CN112950392A

CN112950392A - Information display method, posterior information determination method and device and related equipment

Info

Publication number: CN112950392A
Application number: CN201911170119.2A
Authority: CN
Inventors: 汤泽; 韩沙日拉
Original assignee: Taikang Insurance Group Co Ltd
Current assignee: Taikang Insurance Group Co Ltd
Priority date: 2019-11-26
Filing date: 2019-11-26
Publication date: 2021-06-11

Abstract

The invention relates to the technical field of artificial intelligence, in particular to a posterior information determining method and device, an event information display method, a computer readable storage medium and electronic equipment for a user to trigger a target event, wherein the method comprises the steps of collecting historical characteristic information of a user to be determined, which is related to the target event, and preprocessing the historical characteristic information to obtain characteristic data; acquiring a plurality of reference characteristic paths in a preset target prediction model, wherein each reference characteristic path corresponds to a reference event trigger probability; selecting a reference characteristic path matched with the characteristic data as a target characteristic path according to the characteristic data; and determining the reference event triggering probability corresponding to the target characteristic path as posterior information of the target event triggered by the user to be determined. Compared with the prior art, the technical scheme of the embodiment of the disclosure improves the accuracy of determining the posterior information of the user trigger target event.

Description

Information display method, posterior information determination method and device and related equipment

Technical Field

The present disclosure relates to the technical field of artificial intelligence, and in particular, to a posterior information determination method and apparatus for a user-triggered target event, an event information display method, a computer-readable storage medium, and an electronic device.

Background

In many industries, with the development of data mining technology, posterior information of a user trigger target event is confirmed to adjust or operate related services, so that things which are interested by the user can be effectively presented to the user, and unnecessary operations can be reduced; for example, attribute information of the user about the insurance is determined in the insurance industry to determine whether the user is interested in the insurance, and then whether to promote the insurance.

However, the prior art posterior information determining method for a user triggered target event has insufficient accuracy in determining posterior information of the user triggered target event, which may cause many unnecessary adjustments or operations to be performed on related services, and sometimes may cause great loss.

Therefore, it is necessary to provide a new posterior information determination method and apparatus for user triggered target events, an event information presentation method, a computer readable storage medium, and an electronic device.

It is to be noted that the information disclosed in the above background section is only for enhancement of understanding of the background of the present disclosure, and thus may include information that does not constitute prior art known to those of ordinary skill in the art.

Disclosure of Invention

The present disclosure is directed to a method and an apparatus for determining posterior information of a user-triggered target event, an event information displaying method, a computer-readable storage medium, and an electronic device, so as to overcome, at least to some extent, the deficiency that the accuracy of predicting event triggering is insufficient due to the limitations and defects of the related art, which may result in many unnecessary adjustments or operations on related services, and sometimes cause a large loss.

Additional features and advantages of the disclosure will be set forth in the detailed description which follows, or in part will be obvious from the description, or may be learned by practice of the disclosure.

According to a first aspect of the present disclosure, a method for determining posterior information of a user-triggered target event includes:

acquiring historical characteristic information of a user to be determined, which is related to a target event, and preprocessing the historical characteristic information to obtain characteristic data;

acquiring a plurality of reference characteristic paths in a preset target prediction model, wherein each reference characteristic path corresponds to a reference event trigger probability;

selecting a reference characteristic path matched with the characteristic data as a target characteristic path according to the characteristic data;

and determining the reference event triggering probability corresponding to the target characteristic path as posterior information of the target event triggered by the user to be determined.

In an exemplary embodiment of the present disclosure, preprocessing the historical feature information to obtain feature data includes:

performing data cleaning on the historical characteristic information;

and carrying out discretization processing on the historical characteristic information data after data cleaning to obtain characteristic data.

In an exemplary embodiment of the present disclosure, the method further comprises:

and acquiring a target prediction model between the characteristic data and the event triggering probability.

In an exemplary embodiment of the present disclosure, obtaining a target prediction model between the feature data and the event trigger probability includes:

acquiring training data, wherein the training data comprises a plurality of groups of characteristic data and event trigger states corresponding to the characteristic data one by one;

and obtaining the target prediction model by utilizing machine learning based on the training data.

In an exemplary embodiment of the present disclosure, the obtaining the prediction model by using machine learning based on the training data includes:

dividing a plurality of groups of training data into a training sample set and a test sample set through data sampling;

training an initial prediction model based on the training sample set to obtain a reference prediction model;

and adjusting the reference prediction model based on the test sample set to obtain the target prediction model.

In an exemplary embodiment of the disclosure, the initial prediction model is an ID3 decision tree, the feature data includes a plurality of feature attributes, and training the initial prediction model based on the training sample set to obtain a reference prediction model includes:

calculating the information entropy of each training sample set;

calculating the information gain of each characteristic attribute according to the information entropy based on the event trigger state;

and selecting the characteristic attribute which is not used for division and has the maximum information gain as a division standard at each node of the decision tree until the training is completed to obtain the reference prediction model.

According to one aspect of the present disclosure, there is provided an information presentation method, including:

any one of the above posterior information determination methods for the user triggered target event;

and when the triggering probability of the target event is greater than or equal to a preset value, displaying the information of the target event to the user.

According to an aspect of the present disclosure, there is provided an posterior information determination apparatus for a user-triggered target event, including:

the acquisition module is used for acquiring historical characteristic information of a user to be determined, which is related to a target event, and preprocessing the historical characteristic information to obtain characteristic data;

the device comprises an acquisition module, a processing module and a processing module, wherein the acquisition module is used for acquiring a plurality of reference characteristic paths in a preset target prediction model, and each reference characteristic path corresponds to a reference event trigger probability;

the selection module is used for selecting a reference characteristic path matched with the characteristic data as a target characteristic path according to the characteristic data;

and the determining module is used for determining the reference event triggering probability corresponding to the target characteristic path as the posterior information determination of the target event triggered by the user to be determined.

According to an aspect of the present disclosure, there is provided a computer readable storage medium, having stored thereon a computer program which, when executed by a processor, implements a posterior information determination method for a user triggered target event as described in any one of the above.

