining unit is configured to perform: acquiring working information of each teller in a target network point, wherein the working information at least comprises age and time for entering the office;

The second obtaining unit is configured to perform: acquiring historical handling information of the target network point on a target service to be evaluated, wherein the historical handling information comprises handling quantity and handling time information in the historical period;

the first input unit is configured to perform: inputting first input data into a trained proficiency assessment model, wherein the first input data comprises working information of each teller, history handling information and target business to be assessed;

the third obtaining unit is configured to perform: obtaining a proficiency assessment score output by the proficiency assessment model;

the fourth obtaining unit is configured to perform: obtaining time information of the target network point for handling each request service in a target period, wherein the time information comprises number calling time and waiting time;

the second input unit is configured to perform: inputting second input data into a trained overall waiting time length evaluation model, wherein the second input data is composed of at least one element data which is orderly arranged, and each element data comprises the target service to be evaluated, one request service handled by the target website in the target period and corresponding time information;

The fifth obtaining unit is configured to perform: obtaining the overall waiting time length evaluation score output by the overall waiting time length evaluation model;

the first determination unit is configured to perform: and determining the sum of the reciprocal of the proficiency evaluation score and the overall waiting time evaluation score as the final effective score corresponding to the target service to be evaluated.

Optionally, the target service to be evaluated is a service to be evaluated in a service set to be evaluated, where the service set to be evaluated includes at least one service to be evaluated; the apparatus further comprises: a sixth obtaining unit, a sorting unit, and a second determining unit, wherein:

the sixth obtaining unit is configured to perform: respectively obtaining final effective scores corresponding to all the to-be-evaluated services in the to-be-evaluated service set;

the sorting unit is configured to perform: ranking each of the final significant scores in order from high score to low score;

the second determination unit is configured to perform: and determining the service to be evaluated corresponding to the final effective score with the sequencing sequence number not larger than the preset threshold value as the service to be learned of the network robot.

Optionally, the duration of the target period is one day; the apparatus further comprises: a collecting unit, a third determining unit, and a seventh obtaining unit, wherein:

The collection unit is configured to perform: collecting time information of the first network site handling the request service every day through a number calling machine;

the third determination unit is configured to perform: determining a training sample of the overall waiting time length assessment model by utilizing each request service handled by the first website in the same day and corresponding time information, wherein the training sample consists of at least one ordered sub-data, and each sub-data comprises a first service to be assessed, one request service handled by the first website in the same day and corresponding time information;

the seventh obtaining unit is configured to perform: a positive sample or a negative sample is obtained, the positive sample being the training sample determined manually based on the principle of optimizing the dot efficacy of the first dot, the negative sample being the training sample determined manually based on the principle of optimizing the dot efficacy of the first dot.

Optionally, the apparatus further includes: first marking unit, second marking unit and first training unit, wherein:

the first marking unit is configured to perform: when the training sample is a positive sample, marking the overall waiting time length evaluation score corresponding to the training sample as 1;

The second marking unit is configured to perform: when the training sample is a negative sample, marking the overall waiting time length evaluation score corresponding to the training sample as 0;

the first training unit is configured to perform: and training the overall waiting duration assessment model by using the training sample.

Optionally, the apparatus further includes: an eighth obtaining unit, a ninth obtaining unit, a fourth determining unit, a tenth obtaining unit, a third marking unit, and a second training unit, wherein:

the eighth obtaining unit is configured to perform: acquiring working information of each teller in a first network point;

the ninth obtaining unit is configured to perform: acquiring historical handling information of the first website on a first service to be evaluated;

the fourth determination unit is configured to perform: determining a first training sample, wherein the first training sample comprises work information of each teller in the first website, historical handling information of the first website on a first service to be evaluated and the first service to be evaluated;

the tenth obtaining unit is configured to perform: obtaining a first proficiency assessment score of each teller and service expert for the first service to be assessed;

The third marking unit is configured to perform: marking a proficiency assessment score corresponding to the first training sample as the first proficiency assessment score;

the second training unit is configured to perform: the proficiency assessment model is trained using a first training sample labeled with a first proficiency assessment score.

The data processing method and the data processing device provided by the embodiment can obtain the working information of each teller in the target website, wherein the working information at least comprises age and time of arrival, the historical handling information of the target website on the target to-be-evaluated service is obtained, the historical handling information comprises handling quantity and handling time information in a historical period, the first input data is input into a trained proficiency evaluation model to obtain a proficiency evaluation score output by the proficiency evaluation model, the first input data comprises the working information of each teller, the historical handling information and the target to-be-evaluated service, the time information of each request service handled by the target website in the target period is obtained, the time information comprises number calling time and waiting time, the second input data is input into the trained overall waiting time evaluation model to obtain the overall waiting time evaluation score output by the overall waiting time evaluation model, the second input data comprises orderly arranged at least one element data, each element data comprises the target to-be-evaluated service, one request service handled by the target website in the target period and corresponding time information, the reciprocal and the waiting time score of the overall waiting time are corresponding to the final evaluation score, and the final evaluation score is determined.

