JP7309892B2 - FINANCIAL PRODUCT PRESENTATION DEVICE AND PROGRAM - Google Patents

Description

The present invention relates to a financial product presentation device and program.

2. Description of the Related Art There is a computer system that receives an answer to a customer's question and determines products and services to be proposed to the customer based on the answer (see Patent Document 1, for example).
[Prior art documents]
[Patent Literature]
[Patent Document 1] Patent No. 6326538

Problem to be solved

However, the above computer system requires the customer to answer questions, and is not customer-friendly. For example, when a product or service is related to insurance, which is an example of a broadly defined finance-related business, information is already obtained from existing policyholders at the time of contracting, so it is required to utilize this information. there is

General disclosure

In order to solve the above problems, in a first aspect of the present invention, a storage unit stores information on a contractor and contract specifying information before renewal in association with contract specifying information; a reading unit for reading from the storage unit the information stored in correspondence and the information stored in association with the contract specifying information before renewal associated with the specific contract specifying information, and a contractor; A financial product presentation device comprising a presentation unit that presents financial products in response to the information read by the reading unit using a machine learning model that has learned the relationship between the information about provided.

In a second aspect of the present invention, the computer has a storage function for storing information on the contractor and the contract specifying information before renewal in a storage unit in association with the contract specifying information, and A reading function for reading stored information and information stored in association with the pre-renewal contract identification information associated with the specific contract identification information from the storage unit; A program is provided that implements a presentation function of presenting financial-related products for information read by a reading function, using a machine learning model that has learned relationships with related products.

In the third aspect of the present invention, the feature quantity of the machine learning model that learned the relationship between the information on the contractor and the contracted financial product, and the commentary explaining the feature quantity are stored in association with each other. A commentary storage section and a presentation section for presenting financial-related products to a target person by using a machine learning model for information about the target person, wherein the presentation section is used for presenting the financial-related product. Provided is a financial-related product presentation device that further presents the explanatory text stored in the explanatory text storage unit based on the feature amount obtained.

In a fourth aspect of the present invention, the computer associates the feature quantity of the machine learning model that learned the relationship between the information on the contractor and the contracted financial product, and the commentary text explaining the feature quantity. and a presentation function that presents financial products to the target by using a machine learning model for information about the target. A program is provided that further presents the commentary text stored in the commentary text storage unit based on the feature quantity used for the presentation.

It should be noted that the above summary of the invention does not list all the necessary features of the invention. Subcombinations of these feature groups can also be inventions.

1 shows functional blocks of an insurance product presentation device 10 according to the first embodiment. An example of contractor information stored in the contractor information DB 130 is shown. An example of the operation flow of the insurance product presentation device 10 is shown. An example of the operational flow of step S10 for integrating contractor information is shown. An example of contractor information and the like stored in the integrated information storage unit 100 is shown. It is an example of a decision tree that analyzes policyholders of insurance products. An example of the operation flow of step S30 for presenting an insurance product is shown. The functional block of the insurance product presentation apparatus 12 which concerns on 2nd Embodiment is shown. An example of recruitment information that is information about recruitment stored in the recruitment information DB 136 is shown. An example of the operation flow of the insurance product presentation device 12 is shown. It is an example of a decision tree that analyzes solicitation information for insurance products as explanatory variables. FIG. 11 shows functional blocks of an insurance product presentation device 14 according to a third embodiment; FIG. An example of the commentary text stored in the commentary text storage unit 150 is shown. Another example of the commentary text stored in the commentary text storage unit 150 is shown. An example of the feature quantity of another machine learning model 160 learned and reasoned by the insurance product presentation devices 10, 12, 14 is shown. An example of the feature quantity of still another machine learning model 170 learned and reasoned by the insurance product presentation devices 10, 12, 14 is shown. An example computer 2200 is shown in which aspects of the present invention may be embodied in whole or in part.

