CN111797320B

CN111797320B - Data processing method, device, equipment and storage medium

Info

Publication number: CN111797320B
Application number: CN202010628908.2A
Authority: CN
Inventors: 张萌; 葛俊; 何仲勉; 刘伟
Original assignee: China United Network Communications Group Co Ltd
Current assignee: China United Network Communications Group Co Ltd
Priority date: 2020-07-02
Filing date: 2020-07-02
Publication date: 2023-06-06
Anticipated expiration: 2040-07-02
Also published as: CN111797320A

Abstract

The embodiment of the application provides a data processing method, a device, equipment and a storage medium, wherein the method comprises the following steps: acquiring a feature vector of a target user; obtaining a satisfaction degree prediction result corresponding to the target user through a target prediction model according to the feature vector of the target user, wherein the target prediction model is obtained by training a collaborative filtering model by weights corresponding to individual features of a plurality of historical users, the feature vector of each historical user in the plurality of historical users and the satisfaction degree score; and pushing personalized demand information to the terminal used by the target user according to the satisfaction degree prediction result. The method provided by the embodiment of the invention can solve the problem that the accuracy of personalized recommendation for the user is low because the user experience satisfaction degree cannot be accurately evaluated in the prior art.

Description

Data processing method, device, equipment and storage medium

Technical Field

The embodiment of the application relates to the technical field of data processing, in particular to a data processing method, a device, equipment and a storage medium.

Background

With the increase of the user scale and the rapid increase of the data volume, operators increasingly rely on accurate data statistical analysis to realize scientific management and decision-making in order to cope with strong market competition. In order to accurately push services such as business to users, user experience satisfaction can be evaluated through data statistics analysis.

Currently, for satisfaction evaluation, the existing scheme is to build a four-layer user perception index system: the method comprises the steps of establishing a satisfaction evaluation model by using a neural network algorithm on the basis of a user perception index system, acquiring index weights influencing user perception through a three-layer mapping relation between the user perception index, a service quality index and a network quality index, and finally forming a user experience satisfaction evaluation system.

However, existing solutions utilize neural network algorithms to model the user experience, thereby obtaining an assessment of the user experience (i.e., a satisfaction assessment). Because the neural network requires sufficient training data, most of the collectable user experience data are sparse, and the training requirement of the neural network cannot be met, so that the final training is inaccurate; if the reliable result is to be obtained through the neural network algorithm, a great deal of cost is required and the realization difficulty is high. Therefore, the prior art cannot accurately realize the evaluation of the satisfaction degree of the user experience, and further the accuracy of personalized recommendation for the user is low.

Disclosure of Invention

The embodiment of the application provides a data processing method, a device, equipment and a storage medium, which are used for solving the problem that the accuracy of personalized recommendation for a user is low because the user experience satisfaction degree cannot be accurately evaluated in the prior art.

In a first aspect, an embodiment of the present application provides a data processing method, including:

acquiring a feature vector of a target user;

obtaining a satisfaction degree prediction result corresponding to the target user through a target prediction model according to the feature vector of the target user, wherein the target prediction model is obtained by training a collaborative filtering model by weights corresponding to individual features of a plurality of historical users, the feature vector of each historical user in the plurality of historical users and the satisfaction degree score;

and pushing personalized demand information to the terminal used by the target user according to the satisfaction degree prediction result.

In one possible design, the feature vector includes personality trait data and network quality data;

before the satisfaction evaluation prediction result corresponding to the target user is obtained through the target prediction model, the method further comprises the following steps:

acquiring feature vectors and satisfaction scores of each historical user in the plurality of historical users;

Dividing the plurality of historical users to generate a training set and a verification set, wherein the training set and the verification set both comprise a plurality of samples, and each sample comprises the personality characteristic data, the network quality data and the corresponding satisfaction score;

according to the personalized feature data and the satisfaction score of each sample in the training set, determining the weight corresponding to the personalized feature through a hierarchical analysis method;

training the collaborative filtering model according to the training set and the weight to obtain an initial prediction model;

and verifying the initial prediction model according to the verification set to obtain the target prediction model.

In one possible design, the obtaining feature vectors and satisfaction scores for each of the plurality of historical users includes:

acquiring individual characteristics, network quality and satisfaction degree results of each historical user, wherein the individual characteristics at least comprise gender, current use terminal type, common service type, age, academic, current use package and whether the historical user is in a common area, and the network quality at least comprises signal strength, signal quality and network time delay;

Preprocessing the personality characteristics, the network quality and the satisfaction result to obtain personality characteristic data, network quality data and satisfaction scores of the historical users;

the personalized feature data at least comprises a sex feature value, a current terminal type feature value, a common service type feature value, an age feature value, an academic feature value, a current package feature value and a feature value of a common area, and the network quality data at least comprises a signal strength feature value, a signal quality feature value and a network time delay feature value.