According to an aspect of the present disclosure, there is provided an electronic device including:

a processor; and

a memory for storing one or more programs that, when executed by the one or more processors, cause the one or more processors to implement a method for a posterior information determination of a user triggered target event as described in any of the above.

The technical scheme provided by the embodiment of the disclosure can have the following beneficial effects:

in the posterior information determining method for the user-triggered target event provided by the embodiment of the present disclosure, after extracting feature data from historical feature information of a user to be determined, the historical feature information being related to the target event, posterior information of the user-triggered target event is determined according to the feature data and a feature path selected from a model to be predicted and matched with the feature data; on one hand, the historical characteristic information of the user to be determined is preprocessed to obtain characteristic data, the previous characteristic information of the user to be determined is applied to the posterior information for confirming the user trigger target event, the behavior of the user is predicted according to the behavior habit of the user, the accuracy of determining the posterior information of the user trigger target event can be improved, and unnecessary adjustment or operation on related services is avoided; on the other hand, the feature data of the user to be determined is matched with the feature branch in the model to be predicted, the posterior information of the user trigger target event is determined by using the matching result, and the accuracy of determining the posterior information of the user trigger target event is further improved through the corresponding matching relation. Furthermore, the behavior action of the user is quantified by utilizing the determined posterior information of the user trigger target event, the subsequent behavior of the user is predicted and judged by utilizing the posterior information of the user trigger target event, the loss of the user can be reduced, and the user can be facilitated to trigger the event which is interested by the user.

It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the disclosure.

Drawings

The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present disclosure and together with the description, serve to explain the principles of the disclosure. It is to be understood that the drawings in the following description are merely exemplary of the disclosure, and that other drawings may be derived from those drawings by one of ordinary skill in the art without the exercise of inventive faculty. In the drawings:

FIG. 1 is a flowchart of a method for determining posterior information for a user-triggered target event in an exemplary embodiment of the present disclosure;

FIG. 2 is a flow diagram of pre-processing historical feature information in an exemplary embodiment of the present disclosure;

FIG. 3 is a schematic diagram of a simplified decision tree in an exemplary embodiment of the present disclosure;

FIG. 4 is a flow chart of obtaining a target prediction model in an exemplary embodiment of the present disclosure;

FIG. 5 is a flow chart of deriving the predictive model using machine learning in an exemplary embodiment of the present disclosure;

FIG. 6 is a flow chart of training an initial predictive model to obtain a reference predictive model in an exemplary embodiment of the present disclosure;

FIG. 7 is a graphical illustration of training data results in an exemplary embodiment of the present disclosure;

FIG. 8 is a schematic diagram of node partitioning in an exemplary embodiment of the present disclosure;

FIG. 9 is a schematic diagram of a decision tree with nodes partitioned based only on educational conditions in an exemplary embodiment of the present disclosure;

FIG. 10 is a flow chart of the beginning to the end of an insurance application in an exemplary embodiment of the disclosure;

FIG. 11 is a flow chart of training a predictive model in an exemplary embodiment of the disclosure;

FIG. 12 is a diagram illustrating an exemplary system architecture to which a method for posterior information determination of user-triggered target events according to an embodiment of the present disclosure may be applied;

FIG. 13 is a schematic diagram illustrating a component of an event-triggered probability prediction apparatus in an exemplary embodiment of the present disclosure;

FIG. 14 schematically illustrates a structural diagram of a computer system suitable for use with an electronic device that implements an exemplary embodiment of the present disclosure;

fig. 15 schematically illustrates a schematic diagram of a computer-readable storage medium, according to some embodiments of the present disclosure.

Detailed Description

Example embodiments will now be described more fully with reference to the accompanying drawings. Example embodiments may, however, be embodied in many different forms and should not be construed as limited to the examples set forth herein; rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the concept of example embodiments to those skilled in the art. The described features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.

Furthermore, the drawings are merely schematic illustrations of the present disclosure and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and thus their repetitive description will be omitted. Some of the block diagrams shown in the figures are functional entities and do not necessarily correspond to physically or logically separate entities. These functional entities may be implemented in the form of software, or in one or more hardware modules or integrated circuits, or in different networks and/or processor devices and/or microcontroller devices.

The event occurrence probability prediction method is very important for predicting event occurrence probability in various services, for example, in the insurance industry, the later-stage behaviors of users can be predicted, insurance can be well popularized, and loss of the users is reduced.

Insurance is a popularization business, an important work of an insurance company is to popularize customers, the insurance popularization is important to the normal operation of the insurance company, one part is invested, one part is ensured, and multiple parts are ensured for the customers, so that multiple guarantees of major disease insurance money, specific disease insurance money, high-body disability insurance and personal accident insurance money can be obtained. Both the design of the premium and policy require a more detailed analysis of the promotion.

When the data mining tool is used for risk analysis, a decision-making method can be used for searching a larger field in the insurance policy which is to be popularized on the basis of the insurance policy and the claim information database established by the insurance company, so that the work of the insurance company is guided.

With the rapid development of the economic globalization and insurance industry, the insurance industry faces the huge impact of world insurance, which makes the competition for customers between insurance companies very violent, if the customer can be predicted and corresponding measures can be taken. With the development of data mining technology, many people find that data mining has become very important, and application research aiming at the data mining technology is becoming wider and wider, wherein the analysis of client factors of the insurance industry becomes a big hotspot.

In the related art, the development mode and the development condition of the Chinese insurance industry, the past development profile and the current application condition of the data mining technology are simply introduced, and the feasibility and the necessity of the data mining technology in the client prediction of the insurance industry are analyzed.

The classifier is used for carrying out data mining on the medical insurance data, so that some practical insurance promotion rules are obtained, necessary dangerous variety promotion schemes are made, and the prospect and the actual effect of the data mining method in insurance business application are discussed to be more important. In the related art, after the data acquisition is completed, the acquired basic data is selected and processed to form data more suitable for data mining by data cleaning and data transformation, and then the data is loaded into a database to prepare for the next step of establishing operation analysis. And determining the adopted mining algorithm according to the service model and the service data, repeatedly and iteratively operating the acquired data, and generating a corresponding result for analyzing and predicting. Thereby obtaining the features of the trending customer. However, in the related art, the prediction precision of whether the user purchases insurance is low, and the business direction cannot be adjusted and operated accurately.