Hereinafter, the present invention will be described through embodiments of the invention, but the following embodiments do not limit the invention according to the claims. Also, not all combinations of features described in the embodiments are essential for the solution of the invention.

FIG. 1 shows functional blocks of an insurance product presentation device 10 according to the first embodiment. The insurance product presentation device 10 is, for example, a computer managed by an insurance company. The insurance product presentation device 10 communicates with a terminal 20 used by an agent solicitor through a communication network such as the Internet. The solicitor is a sales representative of an agency or the like who solicits insurance.

The insurance product presentation device 10 includes a policyholder information DB 130 that stores policyholder information about policyholders, a complaint information DB 132 that stores complaint information about complaints from policyholders, and accidents related to accidents involving policyholders. It has an accident information DB 134 that stores information. These policyholder information, complaint information, and accident information are stored in respective databases directly by employees of the insurance company or from the terminal 20 of the solicitor of the agent.

Insurance product presentation device 10 further includes integrated information storage section 100 , reading section 110 , estimation section 112 and presentation section 114 . Presenter 114 includes machine learning model 120 .

FIG. 2 shows an example of contractor information stored in the contractor information DB 130. As shown in FIG. The policyholder information in FIG. 2 exemplifies information about the policyholder of automobile insurance.

In the policyholder information DB 130, policyholder information is stored for each insurance policy number, that is, in association with the insurance policy number. Furthermore, if the insurance contract is renewed due to expiration of the previous insurance contract period, etc., the insurance policy number before renewal is stored in association with the current insurance policy number. In the example of FIG. 2, if the insurance contract is not renewed, an invalid value "*" is stored in the insurance policy number field before renewal.

Contractor information includes contract period, contractor name, contractor age, contractor gender, contractor address, vehicle type, vehicle name, and license color. These pieces of information are essential information at the time of making a contract. The car name includes information indicating whether the car is a domestic car or a foreign car. In the example of FIG. 2, in the car name feed, the car name is followed by "(country)" if it is a domestic car, and "(outside)" if it is a foreign car.

The policyholder information also includes an insurance type that distinguishes automobile insurance, fire insurance, life insurance, etc., and an insurance product ID that identifies the insurance product. In the example shown in FIG. 2, the insurance type and insurance product ID are included in the insurance policy number. The first alphabetic character of the insurance policy number represents the insurance type. For example, automobile insurance is "J", fire insurance is "K", life insurance is "S", and so on. Furthermore, the 2nd to 4th three letters of the alphabet of the insurance policy number represent the insurance product ID. Therefore, in the example of FIG. 2, "JBCD0001" and "JBCD0002" are automobile insurance and are the same insurance product, and "JBCX0195" is automobile insurance but is a different insurance product.

Contractor information also includes optional information that is not mandatory at the time of contract. Examples of arbitrary information in automobile insurance shown in FIG. 2 are marital status and whether the house is owned or rented. Arbitrary information in automobile insurance may include the body color of the car, whether or not the insured person lives with the child, the age of the child, whether or not the insured person lives with the parent, the age of the parent, and the like. Arbitrary information may be obtained by the solicitor from the contractor, or may be obtained from a questionnaire to the contractor attached to the contract. When the insurance type is fire insurance, property value information such as size, area, whether it is a corner room, distance from the station, number of floors, etc. may be included.

Although not shown, the complaint information DB 132 stores, in association with the insurance policy number, the details of complaints filed with an agent or an insurance company regarding insurance by the policyholder of the insurance policy number as complaint information. If the customer has never filed a complaint, information to that effect, for example, an invalid value "*", is stored in the relevant field. Complaint information may include information indicating categories obtained by categorizing the contents of complaints and information indicating the rank of degree of complaint.

Similarly, the accident information DB 134 stores, as accident information, details of accidents in which the policyholder of the insurance policy number was involved in association with the insurance policy number. In particular, if the contract is automobile insurance, information about accidents related to automobiles is stored, and if the contract is fire insurance, information about accidents related to fire is stored. When there is no accident, information indicating that fact, for example, an invalid value "*" is stored in the field. The item information may include information indicating categories into which the details of the accident are categorized, and information indicating the rank of the severity of the accident.

