Summary of the invention
Present specification describes a kind of determination methods of service plan, pass through prediction user telling for client service
It asks, and then provides the service plan for being suitble to the user.
According to the embodiment of first aspect, a kind of determination method of service plan, the executing subject of the method are provided
For server-side, which comprises the triggering in response to active user to the first scene in client service obtains described current
The user characteristics of user;The user characteristics are inputted in disaggregated model, predict current group belonging to the active user;Its
Described in group result of the disaggregated model based on multiple users and train in advance, the group result of the multiple user is based on and institute
State filling in result and determining for the relevant questionnaire of client service;Mapping based on predetermined group and service plan
Relationship determines the corresponding current service scheme of the current group.
In one embodiment, after determining the corresponding current service scheme of the current group, further includes: to client
End pushes the current service scheme, so that the client, which is based on the current service scheme, shows phase to the active user
The service interface answered.
In one embodiment, the user characteristics include user property feature, and/or, for the client service
Interaction feature;Wherein the user property feature includes gender, age, occupation, marriage and childbirth situation, at least one in hobby
Kind, the interaction feature includes that number of clicks uses at least one of duration, place to use, sharing number.
In one embodiment, the disaggregated model is based on following steps and trains in advance: obtaining the multiple user's
It fills in as a result, and obtaining the user characteristics of each user in the multiple user;Filling in based on the multiple user as a result,
The multiple user is grouped;The user characteristics of result and each user based on the grouping, determine multiple instructions
Practice sample;Wherein each training sample includes the user characteristics and sample label of each user, described in the sample label instruction
Group belonging to each user;Utilize the multiple training sample training disaggregated model.
Further, in a specific embodiment, the multiple user includes the first user;Obtain the multiple use
The result of filling at family includes: the triggering in response to first user to the second scene in the client service, Xiang Suoshu
One user pushes the questionnaire;It receives first user and fills in result for the questionnaire;By described first
User's fills in a part filled in result of result as multiple users.
In another specific embodiment, includes multiple options in the questionnaire, the multiple user is carried out
Grouping includes: that the certain customers of the same option will be selected to be classified as one group in the multiple user.
According to the embodiment of second aspect, a kind of determining device of service plan is provided, described device is integrated in service
End, described device includes: acquiring unit, is configured to the triggering in response to active user to the first scene in client service, obtains
Take the user characteristics of the active user;Predicting unit is configured to input the user characteristics in disaggregated model, described in prediction
Current group belonging to active user;Wherein group result of the disaggregated model based on multiple users and train in advance, it is described
The group result of multiple users is determined based on the filling in result of questionnaire relevant to the client service;It determines single
Member is configured to the mapping relations of predetermined group and service plan, determines the corresponding current clothes of the current group
Business scheme.
In one embodiment, further includes: push unit is configured to current service scheme described in client push, with
So that the client is based on the current service scheme and shows corresponding service interface to the active user.
In one embodiment, the user characteristics include user property feature, and/or, for the client service
Interaction feature;Wherein the user property feature includes gender, age, occupation, marriage and childbirth situation, at least one in hobby
Kind, the interaction feature includes that number of clicks uses at least one of duration, place to use, sharing number.
In one embodiment, the disaggregated model is based on following steps and trains in advance: obtaining the multiple user's
It fills in as a result, and obtaining the user characteristics of each user in the multiple user;Filling in based on the multiple user as a result,
The multiple user is grouped;The user characteristics of result and each user based on the grouping, determine multiple instructions
Practice sample;Wherein each training sample includes the user characteristics and sample label of each user, described in the sample label instruction
Group belonging to each user;Utilize the multiple training sample training disaggregated model.
Further, in a specific embodiment, the multiple user includes the first user;Obtain the multiple use
The result of filling at family includes: the triggering in response to first user to the second scene in the client service, Xiang Suoshu
One user pushes the questionnaire;It receives first user and fills in result for the questionnaire;By described first
User's fills in a part filled in result of result as multiple users.
In another specific embodiment, includes multiple options in the questionnaire, the multiple user is carried out
Grouping includes: that the certain customers of the same option will be selected to be classified as one group in the multiple user.
According to the embodiment of the third aspect, a kind of computer readable storage medium is provided, is stored thereon with computer journey
Sequence, when the computer program executes in a computer, enable computer execute first aspect method.
According to the embodiment of fourth aspect, a kind of calculating equipment is provided, including memory and processor, feature exist
In being stored with executable code in the memory, when the processor executes the executable code, realize first aspect
Method.