The invention can use the proficiency evaluation model to obtain the proficiency evaluation score of the teller for the target to-be-evaluated service, can use the whole waiting time evaluation model to obtain the whole waiting time evaluation score required by a customer when handling the target to-be-evaluated service, and can determine the reciprocal of the proficiency evaluation score and the sum value of the whole waiting time evaluation score as the final effective score for evaluating the target to-be-evaluated service, and the final effective score can quantitatively evaluate the improvement amplitude of the net point efficiency of the target net point when the target to-be-evaluated service is used as a learning object of the net point robot, thereby effectively realizing the evaluation of the target to-be-evaluated service.

The foregoing description is only an overview of the present invention, and is intended to be implemented in accordance with the teachings of the present invention in order that the same may be more clearly understood and to make the same and other objects, features and advantages of the present invention more readily apparent.

Drawings

In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings that are required to be used in the embodiments or the description of the prior art will be briefly described below, and it is obvious that the drawings in the following description are only embodiments of the present application, and that other drawings may be obtained according to the provided drawings without inventive effort to a person skilled in the art.

FIG. 1 is a flowchart of a first data processing method according to an embodiment of the present invention;

FIG. 2 is a flowchart of a second data processing method according to an embodiment of the present invention;

FIG. 3 is a flowchart of a third data processing method according to an embodiment of the present invention;

FIG. 4 is a schematic diagram of a first data processing apparatus according to an embodiment of the present invention;

FIG. 5 is a schematic diagram of a second data processing apparatus according to an embodiment of the present invention;

fig. 6 is a schematic structural diagram of a third data processing apparatus according to an embodiment of the present invention.

Detailed Description

Exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. While exemplary embodiments of the present invention are shown in the drawings, it should be understood that the present invention may be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art.

As shown in fig. 1, this embodiment proposes a first data processing method, which may include the following steps:

s101, obtaining working information of each teller in a target network point, wherein the working information at least comprises age and time for entering the office;

The target website may be a banking website.

The work information can also comprise information such as posts and job levels of the teller.

Specifically, the invention can obtain the work information of each teller from the bank system.

S102, obtaining historical handling information of a target network point on a target service to be evaluated, wherein the historical handling information comprises handling quantity and handling time information in a historical period;

the target service to be evaluated may be a certain service to be evaluated in the target network point. It should be noted that, the service to be evaluated may be determined by a technician according to an actual working situation, and the number of the service to be evaluated may be one or more, which is not limited in the present invention.

Optionally, when the number of the to-be-evaluated services is multiple, the method of the present invention may select one to-be-evaluated service from the number to-be-evaluated services to determine as the current target to-be-evaluated service, to perform the method shown in fig. 1, that is, evaluate the magnitude that the performance of the mesh point of the target mesh point can be improved when the to-be-evaluated service is used as the learning object of the mesh point robot, and evaluate each to-be-evaluated service through the cyclic selection and the execution of the method shown in fig. 1.

Wherein the history period may be a certain period in the past. The invention does not limit the specific time range of the history period, for example, the history period can be one past day; for another example, the history period may be one week in the past.

The processing time information may be related information of time spent when the target website processes the target service to be evaluated. Alternatively, the transacting duration information may include at least one of a longest duration, a shortest duration, and an average duration spent by the target site transacting the target business to be evaluated.

S103, inputting first input data into a trained proficiency assessment model, wherein the first input data comprises work information, history handling information and target to-be-assessed business of each teller;

the work information of each teller, the history handling information and the target business to be evaluated in the first input data may be orderly arranged. For example, the data format in the first input data may be { (work information of each teller), the above-mentioned history handling information, target business to be evaluated }.

The invention is not limited to the model structure and training process of the proficiency assessment model. For example, the underlying model structure of the proficiency assessment model may be a deep learning neural network; for example, the present invention may train a proficiency assessment model using a random forest algorithm.

It should be noted that, the invention can evaluate the target to-be-evaluated service based on the proficiency of each teller in the target website to the target to-be-evaluated service, that is, the evaluation is performed on the range of improvement of the website effectiveness of the target website when the target to-be-evaluated service is used as the learning object of the website robot.

S104, obtaining a proficiency evaluation score output by a proficiency evaluation model;

the proficiency evaluation model can output corresponding proficiency evaluation scores which can be used for evaluating the target to-be-evaluated business of the teller aiming at the work information, the historical handling information and the target to-be-evaluated business of the teller.