Furthermore, the insurance product presentation device 10 may store a log of calls from the policyholder to the contact center of the insurance company as incoming call information at the contact center in association with the insurance policy number. The insurance product presentation device 10 may further be connected to an external DB that stores information associated with the insurance policy number.

FIG. 3 shows an example of the operation flow of the insurance product presentation device 10. As shown in FIG. The insurance product presentation device 10 integrates policyholder information such as the policyholder information DB 130 (S10), analyzes each insurance product with a decision tree using the integrated policyholder information, etc., and creates a machine learning model (S20), An insurance product is presented to the terminal 20 using the machine learning model (S30).

FIG. 4 shows an example of the operation flow of step S10 for integrating contractor information and the like. The operation flow is started by a direct instruction to the insurance product presentation device 10 or periodically, such as quarterly.

The reading unit 110 reads the insurance policy number and the policyholder information stored in association with the insurance policy number from the policyholder information DB 130 (S101). Reading unit 110 further reads complaint information stored in complaint information DB 132 and accident information stored in accident information DB 134 in association with the same insurance policy number (same step).

The reading unit 110 determines whether or not the insurance policy number before renewal is stored in association with the insurance policy number read in step S101 (S103). If the insurance policy number before renewal is stored (S103: Yes), reading unit 110 reads the policyholder information associated with this insurance policy number before renewal from the policyholder information DB (S105). Further, reading unit 110 reads complaint information and accident information associated with this insurance policy number before renewal from complaint information DB 132 and accident information DB 134 (same step).

In the example shown in FIG. 2, the insurance policy number "JBCD0002" before renewal is stored in association with the insurance policy number "JBCD0571". Therefore, reading unit 110 reads the policyholder information, complaint information, and accident information of insurance policy number "JBCD0571" in step S103, and reads the policyholder information, complaint information, and accident information of insurance policy number "JBCD0002" before renewal in step S105. Also read information. Incoming call information at the contact center associated with the insurance policy number and external data when an external DB is connected are also read. Contractor information, complaint information, accident information, and other information relating to the contractor may be collectively referred to as contractor information and the like.

Next, the estimating unit 112 compares the policyholder information of the insurance policy number with the policyholder information of the insurance policy number before renewal, and determines whether there is any change other than the age and contract period (S107). ). Here, age and contract period are excluded because it is natural that age and contract period are different before renewal and after renewal. The age here includes not only the age of the contractor but also the ages of children and parents as arbitrary information.

If there is a change between the policyholder information of the insurance policy number and the policyholder information of the insurance policy number before renewal (S107: Yes), the estimation unit 112 estimates the life event of the policyholder. (S109). The estimated life event becomes new information about the insurance policy number.

In the example shown in FIG. 2, there is a change between the policyholder address of insurance policy number "JBCD0571" and the policyholder address of insurance policy number "JBCD0002" before renewal associated with the insurance number. Therefore, the estimation unit 112 estimates that the contractor has a life event of "moving".

The life event is estimated using a lookup table that associates the type of changed information with the life event in advance. An example of correspondence is, as described above, "change of contractor's address" and "moving". In addition, if "the number of children has increased", then "childbirth", if "the name of the car has changed", then "car replacement", and so on. Instead of or in addition to this, a life event may be estimated using a learning model obtained by machine-learning the relationship between the type of changed information and the life event.

After step S109 or when the determination in either step S103 or S107 is No, the estimation unit 112 integrates the read information and the estimated life event and stores them in the integrated information storage unit 100 (S111). As a result, information for each contractor is obtained.

FIG. 5 shows an example of contractor information and the like stored in the integrated information storage unit 100. As shown in FIG. The integrated information storage unit 100 stores the latest insurance policy number in the policyholder name and the policyholder information of the latest insurance policy number in association with the policyholder name.