In one possible design, one individual feature corresponds to a weight, and training the collaborative filtering model according to the training set and the weight to obtain an initial prediction model includes:

obtaining a plurality of similar user groups according to each sample in the training set and the weight through a preset weighted similarity algorithm in the collaborative filtering model, wherein one similar user group corresponds to one similar personalized feature vector;

aiming at each similar user group, taking the similar personality characteristic vector, network quality data and satisfaction score corresponding to each historical user in the similar user group as training samples;

And adjusting parameters of the collaborative filtering model according to each training sample to obtain the initial prediction model.

In one possible design, validating the initial predictive model according to the validation set to obtain the target predictive model includes:

inputting the personality characteristic data and the network quality data of each sample in the verification set into the initial prediction model, and outputting a satisfaction degree predicted value corresponding to each sample in the verification set;

for each sample in the verification set, carrying out error calculation on the satisfaction prediction value and the satisfaction score;

if the error result is in the preset error range, determining the initial prediction model as the target prediction model;

and if the error result is not in the preset error range, adjusting parameters of the initial prediction model until the error result is in the preset error range, and determining the current adjusted prediction model as the target prediction model.

In one possible design, the number of the target users is multiple, and the obtaining, according to the feature vector of the target user, a satisfaction prediction result corresponding to the target user through a target prediction model includes:

According to the individual feature data in the feature vector of the target user, acquiring a target similar user group matched with the target user from the plurality of similar user groups through the weight stored in the target prediction model and a preset weighted similarity algorithm;

and taking the network quality data in the target similar individual feature vector corresponding to the target similar user group and the feature vector of the target user as the input quantity of the target prediction model, and outputting a satisfaction degree prediction result corresponding to the target user through the target prediction model.

In one possible design, the preset weighted similarity algorithm is:

wherein, N (u) and N (v) are the characteristic vector of the user u and the characteristic vector of the user v respectively, i represents the common personality characteristic of the user u and the user v, N (i) represents the common characteristic vector of the user u and the user v, and k (i) represents the weight corresponding to i.

In one possible design, the pushing personalized requirement information to the terminal used by the target user according to the satisfaction prediction result includes:

determining whether to prompt the target user to change the currently used package according to the satisfaction prediction result;

If the prompt is determined, acquiring the personalized feature data and the network quality data corresponding to the target user;

determining a package matched with the target user according to the personalized feature data and the network quality data of the target user;

and taking the package as personalized demand information of the target user, and pushing the personalized demand information to a terminal used by the target user.

In a second aspect, an embodiment of the present application provides a data processing apparatus, including:

the acquisition module is used for acquiring the feature vector of the target user;

the prediction module is used for obtaining a satisfaction degree prediction result corresponding to the target user through a target prediction model according to the feature vector of the target user, wherein the target prediction model is obtained by training a collaborative filtering model by weights corresponding to the individual features of a plurality of historical users, the feature vector of each historical user in the plurality of historical users and the satisfaction degree score;

and the recommending module is used for pushing personalized demand information to the terminal used by the target user according to the satisfaction degree prediction result.

In a third aspect, an embodiment of the present application provides a data processing apparatus, including: at least one processor and memory;

The memory stores computer-executable instructions;

the at least one processor executes computer-executable instructions stored by the memory such that the at least one processor performs the data processing method as described above in the first aspect and possible designs of the first aspect.

In a fourth aspect, embodiments of the present application provide a computer readable storage medium, where computer executable instructions are stored, when executed by a processor, to implement a data processing method according to the first aspect and the possible designs of the first aspect.

According to the data processing method, the device, the equipment and the storage medium, the feature vector of the target user is firstly obtained, then the satisfaction degree prediction result corresponding to the target user is obtained through the target prediction model based on the feature vector of the target user, wherein the target prediction model is obtained by training the collaborative filtering model through weights corresponding to the individual features of a plurality of historical users, the feature vector of each historical user in the plurality of historical users and the satisfaction degree score, and because the collaborative filtering method is adopted, the requirement on the magnitude of a training set is lower, the full-quantity prediction result can be obtained only by sparse matrix samples, therefore, the satisfaction degree prediction value calculation is carried out through the target prediction model, and the user experience satisfaction degree evaluation is accurately realized; and then based on the accurate satisfaction degree prediction result, personalized demand information is pushed to the target user, so that the accuracy of personalized recommendation to the user is higher, and the personalized demand of the user can be met.

Drawings

In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, a brief description will be given below of the drawings that are needed in the embodiments or the prior art descriptions, it being obvious that the drawings in the following description are some embodiments of the present application, and that other drawings may be obtained according to these drawings without inventive effort to a person skilled in the art.

Fig. 1 is a schematic view of a scenario of a data processing method according to an embodiment of the present application;

FIG. 2 is a schematic flow chart of a data processing method according to an embodiment of the present disclosure;

FIG. 3 is a flow chart of a data processing method according to another embodiment of the present disclosure;

FIG. 4 is a flowchart illustrating a data processing method according to another embodiment of the present disclosure;

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

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

Detailed Description

For the purposes of making the objects, technical solutions and advantages of the embodiments of the present application more clear, the technical solutions of the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is apparent that the described embodiments are some embodiments of the present application, but not all embodiments. All other embodiments, which can be made by one of ordinary skill in the art based on the embodiments herein without making any inventive effort, are intended to be within the scope of the present application.