In the determination method for the service plan that this specification embodiment provides, firstly, in response to active user to client
The triggering of the first scene, obtains the user characteristics of active user in the service of end;Then, user characteristics are inputted in disaggregated model,
Predict current group belonging to active user;Then, the mapping relations based on predetermined group and service plan, determination are worked as
The corresponding current service scheme of preceding group.In this way, thousand people, thousand face of service plan may be implemented, meet the needs of different user,
And then improve user experience.
Specific embodiment
With reference to the accompanying drawing, the multiple embodiments disclosed this specification are described.
This specification embodiment provides a kind of determination method of service plan, first below to the inventive concept of the method
It introduces, specific as follows:
Under normal conditions, one or more services can be provided in client, be that comparison is flexible herein for defining for service
's.It in one embodiment, may include multiple sub- APP in client, accordingly it is considered that every sub- APP is corresponding to provide one
Service.It in one example, include that multiple sub- APP, the correspondence such as " life payment ", " credit card repayment " mention in Alipay client
For multinomial services such as " life paying service " and " credit card repayment services ".It in another embodiment, can be by client energy
The multiple functions enough realized are equivalent to multinomial service herein.In one example, client can provide login service, payment
Service and transfer services etc..In another example, drop, which drips client of calling a taxi, can provide " the generation service of driving " and " about vehicle service "
Deng.It should be noted that following be referred to as client service for " a certain service provided in client ".
Further, user will generate multiple service scenarios during using client service.It is specific at one
In embodiment, user is during using login service, it is possible to create following service scenarios: " automated log on success " and " defeated
Log-on message mistake entered " etc..In another specific embodiment, user, may during use " about vehicle service "
Generate following service scenarios: " waiting driver's order ", " user cancels about vehicle " and " about Che Chenggong " etc..
Inventors have found that in certain service scenarios, that there are customer churns is excessively high, is difficult to optimize, interacts the problems such as abnormal.
For example, for the multiple users for the payment page for entering " life paying service ", wherein triggering " user does not complete payment " scene
User's accounting be 50%, that is to say, that in the payment scene of " life paying service ", churn rate is high.At this point, service
There is an urgent need to understand user to adjust to the sense of reality and opinion of these service scenarios, and then to corresponding client service for side
Whole or optimization.
In this regard, inventor proposes, the true idea of user can be understood by way of launching questionnaire.Further
The user for providing identical answer is classified as one group, and carry out to the user in each group special according to the questionnaire collected by ground
Sign is extracted, and then train classification models, can predict group belonging to user that other do not fill in questionnaires by this model, also
It is the answer provided for questionnaire.Furthermore, it is possible to different service plans be customized for different groups of other users, to meet
Different user makes each user that can have good user experience the different demands of client service.Next, to this
The application scenarios of the determination method for the service plan that specification embodiment discloses are introduced, specific as follows:
As shown in Figure 1, user can be by terminal, such as mobile phone, computer, using client service, in response to client
The triggering of a certain scene (to the introduction of scene in specifically seeing below) in the service of end, the user that server-side obtains this user are special
Sign is then based on disaggregated model trained in advance, determines group belonging to this user, then corresponding with this group by what is customized in advance
Service plan be pushed to client.Above-mentioned steps are introduced below with reference to specific embodiment.
Method flow diagram is determined Fig. 2 shows the service plan according to one embodiment, and the executing subject of the method can be with
For the server with processing capacity, device or system, for example, server-side shown in Fig. 1.As shown in Fig. 2, this method process
The following steps are included: step S210, the triggering in response to active user to the first scene in client service obtain active user
User characteristics;Step S220 inputs user characteristics in disaggregated model, predicts current group belonging to active user;Step
S230, the mapping relations based on predetermined group and service plan determine the corresponding current service scheme of current group.With
Upper step is specific as follows:
Firstly, in step S210, triggering in response to active user to the first scene in client service obtains current use
The user characteristics at family.
It is mentioned above, user will generate multiple service scenarios during using client service.Herein first
Scene is one of multiple service scenarios, can specifically be preset according to the actual situation by staff.Have at one
In body embodiment, client service is client login service, and correspondingly, the first scene can be set to " enter client
Login interface ".Correspondingly, the first scene can trigger the opening operation of client by active user.In one example,
Wherein opening operation may include: the clicking operation to client icon in terminal, alternatively, input and " opening visitor in the terminal
The corresponding acoustic control instruction in family end ".In another specific embodiment, client service is life paying service, correspondingly, the
One scene can be set to " entering the payment interface in life paying service ".Correspondingly, the first scene can be by currently using
The opening operation of payment interface is triggered at family.