Furthermore, policyholder information, etc., which was not stored in association with the latest insurance policy number but was stored in association with the insurance policy number before renewal, is also stored. In the example shown in FIG. 5, other information of owned house/rental associated with the latest insurance policy number "JBCD0570" of the policyholder name "Taro Mitsui" was not stored. However, "Rental" is stored in association with the insurance policy number "JBCD0001" before renewal of the insurance policy number "JBCD0570". Additionally, there is no change in policyholder address between these policy numbers. Therefore, the integrated information storage unit 100 stores "rental" in association with the contractor's name "Taro Mitsui".

Furthermore, the life event estimated by the estimation unit 112 is also stored in the integrated information storage unit 100 . In the example shown in FIG. 5, "moving" is stored in the life event field in response to presuming that the contractant name "Hanako Sumitomo" has a life event of "moving" as illustrated in step S109 above. ing.

Thus, step S10 for integrating information is completed. The policyholders stored in the integrated information storage unit 100 are the existing policyholders of the insurance product corresponding to each insurance policy number, and are also candidates for contracts for other insurance products. Next, the presentation unit 114 analyzes the policyholders of the insurance product using a decision tree (S20).

FIG. 6 is an example of a decision tree that analyzes policyholders of insurance products. Decision tree analysis is performed by machine learning. The decision tree constructed by the analysis becomes the machine learning model 120 that has learned the relationship between policyholder information and the like and insurance products.

The objective variable of the decision tree is the policyholders for each insurance product. In the example of FIG. 6, 200 policyholders enrolled in the insurance product "automobile insurance JBCD" are the objective variables.

Contractor information or the like stored in the integrated information storage unit 100 is used as explanatory variables. In the example of FIG. 6, "age" with "35 years old" as a threshold, "foreign car or domestic car", "whether or not the car is moved", etc. are used as explanatory variables. Of these, for example, "age" is directly stored in the contractor information DB 130, and is information taken over to the integrated information storage unit 100 as it is. On the other hand, the “presence or absence of moving” is estimated as a life event by the estimating unit 112 from the “address” information directly stored in the contractor information DB 130 and stored in the integrated information storage unit 100, so to speak, as a synthetic variable. be.

Thus, the step S20 of analyzing with the decision tree is completed. Next, the presentation unit 114 presents insurance products using this decision tree (S30).

FIG. 7 shows an example of the operational flow of step S30 for presenting an insurance product. The operation flow of step S30 is started by an instruction from the solicitor via the terminal 20. FIG.

The presentation unit 114 reads the covenanter information and the like integrated for each covenanter from the integrated information storage unit 100 (S131). The presentation unit 114 applies the policyholder information and the like to the decision tree for the specific insurance product (S133). Specific insurance products in this case exclude insurance products already contracted by the policyholder. This is because there is no need to judge the possibility of making a contract for an insurance product that has already been contracted.

That is, the policyholder information and the like of policyholder X who has contracted insurance product A is applied to the decision tree of insurance product B. This makes it possible to estimate the possibility that policyholder X will also contract insurance product B.

For example, referring to the examples of FIGS. 5 and 6, applying the decision tree of the insurance product 'JBCD' to the policyholder 'Jiro Tokyo' yields 'age over 35', 'domestic car', 'not moving'. ', it can be seen that profiles such as policyholder information match those of only 4% of existing policyholders of the insurance product "JBCD." Therefore, it is estimated that the policyholder “Tokyo Jiro” is unlikely to contract the insurance product “JBCD”.

The presentation unit 114 fits the policyholder to the decision trees of all the insurance products excluding the insurance product that the policyholder has contracted (S135). As a result, the presentation unit 114 outputs to the terminal 20 the insurance product with many policyholders whose profiles match each other as a proposed insurance product (S137). In this case, among the plurality of fitted insurance products, a predetermined number of insurance products with a high percentage of existing policyholders with matching profiles may be proposed. An insurance product with a percentage of a certain value or more, such as 50% or more, may be proposed.