In the exemplary embodiment, a posterior information determination method for a user triggering target event is provided first, and may be applied to prediction of an event occurrence probability in various services, for example, prediction of a probability that a user purchases a new insurance in an insurance service to be able to recommend the new insurance to a user who may purchase the new insurance in time. Referring to fig. 1, the method for determining posterior information of a user-triggered target event may include the following steps:

step S110, collecting historical characteristic information of a user to be determined, which is related to a target event, and preprocessing the historical characteristic information to obtain characteristic data;

step S120, a plurality of reference characteristic paths in a preset target prediction model are obtained, wherein each reference characteristic path corresponds to a reference event trigger probability;

step S130, selecting a reference characteristic path matched with the characteristic data as a target characteristic path according to the characteristic data;

step S140, determining the reference event triggering probability corresponding to the target feature path as posterior information of the target event triggered by the user to be determined.

After characteristic data are extracted from historical characteristic information of a user to be determined, which is related to a target event, posterior information of the user triggering the target event is determined according to the characteristic data and a characteristic path which is selected from a model to be predicted and is matched with the characteristic data; on one hand, the historical characteristic information of the user to be determined is preprocessed to obtain characteristic data, the previous characteristic information of the user to be determined is applied to the posterior information for confirming the user trigger target event, the behavior of the user is predicted according to the behavior habit of the user, the accuracy of determining the posterior information of the user trigger target event can be improved, and unnecessary adjustment or operation on related services is avoided; on the other hand, the feature data of the user to be determined is matched with the feature branch in the model to be predicted, the posterior information of the user trigger target event is determined by using the matching result, and the accuracy of determining the posterior information of the user trigger target event is further improved through the corresponding matching relation. Furthermore, the behavior action of the user is quantified by utilizing the determined posterior information of the user trigger target event, the subsequent behavior of the user is predicted and judged by utilizing the posterior information of the user trigger target event, the loss of the user can be reduced, and the user can be facilitated to trigger the event which is interested by the user.

Hereinafter, the respective steps of the a posteriori information determination method for a trigger target event in the present exemplary embodiment will be described in more detail with reference to the drawings and the embodiments.

In step S110, historical feature information of the user to be determined, which is related to the target event, is collected, and the historical feature information is preprocessed to obtain feature data.

In this example embodiment, the target event may represent an event in a plurality of application scenarios, for example, whether the user purchases insurance, and when the target event is whether the user purchases insurance, the triggering target event may represent that the user purchases insurance. The target event may also be a user opening a browser, a user browsing information, and the like, and a specific form of the target event is not limited in this example embodiment.

The historical characteristic information related to the user to be determined may include personal data information of the user and historical events similar to the target event triggered by the user within a past preset time period, and environments where the user triggers the historical events, etc.

The above-mentioned preset time period may be one year, two years, or more, and is not particularly limited in the present exemplary embodiment.

Specifically, the personal data information of the user may include the age, income, family status, etc. of the user, the above-mentioned historical events may include the same keywords as the target event, for example, when the target event is the purchase of life insurance, insurance may be included in the historical events, in which case insurance is a keyword, namely, the keywords of the target event can be determined, then all the historical events triggered by the user within the past preset time period are extracted, acquiring historical events including the keywords from all historical events, and the environment when the historical events are triggered by the user, the context may include whether the historical event is a new push event when the user triggered the historical event, the service attitude of the person responsible for the historical event when the user triggered the historical event, etc., this is merely a simple example, and the environment in which the user triggers the historical event is not limited to the above example.

The historical characteristic information is collected, so that the personal data information of the user can be firstly obtained in the database, and then the historical events matched with the personal data information and the environment when the historical events are triggered are searched in the database. The manner of collecting the historical feature information may include various manners, and the manner of collecting the historical feature information is not limited in the present exemplary embodiment.

Acquiring historical characteristic information of a user to be determined, which is related to the target event, and extracting characteristic data based on the historical characteristic information, as shown in fig. 2, the method may include the following steps:

step S210; performing data cleaning on the historical characteristic information;

step S220; and carrying out discretization processing on the historical characteristic information data after data cleaning to obtain characteristic data.

The following describes steps S210 to S220 in detail:

the method includes the steps of firstly, collecting historical characteristic information of a user to be determined, wherein the historical characteristic information is related to a target event, in the present exemplary embodiment, the target event may represent events in various services, such as insurance purchase, insurance renewal and the like in insurance services, and then, preprocessing the historical characteristic information to obtain characteristic data, wherein preprocessing the historical characteristic information to obtain the characteristic data may include data cleaning of the historical characteristic information, and performing data cleaning of the historical characteristic information may include supplementing missing data fields in the historical characteristic information, merging repeated data in the historical characteristic information, performing information screening in the historical characteristic information, extracting historical characteristic information with a high event trigger probability, and the like.

The preprocessing of the historical characteristic information can also comprise performing data discretization on the historical characteristic information after data cleaning to obtain characteristic data, namely performing discretization on continuous data fields in the historical characteristic information after data cleaning.

In step S120, a plurality of reference feature paths in a preset target prediction model are obtained, where each reference feature path corresponds to a reference event trigger probability.

In this example embodiment, the prediction model may have a plurality of feature paths therein, and each feature path may correspond to a reference event trigger probability.

In this exemplary embodiment, the target prediction model may be a decision tree model or a random forest model, and the target prediction model is not specifically limited in this exemplary embodiment.