The user of triggering in response to active user to above preset first scene, available active user is special
Sign.In one embodiment, wherein user characteristics include user property feature.Correspondingly, obtaining user characteristics may include: to obtain
Personal information or the identity information etc. of active user are taken, and therefrom extracts user property feature.In a specific embodiment,
The data filled in when available user's registration client.In a specific embodiment, the user property feature of extraction can
To include: gender, age, occupation, marriage and childbirth situation, hobby etc..In another embodiment, wherein user characteristics include using
Family is directed to the interaction feature of client service.Correspondingly, obtaining user characteristics may include: to obtain active user for client
The historical behavior data of service, and therefrom extract interaction feature.In a specific embodiment, the interaction feature of extraction can be with
Including number of clicks, use duration, place to use, sharing number etc..It on the other hand, in one embodiment, can also be directly
The user characteristics of active user are obtained from user characteristics library.In a specific embodiment, it may be predetermined that Duo Geke
The user characteristics of family end subscriber, and store it in user characteristics library.In this way, on the one hand can accelerate to obtain active user's
The process of user characteristics, on the other hand can be to avoid computing repeatedly, and then saves computing resource.
More than, the user characteristics of available active user.Then, in step S220, user characteristics are inputted into classification mould
In type, current group belonging to active user is predicted.
The training process of wherein disaggregated model is introduced first below.As shown in figure 3, disaggregated model can based on
Lower step is trained in advance: step S31, obtains multiple users filling in as a result, and the multiple use of acquisition for questionnaire
The user characteristics of each user in family;Step S32, filling in based on multiple users is as a result, be grouped multiple users;Step
S33, the user characteristics of packet-based result and each user determine multiple training samples;Step S34 utilizes multiple training
Sample training disaggregated model.It is specific as follows:
Firstly, in step S31, multiple users filling in as a result, and obtaining wherein each use for questionnaire is obtained
The user characteristics at family.
As previously mentioned, in certain service scenarios, that there are customer churns is excessively high, is difficult to the problems such as optimizing.In this regard, service side
Urgently want to know about these problems Producing reason.Correspondingly, service side can ask for the investigation of these problems Scenario Design
Volume.Design for questionnaire, in one embodiment, a questionnaire can for one in client service or
Multiple problem scenes.In another embodiment, questionnaire may include one or more problems.In a specific implementation
In example, for each problem therein, it can be set for multiple options of user's selection or input custom content for user
Input frame.
Further, it for the dispensing of questionnaire, in one embodiment, can use the part into client at random
Launch these questionnaires in family.It in another embodiment, can be unified to the user for triggering problem scenes at the scheduled time
Provide questionnaire.In yet another embodiment, can triggering in response to user to the second scene, provide and investigate to this user
Questionnaire.It should be noted that the second scene and problem scenes above-mentioned, the first scene are all the fields under same client service
Scape, both any between this three can be identical, can not also be identical, specifically can by staff according to the actual situation into
Row setting.In this way, questionnaire is provided to the user in the second scene, it is available more immediately, effectively, really to use
Family feedback.
According to a specific embodiment, " payment scene " (problem scenes) of service side's discovery in " life paying service "
In, customer churn is excessively high.Correspondingly, the questionnaire for " payment scene " can be designed.In one example, questionnaire
As shown in figure 4, including problem " may I ask your paying service of wishing to live and carry out the improvement of following which aspect? ", Yi Jigu
Determine option: " A., which wishes to pay the fees with query history, to be recorded ", " B. wish can push payment alert notification ", " C. wishes can be with
Push payment is notified to account " and customized input field.It is possible to further at a time to the user for entering payment scene
Unified payment questionnaire.
According to another specific embodiment, " logging in the scene " (problem of service side's discovery in " client login service "
Scene) in, after entering login interface there are certain customers, the problem of without operating.Correspondingly, it can design to be directed to and " step on
The questionnaire of record scene ".In one example, questionnaire is as shown in figure 5, " detect that you are logging in including problem
Interface stops for a long time, may I ask you and encounters following problems? " and fixed options: " A. do not know how using ", " B. worries
It is dangerous " and " C. does not know about this application ".It is possible to further the triggering in response to user to the second scene, to for launching
Questionnaire.It in one example, can be that " after entering login interface, client does not receive in 10s by the second scene settings
Any instruction ".
More than, questionnaire can be designed and launch, and then acquire multiple users and fill in result for questionnaire.?