The above steps S131 to S137 are executed for all contractors stored in the integrated information storage unit 100 (S139). This completes the operation flow of step S30.

As described above, according to the present embodiment, the decision tree is analyzed using the integrated information, so that machine learning can be performed by selecting appropriate explanatory variables from a larger number of explanatory variable candidates. Furthermore, since the integrated information is used to fit the decision tree, it is possible to present insurance products with a higher probability of contract.

FIG. 8 shows functional blocks of the insurance product presentation device 12 according to the second embodiment. In the insurance product presentation device 12, the same components as in the insurance product presentation device 10 are denoted by the same reference numerals, and descriptions thereof are omitted.

In addition to the configuration of insurance product presentation device 10 , insurance product presentation device 12 further includes recruitment information DB 136 and machine learning model 122 in presentation unit 116 . The presentation unit 116 of the insurance product presentation device 12 presents the information on recruitment to the terminal 20 using the machine learning model 122 that has learned the relationship between the insurance product and the information on recruitment.

FIG. 9 shows an example of recruitment information that is information about recruitment stored in the recruitment information DB 136 . In the recruitment information DB 136, in association with the solicitor ID, the solicitor name, solicitor age, solicitor gender, agency ID, skill level, insurance policy number with which the solicitor concluded a contract, and soliciting activities that contributed to the contract is stored. These pieces of information are stored in the recruitment information DB 136 directly by an employee of the insurance company or from the terminal 20 of the solicitor of the agency.

FIG. 10 shows an example of the operation flow of the insurance product presentation device 12. As shown in FIG. The operation flow of the insurance product presentation device 12 includes, in addition to the operation flow of the insurance product presentation device 10, a step in which the presentation unit 116 reads the offer information (S22), and a step of analyzing with a decision tree using the offer information. (S24), and a step of proposing recruitment using a decision tree of recruitment information (S40).

FIG. 11 is an example of a decision tree obtained by analyzing insurance product offering information as an explanatory variable. The decision tree analysis is performed by machine learning in steps S22 and S24. A decision tree constructed by the analysis becomes a machine learning model 122 that has learned the relationship between recruitment information and insurance products.

The objective variable of the decision tree is the total number of policies for each insurance product. In the example of FIG. 11, the objective variable is 500 contracts for the insurance product "automobile insurance JBCD". Here, the total number of cases is calculated based on the insurance policy number stored in the recruitment information DB 136 and contracted by the solicitor. Therefore, unlike FIG. 6, even if the same policyholder has a total of multiple contracts for the insurance product due to renewals, etc., each case may be counted as a separate number of cases.

Recruitment information stored in the recruitment information DB 136 is used as explanatory variables. Further, as explanatory variables, composite variables based on recruitment information in the recruitment information DB 136 and contractor information in the integrated information storage unit 100 may be used. In this case, the composite variable is generated from the policyholder information and solicitation information associated with the same insurance policy number. The synthesis variable may have its generation procedure stored in advance in the presentation unit 116 .

In the example of FIG. 11 , the explanatory variable of “face-to-face, Internet, or telephone” of “recruitment activity” is “recruitment activity that contributed to closing a contract” stored in the recruitment information DB 136 . On the other hand, the explanatory variable ``whether the solicitor is older or younger than the policyholder'' is the difference between the ``age of solicitor'' in the solicitation information DB 136 and the ``age of the policyholder'' in the integrated information storage unit 100 for the same insurance policy number. It is a composite variable generated by comparison. As another combined variable, even if "closeness of address" generated from "address of agent" (not shown in FIG. 11) of "contractor address and recruitment information DB 136" of integrated information storage unit 100 is used. good.