Taking a target prediction model as an example to explain, the decision tree model may have a plurality of branches, each branch may represent a feature path, and a simple example is explained below, the node division of the decision tree is performed according to the education condition of the user and the acceptance condition of the user on the promotion risk seeds, and the user purchases insurance or does not purchase insurance as output, the decision tree shown in fig. 3 has 8 feature paths, which are S310, S320, S330, S340, S350, S360, S370, and S380; each feature path corresponds to a reference event trigger probability, for example, the probability that the user purchases a new insurance in the feature branch S310 is 0, that is, the probability that the user does not purchase an insurance is 100%; in the characteristic branch S320, the probability of the user purchasing insurance is 75%, and the probability can be obtained by analyzing and calculating the data in the database, for example, when the educational condition is analyzed, it is assumed that the number of users in the subject calendar in the database is 200, the number of users in the subject calendar purchasing new insurance is 150, and the probability of the user purchasing insurance in the subject calendar is 75% by calculation. Similar processes can be adopted for the calculation of other branches; it should be noted that the probability may be calculated in other manners, and is not specifically limited in this exemplary embodiment.

In the present exemplary embodiment, obtaining a target prediction model between the sample feature data and the event trigger probability, as shown in fig. 4, includes the following steps: :

step S410, training data is obtained, wherein the training data comprises a plurality of groups of characteristic data and event trigger states corresponding to the characteristic data one by one.

And step S420, obtaining the target prediction model by machine learning based on the training data.

The following describes steps S410 to S420 in detail:

in this example embodiment, training data may be obtained first, the training data may include a plurality of sets of feature data and event trigger states corresponding to the feature data one to one, the feature data may include a plurality of feature attributes, and the event trigger states may be used to indicate whether an event is triggered, and then, based on the training data, a required target prediction model may be obtained by using machine learning.

For example, when the posterior information determining method for the user triggered target event is used in the insurance field, the collection of the training data may collect the risk information, the marketer information and the policy information in the database at first, obtain multiple groups of user personal data in the database by using the risk information, the marketer information and the policy information, and integrate the user personal data and the risk information, the marketer information and the policy information into multiple groups one by one to obtain the training data. It should be noted that there are various information acquisition manners, and the acquisition order may also include various manners, and the manner and the order of information acquisition are not specifically limited in this exemplary embodiment.

Specifically, first, the collected dangerous species information is explained, and the collected dangerous species information may be referred to table 1:

TABLE 1 dangerous species information Table

As can be seen from table 1, the risk category information may include personal insurance, production insurance health insurance, personal accidental injury insurance, life insurance, survival insurance, dual insurance, major illness insurance, responsibility insurance, public responsibility insurance, product responsibility insurance, employer responsibility insurance, occupational responsibility insurance, credit insurance, business credit insurance, export credit insurance, investment insurance, etc., and may be described by corresponding field representations.

The collected marketer information is then described, which may be referred to in table 2:

TABLE 2 marketer information

Name of field	Data type	Of significance	Whether or not it is empty
				Subsection number	Char	Number of parts	Whether or not
Salesman number	Char	Attendant numbering	Whether or not
				gender	Char	Sex	Is that
age	Char	Age (age)	Is that
				Record of formal schooling	Char	Study calendar	Is that
marriage	Char	Marriage	Is that
				rank	Char	Job level	Is that
Length of service	Char	Age of the job	Is that
				Up to three months commission	Char	Commission for a maximum of three months	Is that

Referring to table 2, marketer information may be collected in the database, and may include a marketer part number, a salesman number, a gender, an age, a scholarship, marital, job level, a working age, a last three-month commission, and the like.

Finally, the collection of policy information is explained, the policy information can be collected in a database, and the policy information can specifically refer to table 3:

TABLE 3 policy information table

As can be seen from Table 3, the policy information may include the current state of the policy, distribution number, policy number, Chinese code, service number, room number, policy type, security seed code, clerk number, etc.

In the above-mentioned policy information table, insurance information table and marketer information table, all may include related field names, in this exemplary embodiment, user personal data information related to the insurance information, policy information table and marketer information may be obtained by matching in the database according to the field names, and then the user personal data information and the insurance information, the marketer information and the insurance information may be integrated and deleted to obtain training data, and when deleting, the integrated information may be deleted according to the above-mentioned keyword of the target event, and data including the keyword in the target event may be saved as the training data. There are various ways of acquiring the training data, and the manner of acquiring the training data is not limited in this example. The training data may include a plurality of sets of feature data and event trigger states corresponding to the feature data one to one, and in this example embodiment, an event trigger state may refer to whether a user has purchased a new insurance during a past preset time period, where the new insurance is used to indicate a new insurance relative to the past preset time period.

In the present exemplary embodiment, each set of feature data includes user feature data, and as shown in fig. 5, each set of user feature data includes feature attributes, and in obtaining the required target prediction model by using machine learning, the method may include the following steps:

step S510, dividing a plurality of groups of training data into a training sample set and a test sample set through data sampling;

step S520, training an initial prediction model based on the training sample set to obtain a reference prediction model;

step S530, the reference prediction model is adjusted based on the test sample set to obtain the target prediction model.

The following describes steps S510 to S530 in detail:

in this exemplary embodiment, the training data may be randomly sampled by data to be divided into a training set and a test set, and of course, when the number of feature data in the training data is small, the data may not be sampled and all the feature data are used as a training sample set; training the initial prediction model by using the training sample set to obtain a reference prediction model, and then adjusting the reference training model by using the test set to obtain a target training model, wherein the initial training model can be an ID3 decision tree.

And the test sample set test set is used for testing the performance of the reference prediction model and adjusting the reference prediction model according to the preset requirement so as to obtain the target prediction model with higher precision.

In another example embodiment, training data may be divided into a training set, a validation set, and a test set by data sampling; the complexity of the whole model can be simplified by adding the verification set, and the training data can be divided according to the fact that the training set, the verification set and the test set respectively account for 50%, 25% and 25% of the training data.

In this exemplary embodiment, referring to fig. 6, in step S520, training the initial prediction model based on the training sample set to obtain the reference prediction model may include the following steps:

step S610, calculating the information entropy of each training sample set;

step S620, calculating the information gain of each characteristic attribute according to the information entropy based on the event trigger state;

step S630, selecting the feature attribute which is not used for partitioning and has the largest information gain as a partitioning standard at each node of the decision tree until training is completed to obtain the reference prediction model.