It can be understood that if the proficiency evaluation score output by the proficiency evaluation model aiming at the target to-be-evaluated service is higher, the invention can determine that the proficiency of each teller in the target website for the target to-be-evaluated service is higher, and the improvement amplitude of the website effectiveness of the target website can be smaller when the target to-be-evaluated service is taken as a learning object of the website robot; if the proficiency evaluation score output by the proficiency evaluation model aiming at the target to-be-evaluated service is lower, the invention can determine that the proficiency of each teller in the target network point for the target to-be-evaluated service is lower, and the network point efficiency of the target network point can be improved to a larger extent when the target to-be-evaluated service is taken as a learning object of the network point robot.

S105, obtaining time information of the target network point for handling each request service in a target period, wherein the time information comprises number calling time and waiting time;

wherein the target period may be a certain period in the past, current or future. For example, the target period may be one day in the past.

Specifically, the invention can obtain the time information of the target network point for handling each request service in the target time period through the number calling machine of the target network point.

S106, inputting second input data into the trained overall waiting time length evaluation model, wherein the second input data is composed of at least one element data which is orderly arranged, and each element data comprises a target service to be evaluated, a request service handled by a target website in a target period and corresponding time information;

the overall waiting time length assessment model can be used for quantitatively assessing the degree of overall needed waiting time length when a client requests to transact a target service to be assessed. It should be noted that, the present invention is not limited to the basic model structure and the training process of the overall waiting time evaluation model.

It can be understood that if the overall waiting time length evaluation score output by the overall waiting time length evaluation model for the target service to be evaluated is higher, the greater the overall waiting time length required by the client when the client requests to transact the target service to be evaluated, the greater the range of improving the network point efficiency of the target network point when the target service to be evaluated is used as the learning object of the network point robot; if the overall waiting time length evaluation model outputs an overall waiting time length evaluation score for the target service to be evaluated, the smaller the overall waiting time length is required to be waited when the client requests to transact the target service to be evaluated, the smaller the amplitude of improving the network point efficiency of the target network point is when the target service to be evaluated is used as a learning object of the network point robot.

Wherein the target to-be-evaluated service, the request service and the time information in the element data can be orderly arranged. For example, the first element data may be { T1, time1, wtime1}, and the second element data may be { T2, time2, wtime2}, where time1 and time2 are times, T1 and T2 are request services handled by the target mesh point at time1 and time2, respectively, and wtime1 and wtime2 are waiting time periods when the client handles the requests at time1 and T2, respectively.

S107, obtaining the overall waiting time length evaluation score output by the overall waiting time length evaluation model;

the overall waiting time length evaluation score can be used for quantitatively evaluating the degree of the overall required waiting time length when a client requests to transact the target service to be evaluated.

Specifically, the invention can obtain the overall waiting time evaluation score of the overall waiting time evaluation model output aiming at the second input data.

S108, determining the sum of the reciprocal of the proficiency evaluation score and the overall waiting time length evaluation score as the final effective score corresponding to the target service to be evaluated.

The sum of the reciprocal of the proficiency evaluation score and the overall waiting time length evaluation score can be used as the final effective score of the quantitative evaluation target to-be-evaluated business.

The final effective score corresponding to the target service to be evaluated can quantitatively evaluate the target service to be evaluated, namely quantitatively evaluate the improvement amplitude of the target network point efficiency generated by the target service to be evaluated when the target service to be evaluated is used as a learning object of the network point robot.

It will be appreciated that the greater the final effective score corresponding to the target service under evaluation, the less proficiency the teller will be in the target service under evaluation, or the longer the overall waiting time the customer will need to handle the target service under evaluation. The invention can quantitatively evaluate the improvement amplitude of the network point efficiency of the target network point when the target service to be evaluated is used as a learning object of the network point robot according to the final effective fraction corresponding to the target service to be evaluated.

When the target service to be evaluated is used as a learning object of the website robot, if the final effective score corresponding to the target service to be evaluated is larger, the improvement amplitude of the website efficiency of the target website is larger, and if the final effective score corresponding to the target service to be evaluated is smaller, the improvement amplitude of the website efficiency of the target website is smaller.

It should be noted that, the invention can use the proficiency evaluation model to obtain the proficiency evaluation score of the teller for the target to-be-evaluated service, can use the overall waiting time evaluation model to obtain the overall waiting time evaluation score required by the customer when handling the target to-be-evaluated service, and determine the sum of the reciprocal of the proficiency evaluation score and the overall waiting time evaluation score as the final effective score for evaluating the target to-be-evaluated service, and the final effective score can quantitatively evaluate the improvement amplitude of the website effectiveness of the target website when the target to-be-evaluated service is used as the learning object of the website robot.