In step S40, the presentation unit 116 proposes a solicitation using the decision tree, which is the machine learning model 122 that has learned the relationship between solicitation information and insurance products. In this case, the presentation unit 116 uses a decision tree obtained by analyzing the relationship between the insurance products presented in step S30 and the solicited information. For example, when the insurance product C is presented to the policyholder Y in step S30, the decision tree of the insurance product C is used to propose a solicitation to the policyholder Y.

In this case, a predetermined number of recruitment information with a high percentage of the number of existing contracts analyzed by the decision tree may be proposed, or the percentage of the number of existing contracts is a certain value or more, for example, 50% or more. may be suggested. In the example of FIG. 11, if the recruitment information with the highest % of the number of existing contracts is proposed, the solicitor whose "age of the solicitor" is "older" than the "age of the policyholder" is selected as the solicitation activity of "face-to-face". In addition, we propose to "inform at the time of car inspection".

As described above, according to the present embodiment, in addition to the effects of the first embodiment, recruitment is proposed using the decision tree analyzed from the recruitment information, so that the possibility of making a contract can be further increased. In addition, the information possessed by insurance companies can be used more effectively.

In any of the embodiments, instead of applying the decision tree of all insurance products for each policyholder in step S30, the policyholder who fits the profile with a high percentage of existing policyholders in each decision tree may be retrieved and insurance products corresponding to the decision tree may be presented to the policyholder.

Furthermore, in any embodiment, the insurance type of the policyholder analyzed by the decision tree and the insurance type of the policyholder fitted to the decision tree may be the same or different. In the above examples, all are automobile insurance and the insurance type is the same. Among the insurance products of the same type, the insurance product with greater coverage than the current insurance product may be selected and presented from among the insurance products extracted in step S30. This allows us to propose renewals to insurance products with greater coverage.

As an example of different types of insurance, by building a decision tree with auto insurance policyholders and applying the policyholder information of fire insurance and life insurance policyholders to the decision tree, You may present your car insurance. This enables efficient overselling of insurance products.

Further, in the second embodiment, a decision tree may be constructed for each insurance type instead of constructing a decision tree for solicitation information for individual insurance products. This makes it possible to propose an appropriate solicitation for each insurance type.

Although decision trees are described as machine learning models 120 and 122 in both embodiments, other machine learning models using supervised learning may be used. Other machine learning models include support vector machines, neural networks, and deep learning.

In any of the embodiments, an example of an individual contractor has been described, but this may also be applied to cases in which the contractor is a corporation. If the contractor is a corporation, the contractor information may include information on the contractor's business partners and related companies. Information on business partners and affiliated companies may be provided by, for example, a business research firm such as Teikoku Databank. The customer information and affiliated company information may be used as explanatory variables for decision trees, or may be targets for which insurance products are presented as candidates for insurance contracts.

FIG. 12 shows functional blocks of an insurance product presentation device 14 according to the third embodiment. In the insurance product presentation device 14, the same configuration as that of the insurance product presentation device 10 is assigned the same reference number, and the description thereof is omitted.

The insurance product presentation device 14 has a commentary storage unit 150 . The presentation unit 114 presents the commentary text stored in the commentary text storage unit 150 together with the insurance product on the terminal 20 based on the feature amount used for presenting the insurance product.

FIG. 13 shows an example of the commentary text stored in the commentary text storage unit 150. As shown in FIG. The commentary text storage unit 150 stores feature quantities of the machine learning model 120 and commentary texts explaining the feature quantities in association with each other.

FIG. 13 is an example in which the machine learning model 120 is a decision tree. More specifically, in FIG. 13, there are two end points (also called terminal nodes, leaf nodes, etc.) presenting the insurance product "workmen's accident insurance" as a decision tree. is associated with an explanation. In this case, policyholders of industrial accident insurance are analyzed by a decision tree, and, for example, the user using the insurance product presentation device 14 extracts the value of the explanatory variable leading to the end point with a high contract rate from the analysis result as a feature quantity. . Further, the user prepares an explanation explaining the reason why the worker's accident insurance is presented to other users using the terminal 20 from the set of feature values, and stores it in the explanation sentence storage unit 150 .