The above steps are explained in detail below:

firstly, in the present exemplary embodiment, the initial prediction model is an ID3 decision tree, and the ID3 decision tree is trained by using a training sample set, in the training process, the information entropy of the training sample set may be first calculated, then the information gains of a plurality of feature attributes are calculated according to the information entropy, in the node division process of the decision tree, the feature attribute with the largest information gain may be selected for division, and after the feature attribute is adopted, the division is not performed according to the feature attribute; and then, calculating the information entropy of the residual characteristic attributes, calculating the information gain of the residual characteristic attributes, selecting the characteristic attribute with the maximum information gain from the residual characteristic attributes for division, and repeating the steps until the training of the initial prediction model is completed to obtain a reference prediction model. In this example embodiment, training of the decision tree may be completed in a pre-pruning manner, or may be completed in a post-pruning manner, which is not specifically limited in this example embodiment.

For example, referring to FIG. 7, the characteristic attributes may include age, income, education status, family life cycle changes, partial purchases, quality of service, good-to-good customers, customer environmental factors, new risk needs, current value, potential value, promotional risks, and output as to whether the user has purchased insurance.

Training the decision tree according to the training data described in fig. 7, first calculating the information entropy of the training data, and obtaining 17 samples from fig. 7, including 10 positive examples and 7 negative examples. The entropy of the current information is then calculated as follows:

in the decision tree classification problem, the information gain is the difference between the information before and after attribute selection and division of the decision tree. The information gain of each feature attribute is calculated, and the feature attribute education status is described as an example in the present exemplary embodiment.

Referring to fig. 8, when divided in an educational condition, it can be divided into three parts.

The first portion 810 has a total of 5 samples, including 3 positive examples and 2 negative examples. The entropy of the current information is then calculated as follows:

the second portion 820 has a total of 8 samples, including 6 positive examples and 2 negative examples. The entropy of the current information is then calculated as follows:

the third portion 830 has a total of 3 samples, including 0 positive examples and 3 negative examples. The entropy of the current information is then calculated as follows:

the divided information entropy is then:

representing the conditional entropy of the sample under the condition of the feature attribute. Then the information gain brought by the finally obtained characteristic attribute is:

IG(T)＝Entropy(S)-Entropy(S|T)＝0.3100665156

referring to fig. 9, assuming that the information gain of the educational condition is the maximum, it may be first divided according to the educational condition. Then, respectively calculating the information gain of each characteristic attribute, firstly, selecting the characteristic attribute with the maximum information gain for division, and after adopting the characteristic attribute, then, not dividing according to the characteristic attribute; and then calculating the information entropy of the residual characteristic attributes, calculating the information gain of the residual characteristic attributes, selecting the characteristic attribute with the maximum information gain from the residual characteristic attributes for division, and repeating the steps until the training of the initial prediction model is completed to obtain a reference prediction model.

In step S130, according to the feature data, a reference feature path matched with the feature data is selected as a target feature path.

In this exemplary embodiment, since the feature data includes a plurality of feature attributes, and the node division of the target prediction model is determined by the feature attributes, a set of feature attributes corresponds to one reference feature branch in the target prediction model, a set of feature data corresponds to one reference feature branch, and a reference feature path matched with the feature data of the user to be determined is selected as the target reference path.

The characteristic attributes in the characteristic data may include age, income, educational status, family life cycle changes, partial purchases, quality of service, good property to customers, customer environmental factors, new risk demand, current value, potential value, promotion risk, output as to whether the user purchases insurance.

In this exemplary embodiment, the target prediction model may include a plurality of reference feature paths formed by dividing each feature attribute into nodes, and the reference feature paths including all feature attributes may be selected as the target feature paths.

Referring to fig. 3, for example, when the characteristic attribute of the user includes that the education status is master, and the user can accept promotion risk and needs to predict the probability of purchasing new insurance, the reference characteristic path S340 may be selected as the target reference path.

In step S140, the reference event triggering probability corresponding to the target feature path is determined as posterior information of the target event triggered by the user to be determined.

In an example embodiment of the present disclosure, the posterior information may be a target event triggering probability of a user triggering a target event, and a reference event triggering probability corresponding to a target feature path may be determined as a probability of the user triggering the target event to be determined, that is, a target event triggering probability, that is, a probability of the user to be tested triggering the target event.

For example, with continued reference to fig. 3, each reference feature path may correspond to a reference event trigger probability, where the reference trigger probability corresponding to the reference feature path S320 is 75%, and the probability of the corresponding trigger target event is 75% if the selected target feature path is the reference feature path S320.

For another example, referring to fig. 3, taking the triggering target event as an example of purchasing a new insurance, when the characteristic attribute includes that the education status is a major, the selected target characteristic path may be S310, and the probability that the user corresponding to the reference characteristic path S310 purchases the new insurance is 0, that is, the probability that the user triggers the target event is 0. It should be noted that the probability in the present disclosure is 0, which means that the probability of the user triggering the target event is very low, but does not mean that the target event is not possible to be triggered at all.

The posterior information determination method for the user trigger target event is further described in the following application scenario with insurance promotion.

Referring to fig. 10, the posterior information determining method for the user triggered target event may be applied to the field of insurance promotion, and a flow from the beginning to the end of insurance application is first described in detail, and referring to fig. 10, the step S1010 is first performed to apply insurance; step S1020, checking and protecting; and step S1030, underwriting. Firstly, the user applies insurance, then the insurance company and the user carry out underwriting, then the user carries out payment, and the step S1040 is carried out after payment, and the operation is effective; the payment insurance takes effect; after the validation, there are several cases, including step S1051, claim settlement; step S1052, meeting insurance; step S1053, expiration; step S1054, modifying; when step S1051 is triggered, the process returns to step S1040, and proceeds to step S1061, where payment is made, and then proceeds to step S1070, where the contract is terminated. When step S1052 is triggered, step S1061 is performed to pay, and step S1070 is performed after payment, and the contract is terminated. When step S1053 is triggered, step S1070 is performed directly, and the contract is terminated. When step S1054 is triggered, the process returns to step S1040.