The terms "first," "second," "third," "fourth" and the like in the description and in the claims of this application and in the above-described figures, if any, are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged where appropriate such that embodiments of the present application described herein may be implemented, for example, in sequences other than those illustrated or described herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.

The existing scheme utilizes a neural network algorithm to build a model for user experience, so that evaluation on the user experience is obtained. Because the neural network requires sufficient training data, most of the collectable user experience data are sparse, and the training requirement of the neural network cannot be met, so that the final training is inaccurate; if the reliable result is to be obtained through the neural network algorithm, a great deal of cost is required and the realization difficulty is high. Therefore, the prior art cannot accurately realize the evaluation of the satisfaction degree of the user experience, and further the accuracy of personalized recommendation for the user is low.

In order to solve the problems, the technical conception of the method is to acquire network experience satisfaction of a small part of users under different characteristics and network conditions, obtain predicted network experience satisfaction scores of the whole users by collaborative filtering of the characteristic attributes of the users, current network KPI data and other basic information, accurately evaluate the user experience satisfaction, and push personalized demand information to target users based on accurate satisfaction prediction results, so that accuracy of personalized recommendation of the users is high, and personalized demands of the users can be met.

The technical scheme of the present application is described in detail below with specific examples. The following embodiments may be combined with each other, and some embodiments may not be repeated for the same or similar concepts or processes.

Referring to fig. 1, fig. 1 is a schematic view of a scenario of a data processing method according to an embodiment of the present application. In practical applications, the execution subject of the data processing method may be a data processing apparatus such as the recommendation apparatus 10. The recommendation device 10 is configured to cooperatively filter the obtained information such as the characteristic attribute of the user and the current network KPI data, and classify the user group based on the characteristic attribute of the user to integrate a plurality of pieces of data, and then train the information such as the user group and the current network KPI data as a sample, and calculate the network experience satisfaction of a small part of the users to obtain a predicted network experience satisfaction score of the whole user, thereby reducing the complexity of the training process data and ensuring that the satisfaction prediction result is more accurate. And then, according to the satisfaction degree prediction result, personalized demand information can be pushed to the terminal 20 used by the target user, so that the accuracy of personalized recommendation to the user is improved, and the personalized demand of the user is further met.

Specifically, referring to fig. 2 for how to implement data processing, fig. 2 is a schematic flow chart of a data processing method according to an embodiment of the present application.

Referring to fig. 2, the data processing method includes:

s201, obtaining the feature vector of the target user.

In this embodiment, the execution subject may be a recommendation device. The recommending means here may be a server.

Specifically, the target users may be in batches, and the individual characteristics of the users to be predicted, that is, the target users, may be obtained by means of a questionnaire, such as gender, type of currently used terminal, type of commonly used service, age, academic history, currently used packages, whether in commonly used areas, and the like. The network condition corresponding to the target user, such as signal strength, signal quality, network time delay, etc., can also be obtained through the user background data. And then preprocessing the personality characteristics and the network quality, and unifying the formats to obtain personality characteristic data and network quality data. The personality characteristic data may include at least a gender characteristic value, a currently used terminal type characteristic value, a commonly used service type characteristic value, an age characteristic value, an academic characteristic value, a currently used package characteristic value, and a commonly used area characteristic value, and the network quality data may include at least a signal strength characteristic value, a signal quality characteristic value, and a network delay characteristic value. Wherein the personality trait data and the network quality data form a trait vector for the target user.

S202, obtaining a satisfaction degree prediction result corresponding to the target user through a target prediction model according to the feature vector of the target user, wherein the target prediction model is obtained by training the collaborative filtering model by weights corresponding to the individual features of a plurality of historical users, the feature vector of each historical user in the plurality of historical users and the satisfaction degree score.

In this embodiment, a target prediction model is established through a collaborative filtering model according to weights corresponding to individual features, feature vectors of each historical user, and satisfaction scores. The target prediction model may be used to predict a satisfaction score corresponding to the target user.

And S203, pushing personalized demand information to the terminal used by the target user according to the satisfaction degree prediction result.

In this embodiment, after obtaining the satisfaction prediction result corresponding to the target user, in order to meet the personalized requirement of the user, requirements such as a service suitable for the target user may be recommended to the target user. In one possible design, how to push personalized requirement information to the terminal used by the target user according to the satisfaction prediction result may include the following steps:

And a1, determining whether to prompt the target user to change the currently used package according to the satisfaction degree prediction result.

And a2, if the prompt is determined, acquiring the personalized feature data and the network quality data corresponding to the target user.

And a3, determining a package matched with the target user according to the personalized feature data and the network quality data of the target user.

And a4, taking the package as personalized demand information of the target user, and pushing the personalized demand information to a terminal used by the target user.

In this embodiment, the satisfaction prediction result may be 1 score, 2 scores, 3 scores, 4 scores, 5 scores, etc. Wherein, 1 may represent very dissatisfaction, 2 represents dissatisfaction, 3 represents general, 4 represents satisfaction, and 5 represents very satisfaction.