It include a problem, and multiple fixed options for this problem in questionnaire in one embodiment.Correspondingly, some
User may include its at least one selected option for filling in result for the questionnaire.In another embodiment, it adjusts
Interrogate includes a problem, and input frame for this problem in volume.It correspondingly, can be with for filling in result for the questionnaire
The content of text inputted in input frame including user.In one example, for questionnaire shown in Fig. 5, acquisition
Filling in result includes: user to the selection of " other " item and custom content " page generation Caton ".
On the other hand, in available the multiple user each user user characteristics.It should be noted that user
The acquisition of feature may refer to associated description above, and therefore not to repeat here.
More than, available multiple users' fills in result and user characteristics.Then, in step S32, multiple users are based on
Fill in as a result, being grouped to multiple users.
In one embodiment, result and scheduled rule of classification are filled in based on multiple users, multiple users is carried out
Grouping.It should be noted that wherein scheduled rule of classification can the content based on questionnaire and set.It is specific at one
In embodiment, it is assumed that include a problem, and multiple fixed options for this problem, and scheduled grouping in questionnaire
Rule is that the user for having the same option will be selected to be classified as one group in multiple users, in this way, being directed to multiple fixations based on multiple users
Multiple users can be divided into multiple groups corresponding with multiple fixed options by the selection result of option.It is specific at another
In embodiment, it is assumed that include a problem, and input frame for this problem, and scheduled rule of classification in questionnaire
One group to be classified as by consistent user is intended in multiple users, in this way, can first cluster to the text of multiple users input
On the one hand processing can be determined that wherein the corresponding user of each class cluster is intended to, separately based on obtained multiple class clusters by staff
On the one hand result can be will fill in it is located at the user of same class cluster and is classified as one group.In another specific embodiment, it is assumed that adjust
Interrogating in volume includes multiple problems, and for multiple fixed options of wherein each problem, and scheduled rule of classification is, will
The completely the same user of result is filled in multiple users is classified as one group.
According to a specific example, for the questionnaire shown in Fig. 5, selection " its in result is filled in firstly, obtaining
He " multiple content of text corresponding to item, and this multiple content of text is clustered, it is assumed that obtain two class clusters, and work
Make personnel and this two problems corresponding to class cluster are identified as " page Caton " and " forgetting login account ".It is based on three as a result,
A fixed options and it is determined here that two problems, can based on multiple users to the selection of wherein fixed options or
The text inputted under " other " item is assigned in corresponding user's group, and then obtains 5 user groups.
More than, multiple users can be grouped, that is, determine the group of wherein each user.Then, in step
The user characteristics of each user, can determine multiple training samples in S33, packet-based result and multiple users.
It is understood that training sample includes sample characteristics and sample label.Correspondingly, in one embodiment, may be used
The user characteristics of each user as sample characteristics, are determined corresponding sample label, and then structure based on the group of the user
Build training sample.Thus, it is possible to determine multiple training samples corresponding with multiple users.
More than, after determining multiple training samples, then in step S34, utilize the training point of multiple training samples
Class model.It is based on above step S31- step S34 as a result, the training to disaggregated model can be completed.
Further, it by the above disaggregated model trained in advance of user characteristics input of active user, can predict
Current group belonging to active user out.
More than, after determining the current group of active user, then in step S230, it is based on predetermined group
With the mapping relations of service plan, the corresponding current service scheme of current group is determined.
The foundation of wherein mapping relations is introduced first below.From the foregoing, it can be understood that the design and dispensing of questionnaire
For determining that user is directed to the sense of reality and demand of problem scenes, and organizing other division that can embody the different of different user needs
It asks.In this way, it can be directed to the corresponding service plan of one of the various Demand Designs after a variety of demands for understanding user, from
And corresponding service plan is provided respectively to each group of other user is belonged to, accomplish before thousand people, and then improve user experience.
In one example, for the questionnaire shown in Fig. 5, for use corresponding to option " do not know how using "
Family group, the service plan of design can be " director information of display operation in login interface ", and option " is worried uneasy
The service plan of user's group corresponding to entirely ", design can be " safe drumbeating content is shown in login interface ", for choosing
User's group corresponding to item " not knowing about this application ", the service plan of design can be " client to be shown in login interface
Recommended information ", etc..Thus, it is possible to set up the mapping relations of group and service plan.
Further, based on the mapping relations above established, the corresponding current service scheme of current group can be determined.
It should be noted that after step S230, can also include: to client push current service scheme so that
Client is based on current service scheme and shows corresponding service interface to active user.