According to the data processing method provided by the embodiment, working information of each teller in the target network point can be obtained, the working information at least comprises age and time of arrival, historical handling information of the target network point on the target to-be-evaluated service is obtained, the historical handling information comprises handling quantity and handling time information in a historical period, first input data are input into a trained proficiency evaluation model to obtain a proficiency evaluation score output by the proficiency evaluation model, the first input data comprise the working information of each teller, the historical handling information and the target to-be-evaluated service, time information of each request service handled by the target network point in the target period is obtained, the time information comprises number calling time and waiting time, second input data are input into the trained overall waiting time evaluation model to obtain an overall waiting time evaluation score output by the overall waiting time evaluation model, the second input data are composed of orderly arranged at least one element data, each element data comprises the target to-be-evaluated service, one request service handled by the target network point in the target period and corresponding time information, the reciprocal of the proficiency evaluation score and the overall waiting time value are determined to be the final evaluation score corresponding to the target waiting time score. The invention can use the proficiency evaluation model to obtain the proficiency evaluation score of the teller for the target to-be-evaluated service, can use the whole waiting time evaluation model to obtain the whole waiting time evaluation score required by a customer when handling the target to-be-evaluated service, and can determine the reciprocal of the proficiency evaluation score and the sum value of the whole waiting time evaluation score as the final effective score for evaluating the target to-be-evaluated service, and the final effective score can quantitatively evaluate the improvement amplitude of the net point efficiency of the target net point when the target to-be-evaluated service is used as a learning object of the net point robot, thereby effectively realizing the evaluation of the target to-be-evaluated service.

Based on the steps shown in fig. 1, as shown in fig. 2, the present embodiment proposes a second data processing method. In the method, the target service to be evaluated is a service to be evaluated in a set of services to be evaluated, the set of services to be evaluated including at least one service to be evaluated. The method may further comprise the steps of:

s201, respectively obtaining final effective scores corresponding to all the services to be evaluated in the set of the services to be evaluated;

wherein the set of services under evaluation may be made up of one or more services under evaluation.

The to-be-evaluated business in the to-be-evaluated business set can be determined by a technician according to actual working conditions, and the invention is not limited to the above. For example, a technician may determine all traffic in the target site as traffic in the set of traffic to be evaluated.

Specifically, the invention can select one service to be evaluated from the service set to be evaluated each time, determine the service to be evaluated as a target service to be evaluated, and then determine the final effective score corresponding to the target service to be evaluated according to the steps shown in fig. 1 until the final effective score corresponding to each service to be evaluated in the service set to be evaluated is determined.

S202, sorting all final effective scores according to the order from high score to low score;

Specifically, the invention can sort the final effective scores corresponding to the services to be evaluated according to the order from high to low.

S203, determining the service to be evaluated corresponding to the final effective score with the sequencing number not larger than the preset threshold value as the service to be learned of the net point robot.

The preset threshold may be formulated by a technician according to actual working conditions, which is not limited by the present invention.

It should be noted that, the invention can determine the service to be evaluated corresponding to the last effective scores of the last several last effective scores which are ranked in the front as the service to be learned of the website robot. For example, when the preset threshold is 5, the invention can determine the service to be evaluated corresponding to the top 5 final effective scores which are ranked in the top 5, namely the top 5 with the highest scores, as the service to be learned of the website robot. At this time, the invention can greatly improve the network point efficiency of the target network point on the basis of improving the equipment utilization rate of the network point robot.

According to the data processing method provided by the embodiment, the service to be evaluated corresponding to the first several final effective scores which are ranked in the front, namely the highest score, can be determined to be the service to be learned of the network point robot, the equipment utilization rate of the network point robot is improved, and the network point efficiency of the target network point is improved to a greater extent.

Based on the steps shown in fig. 1, as shown in fig. 3, the present embodiment proposes a third data processing method. The method may further comprise the steps of:

s301, working information of each teller in a first network point is obtained;

the first website may be a banking website. It should be noted that the first mesh point may or may not be the target mesh point.

It should be noted that, the present invention may determine a training sample for training the proficiency assessment model by using the working information of each teller stored in the first website and the handling information of the first website for daily service requested by the customer, and train the proficiency assessment model by using the training sample.

S302, acquiring historical handling information of a first website on a first service to be evaluated;

the first service to be evaluated may be a certain service. It should be noted that the first service to be evaluated may be a service to be evaluated in the set of services to be evaluated, or may not be a service to be evaluated in the set of services to be evaluated. Further, the first service to be evaluated may be the target service to be evaluated, or may not be the target service to be evaluated.

S303, determining a first training sample, wherein the first training sample comprises work information of each teller in a first website, historical handling information of the first website on a first service to be evaluated and the first service to be evaluated;

specifically, the invention can generate the corresponding training sample by using the work information of the teller in the first website and the history handling information of the first website on the first service to be evaluated.

It will be appreciated that the input data for the actual application of the proficiency assessment model may be consistent with the training samples used in the training phase, both of which include data types and arrangements, etc. For example, the duration of the historical period in the input data when the proficiency assessment model is actually applied may be consistent with the duration of the historical period in its training samples.