Specifically, the satisfaction degree of the target user on the current service and the network quality can be determined according to the satisfaction degree prediction result. The reason for the unsatisfactory degree is probably that the reasons for the unsatisfactory degree are not matched with the packages used by the target user, so if the result of the satisfactory degree prediction shows that the packages are very unsatisfactory, unsatisfactory or general, the packages suitable for the service and the like of the target user can be pushed to the terminal used by the target user. The package may be determined according to the personality characteristic data of the target user and the current network quality data. For example, the current package used by the target user does not match the current network quality, and a package matching the current network quality needs to be recommended to the target user or the current network quality needs to be adjusted to meet the personalized requirements of the target user.

According to the data processing method provided by the embodiment, the feature vector of the target user is obtained, and then the satisfaction degree prediction result corresponding to the target user is obtained through the target prediction model based on the feature vector of the target user, wherein the target prediction model is obtained by training the collaborative filtering model through weights corresponding to the individual features of a plurality of historical users, the feature vector and the satisfaction degree score of each historical user in the plurality of historical users, and due to the adoption of the collaborative filtering method, the requirement on the magnitude of a training set is low, and the full quantity prediction result can be obtained only by sparse matrix samples, so that the satisfaction degree prediction value calculation is carried out through the target prediction model, and the user experience satisfaction degree evaluation is accurately realized; and then based on the accurate satisfaction degree prediction result, personalized demand information is pushed to the target user, so that the accuracy of personalized recommendation to the user is higher, and the personalized demand of the user can be met.

Therefore, the data processing method adopts a collaborative filtering method to establish a user experience model, namely a target prediction model, has low requirements on training data sets, can obtain better user experience prediction results by using fewer data sets, accurately realizes the evaluation of user experience satisfaction degree, and simultaneously improves the accuracy of personalized recommendation for users.

Referring to fig. 3, fig. 3 is a flow chart of a data processing method according to another embodiment of the present application. The present embodiment describes in detail how to build the target prediction model based on the above embodiment, for example, based on the embodiment described in fig. 2. The feature vector may include personality characteristic data and network quality data, and before the satisfaction evaluation prediction result corresponding to the target user is obtained through the target prediction model, the method may further include:

s301, obtaining feature vectors and satisfaction scores of each historical user in the plurality of historical users.

In this embodiment, preprocessing, such as format unified processing, may be performed by acquiring individual characteristics, network quality and corresponding satisfaction results of each of the plurality of historical users, so as to facilitate data processing during subsequent model training. The preprocessing herein may convert personality traits, network quality, and satisfaction results into a simplified digital format.

In one possible design, how to obtain the feature vector and satisfaction score of each of the plurality of historical users may be achieved by:

And b1, acquiring individual characteristics, network quality and satisfaction degree results of each historical user, wherein the individual characteristics at least comprise gender, current terminal type, common service type, age, academic, current package and whether the historical user is in a common area, and the network quality at least comprises signal strength, signal quality and network time delay.

And b2, preprocessing the personality characteristics, the network quality and the satisfaction result to obtain personality characteristic data, network quality data and satisfaction scores of the historical users.

In this embodiment, the personality characteristics of the historical user and the satisfaction results may be acquired by on-line questionnaires, where the personality characteristics may include at least gender, current service type, age, academic history, current package, whether in a common area, and the satisfaction results may include very unsatisfactory, unsatisfied, general, satisfied, very satisfactory, and so on. The network condition corresponding to the target user, such as signal strength, signal quality, network time delay, etc., can also be obtained through background data when the user fills in the questionnaire.

Wherein, the personality characteristics of the user can be seen in table 1:

TABLE 1

Specifically, the personality characteristics in the table are digitally converted, such as male representing 0 and female representing 1 in gender; the ios in the current use terminal represents 0, the android represents 1 and the like, and the gender, the current use terminal type, the common service type, the age, the academic, the current use package, whether the current use package is in a common area, the signal strength, the signal quality, the network delay and the satisfaction result are preprocessed through a preset algorithm to obtain a gender characteristic value, a current use terminal type characteristic value, a common service type characteristic value, an age characteristic value, an academic characteristic value, a current use package characteristic value, whether the current use package is in a common area characteristic value, the signal strength characteristic value, the signal quality characteristic value, the network delay characteristic value and the satisfaction score.

S302, dividing the historical users to generate a training set and a verification set, wherein the training set and the verification set comprise a plurality of samples, and each sample comprises the personality characteristic data, the network quality data and the corresponding satisfaction score.

In this embodiment, the personality characteristic data, the network quality data, and the corresponding satisfaction score of each historical user may be used as a sample, and first, the sample with missing data may be cleaned, i.e., deleted, and then, a plurality of samples corresponding to the remaining historical users may be divided into a training set and a verification set. The training set and the verification set both comprise a plurality of samples which are respectively used for training and verifying the collaborative filtering model, so that the trained collaborative filtering model is stable, i.e. the predicted result tends to the actual value, and the accuracy of the satisfaction degree predicted result is ensured.