When the insurance is popularized, the insurance can be classified and popularized, the insurance types can comprise property insurance, personal insurance, responsibility insurance, credit insurance and the like, and when the insurance is popularized, the insurance is classified and the profit proportion of each dangerous type is analyzed, so that dangerous type information can be collected firstly, and the collection of the dangerous type information is described in detail, and therefore, the detailed description is omitted. The user triggered insurance may then be analyzed as shown with reference to table 4:

TABLE 4 customer insurance trigger event analysis Table

Changes in the life cycle of a family	Height of
		Change in purchased security awareness customers who have purchased part of the risk	Height of
Quality of after-sale service	In
		Changing clients with improved financial status	Height of
Environmental factors of the customer	Height of
		Demand for new dangerous species	In

From table 4, it can be obtained that the influence of the change of the family life cycle, the change of the purchased guarantee knowledge and the purchased part of dangerous species, the change of the financial status improvement and the customer environmental factors on the user purchasing insurance is higher, i.e. the probability of successful popularization is higher; the quality of after-market service and the need for new dangerous species have a moderate impact on the customer's insurance purchases, i.e., the likelihood of successful promotion is moderate. Different triggering events can show that the popularization risk probability is different, and certain measures or strategies are adopted for popularization when part of attributes belong to higher attributes, such as customer care, customer preferential activities and the like. Therefore, the data also need to be analyzed when the user needs to purchase insurance, and the historical characteristic information of the user can be collected to analyze the data, and the detailed description of the historical characteristic information is omitted for the sake of brevity.

Historical characteristic information of the user can be obtained from the database through the information of the dangerous case, the information of the marketer and the policy information, and the historical characteristic information analyzes, compares and screens the change of the life cycle of the family, the change of the purchased guarantee knowledge, the customers who purchase part of the dangerous case, the improvement of the financial condition, the quality of after-sale service, the requirement of new dangerous cases and the like.

Specifically, referring to fig. 11, first, step S1111 is performed to obtain dangerous seed information; step S1112, marketer information; step S1113, insurance policy information; step S1120, dangerous seed information; specifically, the characteristic information of the user is obtained from the database according to the dangerous case information, the marketer information and the policy information.

Then, step S1130 is performed, and the feature information is preprocessed; step S1140, discretizing data; and step S1150, dangerous seed information. Finally, step S1160 is carried out, and data are sampled; step S1172, training a sample set; step S1174, generating a prediction model; step S1180, testing a sample set; step S1190, model evaluation; specifically, the history feature information preprocessing and discretization processing, and the preprocessing and discretization have been described in detail recently, and therefore, the details are not repeated herein; and then, data sampling can be carried out, the initial prediction model is trained to generate a reference prediction model, and then the reference prediction model can be adjusted according to the test sample set to obtain a target prediction sample and finish evaluation.

The following describes the application of the posterior information confirmation method for user triggered target events to the insurance field in detail with reference to the system architecture of an exemplary application environment of the posterior information confirmation method for user triggered target events.

Fig. 12 is a schematic diagram illustrating a system architecture of an exemplary application environment to which a posterior information confirmation method of a user-triggered target event according to an embodiment of the present disclosure may be applied.

As shown in fig. 12, the system architecture 1200 may include a terminal device 1201, a server 1202, and a database 1203. The terminal device 1201 may be a variety of electronic devices having a display screen including, but not limited to, desktop computers, laptop computers, smart phones, tablet computers, and the like. It should be understood that the number of terminal devices, databases, and servers in fig. 1 are merely illustrative. There may be any number of terminal devices, databases, and servers, as desired for implementation. For example, the server 1202 may be a server cluster composed of a plurality of servers.

The posterior information determining method for the user triggered target event provided by the embodiment of the present disclosure is generally executed by the server 1202, but it is easily understood by those skilled in the art that the posterior information determining method for the user triggered target event provided by the embodiment of the present disclosure may also be executed by the terminal device 1201. This is not particularly limited in the present exemplary embodiment. For example, in an exemplary embodiment, a posterior information confirmation instruction of the user triggering the target event may be issued through the terminal device 1201, where the instruction may be a confirmation probability that the user to be determined purchases insurance, or may be another instruction, which is not specifically limited in this exemplary embodiment, and the instruction is uploaded to the server 1205, where the instruction includes identification information of the user to be determined, the server 1202 responds to the instruction, obtains historical feature information about the user to be determined from the database 1203 according to the identification information, performs data cleaning on the historical feature information to obtain feature data, then determines the target event triggering probability that the user triggers the target event by using a plurality of reference feature paths of a target prediction model trained in advance in the server 1202, specifically, the server 1202 may select a reference feature path matching the feature data as the target feature path by using the feature data, and determining a reference event trigger probability corresponding to the target feature path as a target event trigger probability for triggering a target event by a user to be determined, where the target event trigger probability may be a probability for purchasing insurance by the target user, and then feeding back the target event trigger probability to the terminal device 1201 by the server 1202.

Specifically, in the present exemplary embodiment, the server 1202 may first obtain the historical feature information of the user from the database 1203, where the database 1203 includes insurance purchased by the user, the physical condition of the user when the insurance is purchased, the family condition, the service attitude of the user, and the like. The historical characteristic information of the user may include personal data information of the user, historical events similar to the target event triggered by the user within a past preset time period, environments where the user triggers the historical events, and the like, where the historical events may be insurance similar to the target insurance purchased by the user within a preset time period, and the environments where the user triggers the historical events may be environments where the user purchases the insurance, and the like, for example, a service attitude of a receptionist, a financial status of the user, and the like, and the preset time period is described in detail above, and therefore, details are not described here. The server 1202 may then perform data cleansing and discretization on the historical feature information of the user to obtain feature data.

The data cleaning of the historical characteristic information may include supplementing missing data fields in the historical characteristic information, merging duplicate data in the historical characteristic information, performing information screening on the historical characteristic information, extracting historical characteristic information with a high event trigger probability, and the like, and performing data discretization on the historical characteristic information after the data cleaning to obtain characteristic data, that is, performing discretization on continuous data fields in the historical characteristic information after the data cleaning to obtain the characteristic data of the user.