The presentation unit 114 of the insurance product presentation device 14 outputs the insurance product to the terminal 20 in the same manner as the insurance product presentation device 10 by inputting information about the subject into the learned machine learning model 120. Based on the feature quantity used to present the insurance product, the explanatory text stored in the explanatory text storage unit 150 is further output. In the example of FIG. 13, when the information of the new policyholder reaches the end point 1 of the decision tree, the presentation unit 114 outputs the insurance product "workmen's accident insurance" and the commentary corresponding to the "end point 1".

As described above, according to the third embodiment, since the commentary is presented together with the insurance product, the user of the terminal 20 can know the reason why the insurance product was presented and can approach the customer efficiently. can. As a modification of the third embodiment, it is not necessary to integrate the contractor information in step S10 in the first embodiment. In that case, the information read by reading unit 110 to presentation unit 114 may be provided without providing estimation unit 112 and integrated information storage unit 100 of insurance product presentation device 14 .

FIG. 14 shows another example of the commentary text stored in the commentary text storage unit 150. As shown in FIG. In the example of FIG. 14, feature amounts and commentary are grouped for each of multiple explanatory variables. Furthermore, a commentary is associated with each of the plurality of feature quantities in each explanatory variable.

In the example of FIG. 14, the presentation unit 114 presents side by side explanatory texts corresponding to the feature amounts used by the machine learning model 120 to present the insurance product for each explanatory variable. For example, out of the new policyholder information, the machine learning model 120 determines the insurance product "income protection insurance" based on the policyholder attribute "the policyholder is 20" and the contract content "set insurance for three major diseases". Suppose that we have come to present In that case, the presentation unit 114 connects the explanatory texts associated with the respective feature amounts in order from the top, saying, "These are young people who are starting to consider compensation preparations by themselves. Additional proposals are effective because they are customers who place importance on the content of coverage." It is typical.” is presented.

The example of FIG. 14 is preferably used when the combination of feature amounts cannot be specified in advance, such as when the machine learning model 120 is not a decision tree. In this case, it is preferable that the total number of characters of the commentary corresponding to the combination of any feature amount extracted from each explanatory variable is within a predetermined number of characters. In the example of FIG. 14, the total number of characters of the explanatory text is 143 characters or less for any combination of the extracted feature amounts among the four explanatory variables. Machine learning models other than decision trees such as Support Vector Machine, Random Forest, XG Boost, LightGBM, Logistics Regression, etc. can be used.

In the example of FIG. 14, explanatory sentences corresponding to feature amounts extracted one by one from each of the four explanatory variables are presented. Instead of this, it is also possible to present commentary texts corresponding to a predetermined smaller number of feature amounts. In this case, among the respective feature amounts, the number of feature amounts is limited in descending order of importance for presenting the insurance product, for example, in descending order of weighting, and commentary texts corresponding to the feature amounts are provided. may be presented.

In the examples of FIGS. 13 and 14, in addition to the commentary text, the feature quantity associated with the commentary text may also be presented. In this case, a character string expressing the content indicated by the feature quantity may be stored in advance in the explanatory text storage unit 150 instead of the feature quantity as it is, and may be presented. For example, if the feature quantity is a numerical value of the contractor's age "20 or more and less than 30", "the contractor is in his twenties" representing the content indicated by the characteristic quantity may be stored in advance and presented.

FIG. 15 shows an example of feature quantities of another machine learning model 160 learned and inferred by the insurance product presentation devices 10, 12, and 14. As shown in FIG. The machine learning model 160 of FIG. 15 predicts the replacement of a car from the policyholder information as an insurance business related to the presentation of insurance products.