S303, determining the weight corresponding to the personalized features through a hierarchical analysis method according to the personalized feature data and the satisfaction score of each sample in the training set.

In this embodiment, before the collaborative filtering model is trained through the samples, in order to reduce the complexity of the operation of the data processing system and simplify the data processing process, the users may be first grouped through the first layer architecture of the collaborative filtering model, i.e. the similar users are grouped into a group. Here the look-up or determination of similar users may be calculated by combining the feature values with weights.

The method comprises the steps of determining weights, namely, selecting personalized feature data of each sample in a training set and the satisfaction score as analysis objects through a hierarchical analysis method, decomposing decision-making problems into different hierarchical structures according to the sequence of a total target, each layer of sub-targets and an evaluation criterion until a specific scheme, solving a method for judging matrix feature vectors to obtain the priority weights of each element of each layer on a certain element of the previous layer, and finally obtaining the weights corresponding to each personalized feature in a hierarchical way through a weighted sum method. One individual feature here corresponds to one weight, and different individual features correspond to different weights.

S304, training the collaborative filtering model according to the training set and the weight to obtain an initial prediction model.

In this embodiment, unlike the recommendation system that uses the interested items to find a user group similar to the target user, the present embodiment uses the preprocessed feature vector of the user and the weights of different personalized feature values in the feature vector to perform grouping. The personalized feature value is the personalized feature data.

In one possible design, how to determine the weights and enable the establishment of an initial predictive model may be accomplished by:

and c1, obtaining a plurality of similar user groups according to each sample in the training set and the weight through a preset weighted similarity algorithm in the collaborative filtering model, wherein one similar user group corresponds to one type of similar personalized feature vector.

And c2, aiming at each similar user group, taking the similar personality characteristic vector, network quality data and satisfaction scores corresponding to each historical user in the similar user group as training samples.

And step c3, adjusting parameters of the collaborative filtering model according to each training sample to obtain the initial prediction model.

In this embodiment, similarity calculation between users can be achieved through a preset weighted similarity algorithm fused in the collaborative filtering model, so that user clustering is achieved. The preset weighted similarity algorithm may be a weighted similarity (i.e. Jaccard) formula, i.e.:

wherein N (u) and N (v) are the feature vector of the user u and the feature vector of the user v respectively, i represents the same personality of the user u and the user v, N (i) represents the feature vector containing the same personality, and k (i) represents the weight corresponding to i. Here, N (i) may be formed by uniformly classifying the feature values of different individual features into 0 by using the feature value of the same individual feature as the actual feature value corresponding to i.

Specifically, any sample in the training set contains feature vectors of corresponding users, the similarity between any users can be calculated through the formula, then each user is divided into a plurality of similar user groups based on the similarity, and each similar user group can be represented by a feature value to replace the personalized feature data corresponding to each similar user in each similar user group. And then taking the similar individual feature vector corresponding to each similar user group, the network quality data and the satisfaction score corresponding to each historical user in the similar user group as training samples, and adjusting the parameters of the collaborative filtering model to obtain an initial prediction model. The determination of the similar user group optimizes the data of the subsequent model training, and avoids the complexity of operation caused by taking a large amount of data as training data.

And S305, verifying the initial prediction model according to the verification set to obtain the target prediction model.

In this embodiment, after the initial prediction model is obtained, in order to verify the stability of the model, the initial prediction model may be verified by a verification set.

Specifically, the verification process may be implemented by:

and d1, inputting the personality characteristic data and the network quality data of each sample in the verification set into the initial prediction model, and outputting a satisfaction degree predicted value corresponding to each sample in the verification set.

And d2, carrying out error calculation on the satisfaction prediction value and the satisfaction score for each sample in the verification set.

And d3, if the error result is within a preset error range, determining the initial prediction model as the target prediction model.

And d4, if the error result is not in the preset error range, adjusting parameters of the initial prediction model until the error result is in the preset error range, and determining that the current adjusted prediction model is the target prediction model.

In this embodiment, first, the personality characteristic data and the network quality data in each sample of the verification set are selected, the personality characteristic data and the network quality data are used as input amounts of an initial prediction model, the initial prediction model is used for predicting, and a satisfaction degree prediction value corresponding to each sample is output. And then calculating the deviation between the satisfaction prediction value and the corresponding satisfaction score (namely the actual value), and continuously optimizing the initial prediction model through feedback adjustment.

If the error result is within the preset error range, the initial prediction model is stable, and the initial prediction model can be used as a target prediction model; if the error result is within the preset error range, the inaccuracy of the prediction result of the initial prediction model is indicated, the parameters of the initial prediction model can be adjusted through the error value between the satisfaction degree prediction value and the satisfaction degree score, then verification is carried out until the error result is adjusted within the preset error range, the optimization of the initial prediction model can be stopped, and the current adjusted prediction model is taken as the target prediction model.