For example, if the current physical condition information of the user is missing from the historical feature information, the current physical condition of the user can be estimated by using the physical conditions of the user in the time periods before and after the current time and supplemented to the historical feature information.

In the present exemplary embodiment, the server 1202 may obtain the information of the risk category, the policy information, and the information of the marketer from the database 1203, and obtain a plurality of sets of feature data from the database according to the information of the risk category, the policy information, and the information of the marketer. The server 1202 calculates data with large influence on insurance purchase of the user by each feature information in the database 1203, the server 1202 extracts the data with large influence as feature data by adopting a data extraction technology, and simultaneously extracts an event trigger state corresponding to the feature data from the database, wherein the event trigger state can be whether the user purchases insurance or not. The feature data and the event trigger state can be integrated based on a data integration technology to obtain training data.

In this exemplary embodiment, the server 1202 may divide the training data into a training sample set and a testing sample set based on a data sampling technique, may train an initial prediction model based on the training sample set to obtain a reference prediction model, and may first calculate an information entropy of the training sample set during training; then, the information gain of each characteristic attribute can be calculated according to the information entropy by utilizing an event triggering state;

in this example embodiment, a decision tree model may be used as an initial prediction model, and a feature attribute that is not used for partitioning and has the largest information gain may be selected as a partition standard at each node of the decision tree until training is completed to obtain a reference prediction model, and then a target prediction model is obtained through testing. Then, the characteristic information of the user to be determined can be input into the target prediction model to determine the target event triggering probability of the user to be determined triggering the target event.

Further, the present disclosure also provides an information displaying method, where the information displaying method includes the method for determining posterior information of the user-triggered target event, and the method for determining posterior information of the user-triggered target event has been described in detail above, and therefore, details are not repeated here.

In this exemplary embodiment, the posterior information may be a target event triggering probability for triggering a target event by a user, and when the target event triggering probability is greater than or equal to a preset value, the information of the target event is displayed to the user, where the preset value may be 60%, or may be self-defined according to a requirement, and the preset value is not specifically limited in this exemplary embodiment.

In addition, the information of the target event to be displayed may be directly sent to the terminal of the user, or may be introduced to the user in a call form, and a method and a channel for displaying the information are not specifically limited in this exemplary embodiment.

It is noted that the above-mentioned figures are merely schematic illustrations of processes involved in methods according to exemplary embodiments of the present disclosure, and are not intended to be limiting. It will be readily understood that the processes shown in the above figures are not intended to indicate or limit the chronological order of the processes. In addition, it is also readily understood that these processes may be performed synchronously or asynchronously, e.g., in multiple modules.

The following describes embodiments of the apparatus of the present disclosure, which may be used to execute the method for determining posterior information of the user triggered target event in the present disclosure. In addition, in an exemplary embodiment of the disclosure, an event-triggered probability prediction device is also provided. Referring to fig. 13, the posterior information determining apparatus 1300 for the user triggered target event includes: an acquisition module 1310, an acquisition module 1320, a selection module 1330, and a determination module 1340.

The acquiring module 1310 may be configured to acquire historical feature information of a user to be determined, which is related to a target event, and preprocess the historical feature information to obtain feature data; the obtaining module 1320 may be configured to obtain a plurality of reference feature paths in a preset target prediction model, where each of the reference feature paths corresponds to a reference event trigger probability. A selecting module 1330, configured to select, according to the feature data, a reference feature path matched with the feature data as a target feature path; the determining module 1340 is configured to determine the reference event triggering probability corresponding to the target feature path as posterior information of triggering the target event by the user to be determined.

For details that are not disclosed in the embodiments of the apparatus of the present disclosure, please refer to the embodiments of the method for predicting the probability of event trigger described above for the details that are not disclosed in the embodiments of the apparatus of the present disclosure.

It should be noted that although in the above detailed description several modules or units of the device for action execution are mentioned, such a division is not mandatory. Indeed, the features and functionality of two or more modules or units described above may be embodied in one module or unit, according to embodiments of the present disclosure. Conversely, the features and functions of one module or unit described above may be further divided into embodiments by a plurality of modules or units.

In addition, in an exemplary embodiment of the present disclosure, an electronic device capable of implementing the above-described method for predicting the probability of event triggering is also provided.

As will be appreciated by one skilled in the art, aspects of the present disclosure may be embodied as a system, method or program product. Accordingly, various aspects of the present disclosure may be embodied in the form of: an entirely hardware embodiment, an entirely software embodiment (including firmware, microcode, etc.) or an embodiment combining hardware and software aspects that may all generally be referred to herein as a "circuit," module "or" system.

An electronic device 1400 according to such an embodiment of the present disclosure is described below with reference to fig. 14. The electronic device 1400 shown in fig. 14 is only an example and should not bring any limitations to the functionality and scope of use of the embodiments of the present disclosure.

As shown in fig. 14, the electronic device 1400 is embodied in the form of a general purpose computing device. The components of the electronic device 1400 may include, but are not limited to: the at least one processing unit 1410, the at least one memory unit 1420, the bus 1430 that connects the various system components (including the memory unit 1420 and the processing unit 1410), and the display unit 1440.

Wherein the storage unit stores program code that is executable by the processing unit 1410, such that the processing unit 1410 performs steps according to various exemplary embodiments of the present disclosure described in the "exemplary methods" section above in this specification. For example, the processing unit 1410 may perform step S110 as shown in fig. 14: acquiring historical characteristic information of a user to be determined, which is related to a target event, and preprocessing the historical characteristic information to obtain characteristic data; s140: acquiring a plurality of reference characteristic paths in a preset target prediction model, wherein each reference characteristic path corresponds to a reference event trigger probability; s140: selecting a reference characteristic path matched with the characteristic data as a target characteristic path according to the characteristic data; s140: and determining the reference event triggering probability corresponding to the target characteristic path as posterior information of the target event triggered by the user to be determined.