S304, obtaining first proficiency evaluation scores of each teller and service expert for a first service to be evaluated;

the first proficiency evaluation score is the proficiency evaluation score of each teller and business expert for the first business to be evaluated. Alternatively, the present invention may calculate the average value of the proficiency assessment scores of each teller and business specialist for the first business to be assessed, and determine the calculated average value as the first proficiency assessment score. Of course, the present invention may also calculate the first proficiency assessment score according to the proficiency assessment scores of each teller and business expert for the first business to be assessed by other calculation methods.

S305, marking a proficiency evaluation score corresponding to the first training sample as a first proficiency evaluation score;

in particular, the present invention may use a first proficiency assessment score to label a first training sample.

S306, training the proficiency assessment model by using a first training sample marked with a first proficiency assessment score.

It should be noted that, according to steps S301, S302, S303, S304, and S305, a plurality of training samples marked with a proficiency assessment score may be obtained, and the proficiency assessment model may be trained using the plurality of training samples.

It will be appreciated that the present invention may obtain training samples from a plurality of dots for training a proficiency assessment model, such as a second dot and a third dot.

Specifically, the invention can adopt a random forest algorithm to carry out regression prediction on the proficiency assessment model. At this time, the proficiency assessment model may be a regression prediction model.

The training time for training the proficiency assessment model is not limited. For example, as shown in fig. 3, when the proficiency assessment model is not trained or is not trained, the training sample is used for training the proficiency assessment model until the trained proficiency assessment model is obtained; for another example, after the proficiency assessment model is trained, the method and the device can generate the training sample with good timeliness during the application of the proficiency assessment model, and train the trained proficiency assessment model continuously by using the training sample with good timeliness, so that the assessment accuracy of the proficiency assessment model is effectively ensured.

The data processing method provided by the embodiment can train the proficiency assessment model to obtain a trained proficiency assessment model, and can improve the assessment accuracy of proficiency assessment.

Based on the steps shown in fig. 1, the present embodiment proposes a fourth data processing method. In the method, the duration of the target period is one day. The method may further comprise the steps of:

s401, collecting time information of the first network point handling the request service every day through a number calling machine;

the first website may be a banking website. The first mesh point and the first mesh point in the third data processing method may be the same mesh point or different mesh points, which is not limited in the present invention.

Specifically, the invention can collect the time information of the first website for transacting the request business of the client every day through the number calling machine arranged at the first website.

S402, determining a training sample of the overall waiting time length assessment model by utilizing each request service handled by a first website in the same day and corresponding time information, wherein the training sample consists of at least one ordered sub-data, and each sub-data comprises a first service to be assessed, one request service handled by the first website in the same day and corresponding time information;

It should be noted that, when the overall waiting duration evaluation model is actually applied, the input data and the training samples used in the training stage may be identical in data type and arrangement form. For example, the arrangement form of the element data in the input data when the overall waiting time assessment model is actually applied can be consistent with the arrangement form of the sub-data in the training sample.

Specifically, the invention can generate a corresponding training sample by utilizing each request service handled by the first website in one day and corresponding time information.

It can be appreciated that the present invention may utilize each request service handled by the first website in different days and corresponding time information to generate a plurality of training samples.

S403, obtaining a positive sample or a negative sample, wherein the positive sample is a training sample determined manually based on the principle of optimizing the dot efficiency of the first dot, and the negative sample is a training sample determined manually based on the principle of optimizing the dot efficiency of the first dot.

Specifically, after determining a training sample, the invention can manually determine whether the training sample is a positive sample or a negative sample based on the principle of optimizing the network point efficiency of the first network point, namely according to the proficiency of each teller for the first service to be evaluated in the training sample and according to the overall waiting time of a customer when handling the requested service contained in the training sample.

It should be noted that, the present invention may obtain training samples for training the overall waiting duration evaluation model from a plurality of mesh points, such as the second mesh point and the third mesh point.

Optionally, in the other data processing method provided in this embodiment, the method may further include:

when the training sample is a positive sample, marking the whole waiting time length corresponding to the training sample as 1;

when the training sample is a negative sample, marking the whole waiting time length corresponding to the training sample as 0;

the training sample is used for training the whole waiting duration assessment model.

It should be noted that, 1 and 0 used in marking the positive sample and the negative sample in the present invention may be the feasible probability that the first service to be evaluated in the training sample can be used as the learning object of the mesh point robot. For example, the positive sample is marked as 1, that is, the feasible probability that the first service to be evaluated in the positive sample can be used as the learning object of the network point robot is 1; for another example, the negative sample is marked with 0, which means that the feasible probability that the first service to be evaluated in the negative sample can be the learning object of the mesh point robot is 0. At this time, the overall waiting time length estimation model may be a classification model, and the present invention may train the overall waiting time length estimation model using positive and negative samples labeled 1 and 0, respectively, to obtain a trained overall waiting time length estimation model. At this time, when the trained overall waiting time length assessment model is applied, the overall waiting time length assessment score output by the overall waiting time length assessment model is the feasible probability that the service to be assessed in the input data is used as the net point robot.