Referring to fig. 4, fig. 4 is a flowchart of a data processing method according to still another embodiment of the present application, and the embodiment describes in detail how to obtain a satisfaction prediction result corresponding to the target user based on the above embodiment. The obtaining, according to the feature vector of the target user, a satisfaction prediction result corresponding to the target user through a target prediction model may include:

s401, acquiring target similar user groups matched with the target user from the plurality of similar user groups through the weights stored in the target prediction model and a preset weighted similarity algorithm according to the individual feature data in the feature vector of the target user.

S402, taking network quality data in the target similar individual feature vector corresponding to the target similar user group and the feature vector of the target user as input quantity of the target prediction model, and outputting a satisfaction degree prediction result corresponding to the target user through the target prediction model.

In this embodiment, since the weight corresponding to each individual feature is already determined by the training set during the training process, the weight does not need to be calculated when the target user satisfaction is verified or predicted by the verification set, and the weight is already stored in the target prediction model. Therefore, during verification, the similar user group to which the user corresponding to each sample in the verification set belongs can be determined directly through the calculation of the similarity; when the satisfaction degree of the target users in batches is predicted, the similar user group of each target user can be determined directly through the calculation of the similarity degree.

Specifically, a similar user group matched with a target user, namely a target similar user group, is calculated in a target prediction model, then a target similar personalized feature vector corresponding to the target similar user group is obtained, network quality data in the target similar personalized feature vector and the feature vector of the target user are used as input quantity for prediction of the target prediction model, a satisfaction degree prediction value corresponding to the target user is output through the target prediction model, and a satisfaction degree prediction result is determined. The prediction method has the advantages that the prediction process is simple, complex calculation is not needed for data of the magnitude required by the neural network training, resources are saved, the accuracy of the prediction result is high, a basis is provided for the user to recommend personalized demand information, and the personalized demand of the user can be better met.

In practical applications, the present application collaborative filters the user feature matrix (i.e., feature vector) to predict user network experiences. Specifically, the experience situation of the user can be predicted according to the characteristics of the gender, age, service type, network situation and the like of the user. Because the method is used for a scene of processing based on a large amount of eigenvalue data, a distributed system infrastructure Hadoop and other parallel storage systems can be adopted for processing the data.

User-based collaborative filtering algorithms (collaborative filtering, CF) can handle user network experience prediction problems, predicting experiences through user characteristics. Specifically, the method and the device mainly carry out collaborative filtering on the characteristic attribute of the user and basic information such as the current network KPI data (such as network quality) and the like, acquire network experience satisfaction of a small part of users under different characteristics and network conditions through methods such as online questionnaires and the like, and further obtain predicted network experience satisfaction scores of the whole users through a prediction model. Therefore, the processes of data collection, collaborative filtering method for network experience prediction, user experience prediction, prediction result optimization and the like can predict the network experience conditions of users without survey data in the user 1 and other user groups A by analyzing the current network conditions of the users in the user groups A and the network experience conditions of the users with survey data, so as to obtain different user experience evaluation types, and according to the characteristics of network experience evaluation of different user groups, a more accurate user experience value, namely a satisfaction prediction score, is predicted. And along with the increase of user feedback, the target prediction model can also compare the user network experience prediction data with newly added real data, and the user network experience prediction result can be continuously optimized along with the increase of the user feedback data, so that the prediction deviation is reduced until the prediction deviation approaches to the real situation infinitely.

In order to implement the data processing method, the present embodiment provides a data processing apparatus. Referring to fig. 5, fig. 5 is a schematic structural diagram of a data processing apparatus according to an embodiment of the present application; the data processing device 50 includes: an acquisition module 501, a prediction module 502, and a recommendation module 503; an obtaining module 501, configured to obtain a feature vector of a target user; the prediction module 502 is configured to obtain a satisfaction prediction result corresponding to the target user through a target prediction model according to feature vectors of the target user, where the target prediction model is obtained by training a collaborative filtering model by weights corresponding to personality features of a plurality of historical users, feature vectors of each historical user in the plurality of historical users, and satisfaction scores; and a recommending module 503, configured to push personalized demand information to a terminal used by the target user according to the satisfaction prediction result.

According to the embodiment, the obtaining module 501, the predicting module 502 and the recommending module 503 are configured to obtain the feature vector of the target user, and then obtain the satisfaction degree prediction result corresponding to the target user through the target prediction model based on the feature vector of the target user, where the target prediction model is obtained by training the collaborative filtering model through weights corresponding to the individual features of a plurality of historical users, the feature vector and the satisfaction degree score of each historical user in the plurality of historical users, and because the collaborative filtering method is adopted, the requirement on the magnitude of the training set is lower, the full quantity prediction result can be obtained only by sparse matrix samples, and therefore, the satisfaction degree prediction value calculation is performed by using the target prediction model, and the user experience satisfaction degree evaluation is accurately realized; and then based on the accurate satisfaction degree prediction result, personalized demand information is pushed to the target user, so that the accuracy of personalized recommendation to the user is higher, and the personalized demand of the user can be met.

The device provided in this embodiment may be used to implement the technical solution of the foregoing method embodiment, and its implementation principle and technical effects are similar, and this embodiment will not be described herein again.