In one example, the current service scheme of push is " client recommended information is shown in login interface ", phase
Ying Di, service interface shown in client can be as shown in Figure 6.In this way, client recommended information is shown in login interface, it can
To assist user to determine whether to register this client, use.
In summary, method is determined using the service plan that this specification embodiment provides, firstly, in response to active user
Triggering to the first scene in client service obtains the user characteristics of active user;Then, user characteristics are inputted into classification mould
In type, current group belonging to active user is predicted;Then, the mapping relations based on predetermined group and service plan,
Determine the corresponding current service scheme of current group.In this way, thousand people, thousand face of service plan may be implemented, meet different user
Demand, and then improve user experience.
According to the embodiment of another aspect, a kind of service plan determining device is also provided, described device is integrated in server-side.
Fig. 7 shows the service plan determining device structure chart according to one embodiment, as shown in fig. 6, described device includes:
Acquiring unit 710, is configured to the triggering in response to active user to the first scene in client service, described in acquisition
The user characteristics of active user.
Predicting unit 720 is configured to input the user characteristics in disaggregated model, predict belonging to the active user
Current group;Wherein group result of the disaggregated model based on multiple users and train in advance, the grouping of the multiple user
As a result it is determined based on the filling in result of questionnaire relevant to the client service.
Determination unit 730 is configured to the mapping relations of predetermined group and service plan, determines described current
The corresponding current service scheme of group.
In one embodiment, further includes: push unit 740 is configured to current service scheme described in client push,
So that the client, which is based on the current service scheme, shows corresponding service interface to the active user.
In one embodiment, the user characteristics include user property feature, and/or, for the client service
Interaction feature;Wherein the user property feature includes gender, age, occupation, marriage and childbirth situation, at least one in hobby
Kind, the interaction feature includes that number of clicks uses at least one of duration, place to use, sharing number.
In one embodiment, the disaggregated model is based on following steps and trains in advance: obtaining the multiple user's
It fills in as a result, and obtaining the user characteristics of each user in the multiple user;Filling in based on the multiple user as a result,
The multiple user is grouped;The user characteristics of result and each user based on the grouping, determine multiple instructions
Practice sample;Wherein each training sample includes the user characteristics and sample label of each user, described in the sample label instruction
Group belonging to each user;Utilize the multiple training sample training disaggregated model.
Further, in a specific embodiment, the multiple user includes the first user;Obtain the multiple use
The result of filling at family includes: the triggering in response to first user to the second scene in the client service, Xiang Suoshu
One user pushes the questionnaire;It receives first user and fills in result for the questionnaire;By described first
User's fills in a part filled in result of result as multiple users.
In another specific embodiment, includes multiple options in the questionnaire, the multiple user is carried out
Grouping includes: that the certain customers of the same option will be selected to be classified as one group in the multiple user.
In summary, the service plan determining device provided using this specification embodiment, acquiring unit is in response to current
The triggering of the first scene, obtains the user characteristics of active user in user customer service;Predicting unit is defeated by user characteristics
Enter in disaggregated model, predicts current group belonging to active user;Determination unit is based on predetermined group and service plan
Mapping relations, determine the corresponding current service scheme of current group.In this way, thousand people, thousand face of service plan may be implemented, it is full
The demand of sufficient different user, and then improve user experience.
As above, according to the embodiment of another further aspect, a kind of computer readable storage medium is also provided, is stored thereon with calculating
Machine program enables computer execute and combines method described in Fig. 2 or Fig. 3 when the computer program executes in a computer.
According to the embodiment of another further aspect, a kind of calculating equipment, including memory and processor, the memory are also provided
In be stored with executable code, when the processor executes the executable code, realize and combine side described in Fig. 2 or Fig. 3
Method.
It will be appreciated that in said one or multiple examples, this specification discloses more those skilled in the art
A embodiment described function can be realized with hardware, software, firmware or their any combination.When use software realization
When, these functions can be stored in computer-readable medium or be referred to as the one or more on computer-readable medium
It enables or code is transmitted.
Above-described specific embodiment to the purpose of multiple embodiments of this specification disclosure, technical solution and has
Beneficial effect has been further described, it should be understood that the foregoing is merely multiple embodiments that this specification discloses
Specific embodiment, be not used to limit this specification disclose multiple embodiments protection scope, it is all in this explanation
On the basis of the technical solution for multiple embodiments that book discloses, any modification, equivalent substitution, improvement and etc. done should all be wrapped
It includes within the protection scope for multiple embodiments that this specification discloses.