It will be appreciated that the present invention may obtain a plurality of positive and/or negative samples and may train the overall latency assessment model using the obtained positive and/or negative samples.

Of course, the present invention can also use other values to label positive and negative samples, and is not limited to the use of 1 and 0. Optionally, the overall waiting duration evaluation model may be a regression prediction model, the present invention may mark corresponding scores for each training sample, the marked scores may represent the feasibility degree (the higher the score may be, the greater the feasibility degree) of the service to be evaluated in the training sample as a mesh point robot learning object, and each training sample marked with the corresponding score is used to train the overall waiting duration evaluation model to obtain a trained overall waiting duration evaluation model, so that when the trained overall waiting duration evaluation model is actually applied, the feasibility degree for representing the target service to be evaluated as the mesh point robot learning object may be output according to each element data contained in the input data.

The overall wait time assessment model may be a gated loop unit (GRU, gate Recurrent Unit) model, among others. When the integral waiting time length evaluation model is trained, the training effect of the integral waiting time length evaluation model can be evaluated by using the evaluation function with the function value being the net point total waiting time length, and when the continuous iterative optimization is performed until the function value of the evaluation function is minimum, the integral waiting time length evaluation model can be determined to be trained well.

The invention is not limited to the training time for training the whole waiting time evaluation model.

The data processing method provided by the embodiment can train the whole waiting time length assessment model to obtain a trained whole waiting time length assessment model, and ensure the assessment accuracy of the whole waiting time length.

Corresponding to the method shown in fig. 1, as shown in fig. 4, this embodiment proposes a first data processing apparatus, which may include: a first obtaining unit 101, a second obtaining unit 102, a first input unit 103, a third obtaining unit 104, a fourth obtaining unit 105, a second input unit 106, a fifth obtaining unit 107, and a first determining unit 108, wherein:

the first obtaining unit 101 is configured to perform: acquiring working information of each teller in the target network point, wherein the working information at least comprises age and time for entering the office;

the target website may be a banking website.

A second obtaining unit 102 configured to perform: acquiring historical transacting information of a target network point on a target service to be evaluated, wherein the historical transacting information comprises transacting quantity and transacting time information in a historical period;

Wherein the history period may be a certain period in the past. The invention is not limited to a specific time range of the history period.

The first input unit 103 is configured to perform: inputting first input data into a trained proficiency assessment model, wherein the first input data comprises work information, historical handling information and target to-be-assessed business of each teller;

the work information of each teller, the history handling information and the target business to be evaluated in the first input data may be orderly arranged.

For example, the data format in the first input data may be { (work information of each teller), the above-mentioned history handling information, target business to be evaluated }.

The invention is not limited to the model structure and training process of the proficiency assessment model.

The third obtaining unit 104 is configured to perform: obtaining a proficiency assessment score output by a proficiency assessment model;

A fourth obtaining unit 105 configured to perform: obtaining time information of handling each request service in a target period by a target network point, wherein the time information comprises number calling time and waiting time;

A second input unit 106 configured to perform: inputting second input data into a trained overall waiting time length evaluation model, wherein the second input data consists of at least one element data which is orderly arranged, and each element data comprises a target service to be evaluated, a request service processed by a target website in a target period and corresponding time information;

A fifth obtaining unit 107 configured to perform: obtaining an overall waiting time length evaluation score output by an overall waiting time length evaluation model;

The first determination unit 108 is configured to perform: and determining the sum of the reciprocal of the proficiency evaluation score and the overall waiting time length evaluation score as the final effective score corresponding to the target service to be evaluated.

The data processing device provided by the embodiment can effectively realize the evaluation of the target service to be evaluated.

Based on the schematic structural diagram shown in fig. 4, as shown in fig. 5, the present embodiment proposes a second data processing apparatus. In the device, the target service to be evaluated is a service to be evaluated in a service set to be evaluated, and the service set to be evaluated comprises at least one service to be evaluated; the apparatus may further include: a sixth obtaining unit 201, a sorting unit 202, and a second determining unit 203, wherein:

The sixth obtaining unit 201 is configured to perform: respectively obtaining final effective scores corresponding to all the services to be evaluated in the set of the services to be evaluated;

Specifically, the invention can select one service to be evaluated from the service set to be evaluated each time, determine the service to be evaluated as a target service to be evaluated, and determine the final effective score corresponding to the target service to be evaluated until the final effective score corresponding to each service to be evaluated in the service set to be evaluated is determined.

A sorting unit 202 configured to perform: ranking the final effective scores in order from high score to low score;

A second determination unit 203 configured to perform: and determining the service to be evaluated corresponding to the final effective score with the sequencing sequence number not larger than the preset threshold value as the service to be learned of the network robot.

It should be noted that, the invention can determine the service to be evaluated corresponding to the last effective scores of the last several last effective scores which are ranked in the front as the service to be learned of the website robot. At this time, the invention can greatly improve the network point efficiency of the target network point on the basis of improving the equipment utilization rate of the network point robot.