In one possible design, the feature vector includes personality trait data and network quality data; the apparatus further comprises: a model building module 504; the model building module 504 is configured to obtain feature vectors and satisfaction scores of each of the plurality of historical users before the satisfaction evaluation prediction result corresponding to the target user is obtained through the target prediction model; dividing the plurality of historical users to generate a training set and a verification set, wherein the training set and the verification set both comprise a plurality of samples, and each sample comprises the personality characteristic data, the network quality data and the corresponding satisfaction score; according to the personalized feature data and the satisfaction score of each sample in the training set, determining the weight corresponding to the personalized feature through a hierarchical analysis method; training the collaborative filtering model according to the training set and the weight to obtain an initial prediction model; and verifying the initial prediction model according to the verification set to obtain the target prediction model.

In one possible design, the modeling module 504 is specifically configured to: acquiring individual characteristics, network quality and satisfaction degree results of each historical user, wherein the individual characteristics at least comprise gender, current use terminal type, common service type, age, academic, current use package and whether the historical user is in a common area, and the network quality at least comprises signal strength, signal quality and network time delay; preprocessing the personality characteristics, the network quality and the satisfaction result to obtain personality characteristic data, network quality data and satisfaction scores of the historical users; the personalized feature data at least comprises a sex feature value, a current terminal type feature value, a common service type feature value, an age feature value, an academic feature value, a current package feature value and a feature value of a common area, and the network quality data at least comprises a signal strength feature value, a signal quality feature value and a network time delay feature value.

In one possible design, one individual feature corresponds to one weight, and the model building module 504 is further specifically configured to: obtaining a plurality of similar user groups according to each sample in the training set and the weight through a preset weighted similarity algorithm in the collaborative filtering model, wherein one similar user group corresponds to one similar personalized feature vector; aiming at each similar user group, taking the similar personality characteristic vector, network quality data and satisfaction score corresponding to each historical user in the similar user group as training samples; and adjusting parameters of the collaborative filtering model according to each training sample to obtain the initial prediction model.

In one possible design, the modeling module 504 is further specifically configured to: inputting the personality characteristic data and the network quality data of each sample in the verification set into the initial prediction model, and outputting a satisfaction degree predicted value corresponding to each sample in the verification set; for each sample in the verification set, carrying out error calculation on the satisfaction prediction value and the satisfaction score; if the error result is in the preset error range, determining the initial prediction model as the target prediction model; and if the error result is not in the preset error range, adjusting parameters of the initial prediction model until the error result is in the preset error range, and determining the current adjusted prediction model as the target prediction model.

In one possible design, the prediction module 502 is specifically configured to: according to the individual feature data in the feature vector of the target user, acquiring a target similar user group matched with the target user from the plurality of similar user groups through the weight stored in the target prediction model and a preset weighted similarity algorithm; and taking the network quality data in the target similar individual feature vector corresponding to the target similar user group and the feature vector of the target user as the input quantity of the target prediction model, and outputting a satisfaction degree prediction result corresponding to the target user through the target prediction model.

In one possible design, the preset weighted similarity algorithm is:

In one possible design, the recommendation module 503 is specifically configured to: determining whether to prompt the target user to change the currently used package according to the satisfaction prediction result; if the prompt is determined, acquiring the personalized feature data and the network quality data corresponding to the target user; determining a package matched with the target user according to the personalized feature data and the network quality data of the target user; and taking the package as personalized demand information of the target user, and pushing the personalized demand information to a terminal used by the target user.

In order to implement the data processing method, the present embodiment provides a data processing apparatus. Fig. 6 is a schematic structural diagram of a data processing apparatus according to an embodiment of the present application. As shown in fig. 6, the data processing apparatus 60 of the present embodiment includes: a processor 601 and a memory 602; wherein the memory 602 is configured to store computer-executable instructions; a processor 601 for executing computer-executable instructions stored in a memory to perform the steps performed in the above embodiments. See in particular the description of the method embodiments described above.

The embodiment of the application also provides a computer readable storage medium, wherein computer execution instructions are stored in the computer readable storage medium, and when a processor executes the computer execution instructions, the data processing method is realized.

In the several embodiments provided in this application, it should be understood that the disclosed apparatus and method may be implemented in other ways. For example, the above-described embodiments of the apparatus are merely illustrative, and for example, the division of the modules is merely a logical function division, and there may be additional divisions when actually implemented, for example, multiple modules may be combined or integrated into another system, or some features may be omitted or not performed. Alternatively, the coupling or direct coupling or communication connection shown or discussed with each other may be an indirect coupling or communication connection via some interfaces, devices or modules, which may be in electrical, mechanical, or other forms. In addition, each functional module in each embodiment of the present application may be integrated in one processing unit, or each module may exist alone physically, or two or more modules may be integrated in one unit. The units formed by the modules can be realized in a form of hardware or a form of hardware and software functional units.