As another example, the electronic device may implement the steps shown in fig. 1 to 5.

The storage unit 1420 may include readable media in the form of volatile memory units, such as a random access memory unit (RAM)1421 and/or a cache memory unit 1422, and may further include a read only memory unit (ROM) 1423.

Storage unit 1420 may also include a program/utility 1424 having a set (at least one) of program modules 1425, such program modules 1425 including, but not limited to: an operating system, one or more application programs, other program modules, and program data, each of which, or some combination thereof, may comprise an implementation of a network environment.

Bus 1430 may be any type of bus structure including a memory cell bus or memory cell controller, a peripheral bus, an accelerated graphics port, a processing unit, or a local bus using any of a variety of bus architectures.

The electronic device 1400 may also communicate with one or more external devices 1470 (e.g., keyboard, pointing device, bluetooth device, etc.), with one or more devices that enable a user to interact with the electronic device 1400, and/or with any devices (e.g., router, modem, etc.) that enable the electronic device 1400 to communicate with one or more other computing devices. Such communication can occur via an input/output (I/O) interface 1450. Also, the electronic device 1400 may communicate with one or more networks (e.g., a Local Area Network (LAN), a Wide Area Network (WAN), and/or a public network, such as the internet) via the network adapter 1460. As shown, the network adapter 1460 communicates with the other modules of the electronic device 1400 via the bus 1430. It should be appreciated that although not shown in the figures, other hardware and/or software modules may be used in conjunction with the electronic device 1400, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems, among others.

Through the above description of the embodiments, those skilled in the art will readily understand that the exemplary embodiments described herein may be implemented by software, or by software in combination with necessary hardware. Therefore, the technical solution according to the embodiments of the present disclosure may be embodied in the form of a software product, which may be stored in a non-volatile storage medium (which may be a CD-ROM, a usb disk, a removable hard disk, etc.) or on a network, and includes several instructions to enable a computing device (which may be a personal computer, a server, a terminal device, or a network device, etc.) to execute the method according to the embodiments of the present disclosure.

In an exemplary embodiment of the present disclosure, there is also provided a computer-readable storage medium having stored thereon a program product capable of implementing the above-described method of the present specification. In some possible embodiments, aspects of the present disclosure may also be implemented in the form of a program product comprising program code for causing a terminal device to perform the steps according to various exemplary embodiments of the present disclosure described in the "exemplary methods" section above of this specification, when the program product is run on the terminal device.

Referring to fig. 15, a program product 1500 for implementing the above method according to an embodiment of the present disclosure is described, which may employ a portable compact disc read only memory (CD-ROM) and include program code, and may be run on a terminal device, such as a personal computer. However, the program product of the present disclosure is not limited thereto, and in this document, a readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device.

The program product may employ any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the foregoing. Examples of physical (a non-exhaustive list) of readable storage media include: an electrical connection having one or more wires, a portable disk, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

A computer readable signal medium may include a propagated data signal with readable program code embodied therein, for example, in baseband or as part of a carrier wave. Such a propagated data signal may take many forms, including, but not limited to, electro-magnetic, optical, or any suitable combination thereof. A readable signal medium may also be any readable medium that is not a readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device.

Program code embodied on a readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

Program code for carrying out operations for the present disclosure may be written in any combination of one or more programming languages, including an object oriented programming language such as Java, C + + or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computing device, partly on the user's device, as a stand-alone software package, partly on the user's computing device and partly on a remote computing device, or entirely on the remote computing device or server. In the case of a remote computing device, the remote computing device may be connected to the user computing device through any kind of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or may be connected to an external computing device (e.g., through the internet using an internet service provider).

Furthermore, the above-described figures are merely schematic illustrations of processes included in methods according to exemplary embodiments of the present disclosure, and are not intended to be limiting. It will be readily understood that the processes shown in the above figures are not intended to indicate or limit the chronological order of the processes. In addition, it is also readily understood that these processes may be performed synchronously or asynchronously, e.g., in multiple modules.

Other embodiments of the disclosure will be apparent to those skilled in the art from consideration of the specification and practice of the disclosure disclosed herein. This disclosure is intended to cover any variations, uses, or adaptations of the disclosure following, in general, the principles of the disclosure and including such departures from the present disclosure as come within known or customary practice within the art to which the disclosure pertains. It is intended that the specification and examples be considered as exemplary only, with a true scope and spirit of the disclosure being indicated by the following claims.

It will be understood that the present disclosure is not limited to the precise arrangements described above and shown in the drawings and that various modifications and changes may be made without departing from the scope thereof. The scope of the present disclosure is to be limited only by the terms of the appended claims.

Claims

1. A posterior information determination method for a user triggering target event is characterized by comprising the following steps:

2. The method of claim 1, wherein preprocessing the historical feature information to obtain feature data comprises:

performing data cleaning on the historical characteristic information;

3. The method of claim 2, further comprising:

4. The method of claim 3, wherein obtaining a target prediction model between the feature data and the event trigger probability comprises:

5. The method of claim 4, wherein deriving the predictive model using machine learning based on the training data comprises:

6. The method of claim 5, wherein the initial prediction model is an ID3 decision tree, wherein the feature data comprises a plurality of feature attributes, and wherein training the initial prediction model based on the set of training samples yields a reference prediction model, comprising:

calculating the information entropy of each training sample set;

7. An information display method, comprising:

the method for posterior information determination of a user-triggered target event according to any of claims 1 to 6;

and when the posterior information is more than or equal to a preset value, displaying the information of the target event to the user.

8. An apparatus for determining posterior information of a user-triggered target event, comprising:

and the determining module is used for determining the reference event triggering probability corresponding to the target characteristic path as the posterior information of the target event triggered by the user to be determined.

9. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, implements the a posteriori information determination method of a user triggered target event according to any one of claims 1 to 6.

10. An electronic device, comprising:

a processor; and

a memory for storing one or more programs that, when executed by the one or more processors, cause the one or more processors to implement the method for a posterior information determination of a user triggered target event as recited in any of claims 1 to 6.