The data processing device provided by the embodiment can determine the service to be evaluated corresponding to the last effective scores which are ranked in the front, namely the last effective score with the highest score, as the service to be learned of the website robot, improves the equipment utilization rate of the website robot, and greatly improves the website efficiency of the target website.

Based on the schematic structural diagram shown in fig. 4, as shown in fig. 6, the present embodiment proposes a third data processing apparatus. The apparatus may further include: an eighth obtaining unit 301, a ninth obtaining unit 302, a fourth determining unit 303, a tenth obtaining unit 304, a third marking unit 305, and a second training unit 306, wherein:

an eighth obtaining unit 301 configured to perform: acquiring working information of each teller in a first network point;

A ninth obtaining unit 302 configured to perform: acquiring historical handling information of a first website on a first service to be evaluated;

A fourth determination unit 303 configured to perform: determining a first training sample, wherein the first training sample comprises work information of each teller in a first website, historical handling information of the first website on a first service to be evaluated and the first service to be evaluated;

It will be appreciated that the input data for the actual application of the proficiency assessment model may be consistent with the training samples used in the training phase, both of which include data types and arrangements, etc.

A tenth obtaining unit 304 configured to perform: obtaining first proficiency assessment scores of each teller and service expert for a first service to be assessed;

The third marking unit 305 is configured to perform: marking a proficiency assessment score corresponding to the first training sample as a first proficiency assessment score;

A second training unit 306 configured to perform: the proficiency assessment model is trained using a first training sample labeled with a first proficiency assessment score.

It will be appreciated that the present invention may obtain a plurality of training samples labeled with proficiency assessment scores and may use the plurality of training samples to train the proficiency assessment model.

The training time for training the proficiency assessment model is not limited.

The data processing device provided by the embodiment can train the proficiency assessment model to obtain a trained proficiency assessment model, and can improve the assessment accuracy of proficiency assessment.

Based on the schematic structure shown in fig. 4, the present embodiment proposes a fourth data processing apparatus. In the device, the duration of the target period is one day; the apparatus may further include: a collecting unit, a third determining unit, and a seventh obtaining unit, wherein:

A collection unit configured to perform: collecting time information of the first network site handling the request service every day through a number calling machine;

the first website may be a banking website. The first mesh point and the first mesh point in the third data processing device may be the same mesh point or different mesh points, which is not limited in the present invention.

A third determination unit configured to perform: determining a training sample of the overall waiting time length assessment model by utilizing each request service handled by the first website in the same day and corresponding time information, wherein the training sample consists of at least one sub data which is orderly arranged, and each sub data comprises a first service to be assessed, one request service handled by the first website in the same day and corresponding time information;

it should be noted that, when the overall waiting duration evaluation model is actually applied, the input data and the training samples used in the training stage may be identical in data type and arrangement form.

A seventh obtaining unit configured to perform: a positive sample, which is a training sample determined manually based on the principle of optimizing the dot efficiency of the first dot, or a negative sample, which is a training sample determined manually based on the principle of optimizing the dot efficiency of the first dot, is obtained.

Optionally, in the other data processing apparatus provided in this embodiment, the method may further include:

first marking unit, second marking unit and first training unit, wherein:

a first marking unit configured to perform: when the training sample is a positive sample, marking the overall waiting time length evaluation score corresponding to the training sample as 1;

a second marking unit configured to perform: when the training sample is a negative sample, marking the overall waiting time length evaluation score corresponding to the training sample as 0;

a first training unit configured to perform: the training sample is used for training the whole waiting duration assessment model.

It should be noted that, 1 and 0 used in marking the positive sample and the negative sample in the present invention may be the feasible probability that the first service to be evaluated in the training sample can be used as the learning object of the mesh point robot. At this time, the overall waiting time length estimation model may be a classification model, and the present invention may train the overall waiting time length estimation model using positive and negative samples labeled 1 and 0, respectively, to obtain a trained overall waiting time length estimation model. At this time, when the trained overall waiting time length assessment model is applied, the overall waiting time length assessment score output by the overall waiting time length assessment model is the feasible probability that the service to be assessed in the input data is used as the net point robot.

It will be appreciated that the present invention may obtain a plurality of positive and/or negative samples and may train the overall latency assessment model using the obtained positive and/or negative samples. Of course, the present invention can also use other values to label positive and negative samples, and is not limited to the use of 1 and 0.

The data processing device provided by the embodiment can train the whole waiting duration evaluation model to obtain a trained whole waiting duration evaluation model, and ensure the evaluation accuracy of the whole waiting duration.

It should also be noted that the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article or apparatus that comprises an element.

The foregoing is merely exemplary of the present application and is not intended to limit the present application. Various modifications and changes may be made to the present application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc. which are within the spirit and principles of the present application are intended to be included within the scope of the claims of the present application.