The integrated modules, which are implemented in the form of software functional modules, may be stored in a computer readable storage medium. The software functional module is stored in a storage medium, and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) or a processor (english: processor) to perform some of the steps of the methods described in the embodiments of the present application. It should be understood that the above processor may be a central processing unit (english: central Processing Unit, abbreviated as CPU), or may be other general purpose processors, digital signal processors (english: digital Signal Processor, abbreviated as DSP), application specific integrated circuits (english: application Specific Integrated Circuit, abbreviated as ASIC), or the like. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like. The steps of a method disclosed in connection with the present invention may be embodied directly in a hardware processor for execution, or in a combination of hardware and software modules in a processor for execution.

The memory may comprise a high-speed RAM memory, and may further comprise a non-volatile memory NVM, such as at least one magnetic disk memory, and may also be a U-disk, a removable hard disk, a read-only memory, a magnetic disk or optical disk, etc. The bus may be an industry standard architecture (Industry Standard Architecture, ISA) bus, an external device interconnect (Peripheral Component, PCI) bus, or an extended industry standard architecture (Extended Industry Standard Architecture, EISA) bus, among others. The buses may be divided into address buses, data buses, control buses, etc. For ease of illustration, the buses in the drawings of the present application are not limited to only one bus or one type of bus. The storage medium may be implemented by any type or combination of volatile or nonvolatile memory devices such as Static Random Access Memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic or optical disk. A storage media may be any available media that can be accessed by a general purpose or special purpose computer.

An exemplary storage medium is coupled to the processor such the processor can read information from, and write information to, the storage medium. In the alternative, the storage medium may be integral to the processor. The processor and the storage medium may reside in an application specific integrated circuit (Application Specific Integrated Circuits, ASIC for short). It is also possible that the processor and the storage medium reside as discrete components in an electronic device or a master device.

Those of ordinary skill in the art will appreciate that: all or part of the steps for implementing the method embodiments described above may be performed by hardware associated with program instructions. The foregoing program may be stored in a computer readable storage medium. The program, when executed, performs steps including the method embodiments described above; and the aforementioned storage medium includes: various media that can store program code, such as ROM, RAM, magnetic or optical disks.

Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present application, and not for limiting the same; although the present application has been described in detail with reference to the foregoing embodiments, it should be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some or all of the technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit of the corresponding technical solutions from the scope of the technical solutions of the embodiments of the present application.

Claims

1. A method of data processing, comprising:

acquiring a feature vector of a target user;

according to the satisfaction degree prediction result, personalized demand information is pushed to a terminal used by the target user;

the feature vector comprises personalized feature data and network quality data;

verifying the initial prediction model according to the verification set to obtain the target prediction model;

and a weight is corresponding to one individual feature, the collaborative filtering model is trained according to the training set and the weight, and an initial prediction model is obtained, and the method comprises the following steps:

adjusting parameters of the collaborative filtering model according to each training sample to obtain the initial prediction model;

The preset weighted similarity algorithm is as follows:

2. The method of claim 1, wherein the obtaining feature vectors and satisfaction scores for each of the plurality of historical users comprises:

3. The method of claim 1, wherein validating the initial predictive model from the validation set results in the target predictive model, comprising:

4. The method of claim 3, wherein the number of target users is plural, and the obtaining, according to the feature vector of the target user, the satisfaction prediction result corresponding to the target user through a target prediction model includes:

5. The method according to any one of claims 2-4, wherein pushing personalized demand information to a terminal used by the target user according to the satisfaction prediction result includes:

6. A data processing apparatus, comprising:

the recommendation module is used for pushing personalized demand information to a terminal used by the target user according to the satisfaction degree prediction result;

the feature vector comprises personalized feature data and network quality data; the apparatus further comprises: a model building module;

the model building module is used for acquiring the characteristic vector and the satisfaction score of each historical user in the plurality of historical users before the satisfaction evaluation prediction result corresponding to the target user is obtained through the target prediction model; dividing the plurality of historical users to generate a training set and a verification set, wherein the training set and the verification set both comprise a plurality of samples, and each sample comprises the personality characteristic data, the network quality data and the corresponding satisfaction score; according to the personalized feature data and the satisfaction score of each sample in the training set, determining the weight corresponding to the personalized feature through a hierarchical analysis method; training the collaborative filtering model according to the training set and the weight to obtain an initial prediction model; verifying the initial prediction model according to the verification set to obtain the target prediction model;

One of the individual features corresponds to one of the weights, and the model building module is further specifically configured to: obtaining a plurality of similar user groups according to each sample in the training set and the weight through a preset weighted similarity algorithm in the collaborative filtering model, wherein one similar user group corresponds to one similar personalized feature vector; aiming at each similar user group, taking the similar personality characteristic vector, network quality data and satisfaction score corresponding to each historical user in the similar user group as training samples; adjusting parameters of the collaborative filtering model according to each training sample to obtain the initial prediction model;

the preset weighted similarity algorithm is as follows:

7. A data processing apparatus, comprising: at least one processor and memory;

the memory stores computer-executable instructions;

the at least one processor executing computer-executable instructions stored in the memory causes the at least one processor to perform the data processing method of any one of claims 1-5.

8. A computer readable storage medium having stored therein computer executable instructions which, when executed by a processor, implement the data processing method of any of claims 1-5.