Claims

1. A method of data processing, comprising:

acquiring historical handling information of the target network point on a target service to be evaluated, wherein the historical handling information comprises handling quantity and handling time information in a historical period;

inputting first input data into a trained proficiency assessment model to obtain a proficiency assessment score output by the proficiency assessment model, wherein the first input data comprises work information of each teller, history handling information and the target business to be assessed; the proficiency assessment model is a model obtained by training based on a first training sample marked with a first proficiency assessment score; the first training sample comprises work information of each teller in a first website, history handling information of the first website on a first service to be evaluated and the first service to be evaluated; the first proficiency assessment score is an assessment score of each teller and business expert for the first business to be assessed;

inputting second input data into a trained overall waiting time length evaluation model to obtain an overall waiting time length evaluation score output by the overall waiting time length evaluation model, wherein the second input data is composed of at least one element data which is orderly arranged, and each element data comprises the target service to be evaluated, one request service processed by the target network point in the target time period and corresponding time information; the overall waiting time length evaluation model is a model obtained by training based on a positive sample and a negative sample, wherein the positive sample and the negative sample are training samples determined manually based on a principle of optimizing the dot efficiency of the first dot; each training sample is a sample determined by utilizing each request service handled by the first website in the same day and corresponding time information, and comprises at least one piece of sub-data which is orderly arranged, wherein each piece of sub-data comprises a first service to be evaluated, one request service handled by the first website in the same day and corresponding time information; the principle of the website effectiveness comprises the proficiency of each teller on the first service to be evaluated in a training sample and the overall waiting time of a client when the client handles the request service contained in the training sample;

2. The method of claim 1, wherein the target service under evaluation is a service under evaluation in a set of services under evaluation, the set of services under evaluation comprising at least one service under evaluation; the method further comprises the steps of:

3. The method of claim 1, wherein the target period of time is one day long; the method further comprises the steps of:

and collecting the time information of the first network site handling the request service every day through the number calling machine.

4. A method according to claim 3, characterized in that the method further comprises:

And when the training sample is a negative sample, marking the overall waiting time length evaluation score corresponding to the training sample as 0.

5. The method according to claim 1, wherein the method further comprises:

acquiring working information of each teller in a first network point;

a first training sample is determined.

6. A data processing apparatus, comprising: a first obtaining unit, a second obtaining unit, a first input unit, a third obtaining unit, a fourth obtaining unit, a second input unit, a fifth obtaining unit, and a first determining unit, wherein:

the second obtaining unit is configured to perform: acquiring historical handling information of the target network point on a target service to be evaluated, wherein the historical handling information comprises handling quantity and handling time information in a historical period;

The third obtaining unit is configured to perform: obtaining a proficiency assessment score output by the proficiency assessment model; the proficiency assessment model is a model obtained by training based on a first training sample marked with a first proficiency assessment score; the first training sample comprises work information of each teller in a first website, history handling information of the first website on a first service to be evaluated and the first service to be evaluated; the first proficiency assessment score is an assessment score of each teller and business expert for the first business to be assessed;

the second input unit is configured to perform: inputting second input data into a trained overall waiting time length evaluation model, wherein the second input data is composed of at least one element data which is orderly arranged, and each element data comprises the target service to be evaluated, one request service handled by the target website in the target period and corresponding time information; the overall waiting time length evaluation model is a model obtained by training based on a positive sample and a negative sample, wherein the positive sample and the negative sample are training samples determined manually based on a principle of optimizing the dot efficiency of the first dot; each training sample is a sample determined by utilizing each request service handled by the first website in the same day and corresponding time information, and comprises at least one piece of sub-data which is orderly arranged, wherein each piece of sub-data comprises a first service to be evaluated, one request service handled by the first website in the same day and corresponding time information; the principle of the website effectiveness comprises the proficiency of each teller on the first service to be evaluated in a training sample and the overall waiting time of a client when the client handles the request service contained in the training sample;

7. The apparatus of claim 6, wherein the target service under evaluation is a set of services under evaluation, the set of services under evaluation comprising at least one service under evaluation; the apparatus further comprises: a sixth obtaining unit, a sorting unit, and a second determining unit, wherein:

8. The apparatus of claim 6, wherein the target period of time is one day long; the apparatus further comprises: a collecting unit, a third determining unit, and a seventh obtaining unit, wherein:

the third determination unit is configured to perform: determining a training sample of the overall waiting duration evaluation model by utilizing each request service handled by the first website in the same day and corresponding time information;

the seventh obtaining unit is configured to perform: either a positive or negative sample is obtained.

9. The apparatus of claim 8, wherein the apparatus further comprises: first marking unit, second marking unit and first training unit, wherein:

10. The apparatus of claim 6, wherein the apparatus further comprises: an eighth obtaining unit, a ninth obtaining unit, a fourth determining unit, a tenth obtaining unit, a third marking unit, and a second training unit, wherein:

the fourth determination unit is configured to perform: determining a first training sample;