In the example of FIG. 15, heterogeneous mixture learning is performed by analyzing the presence or absence of car replacement records for automobile insurance policyholders by combining decision trees and linear regression. The insurance product presentation device 10 or the like uses the machine learning model 160 based on the learning to estimate the presence/absence of replacement based on the feature quantity corresponding to the importance of the subject information. Estimated replacement predictions can be utilized in presenting insurance products and in insurance solicitation.

FIG. 16 shows an example of feature amounts of still another machine learning model 170 learned and inferred by the insurance product presentation devices 10, 12, and 14. As shown in FIG. The machine learning model 170 of FIG. 16 analyzes the predictors of policyholder dropout as an insurance activity associated with the presentation of insurance products.

In the example of FIG. 16, the existence or non-existence of the contract omission of the life insurance is analyzed by the LightGBM of the ensemble model for the policyholders of the life insurance. The ensemble model uses a plurality of weak models for estimation, and random forest, XGBoost, etc. can also be used. AUC is an area under an ROC curve and is an evaluation index of the model. The insurance product presentation device 10 or the like uses the machine learning model 170 based on the learning to estimate the presence or absence of contract cancellation based on the feature amount from the information of the subject. The estimated presence or absence of drop-outs can be utilized for presentation of insurance products and insurance solicitation.

In any of the embodiments, an example of presenting an insurance product to an existing customer using existing customer information has been described. Not limited to this, even if it is not an existing customer, the insurance product is presented by applying the embodiment using the same type of information as a target person, so to speak, a potential customer who has not yet become a customer. be able to.

In both embodiments, insurance was described as an example. However, the embodiment is not limited to insurance, and can be applied to banking products such as loans, and narrowly defined financial products such as securities and bonds. In other words, the present embodiment can be applied to financial products in a broad sense, including products related to insurance and banking as well as financial products in a narrow sense. In that case, the insurance policy number in the above embodiment is an example of contract specifying information that specifies a contract for buying and selling financial related products.

Various embodiments of the invention may be described with reference to flowchart illustrations and block diagrams, where blocks refer to (1) steps in a process in which operations are performed or (2) devices responsible for performing the operations. may represent a section of Certain steps and sections may be implemented by dedicated circuitry, programmable circuitry provided with computer readable instructions stored on a computer readable medium, and/or processor provided with computer readable instructions stored on a computer readable medium. you can Dedicated circuitry may include digital and/or analog hardware circuitry, and may include integrated circuits (ICs) and/or discrete circuitry. Programmable circuits include logic AND, logic OR, logic XOR, logic NAND, logic NOR, and other logic operations, memory elements such as flip-flops, registers, field programmable gate arrays (FPGAs), programmable logic arrays (PLAs), etc. and the like.

Computer-readable media may include any tangible device capable of storing instructions to be executed by a suitable device, such that computer-readable media having instructions stored thereon may be designated in flowcharts or block diagrams. It will comprise an article of manufacture containing instructions that can be executed to create means for performing the operations described above. Examples of computer-readable media may include electronic storage media, magnetic storage media, optical storage media, electromagnetic storage media, semiconductor storage media, and the like. More specific examples of computer readable media include floppy disks, diskettes, hard disks, random access memory (RAM), read only memory (ROM), erasable programmable read only memory (EPROM or flash memory), Electrically Erasable Programmable Read Only Memory (EEPROM), Static Random Access Memory (SRAM), Compact Disc Read Only Memory (CD-ROM), Digital Versatile Disc (DVD), Blu-ray (RTM) Disc, Memory Stick, Integration Circuit cards and the like may be included.

The computer readable instructions may be assembler instructions, Instruction Set Architecture (ISA) instructions, machine instructions, machine dependent instructions, microcode, firmware instructions, state setting data, or instructions such as Smalltalk, JAVA, C++, etc. any source or object code written in any combination of one or more programming languages, including object-oriented programming languages, and conventional procedural programming languages such as the "C" programming language or similar programming languages; may include

Computer readable instructions may be transferred to a processor or programmable circuitry of a general purpose computer, special purpose computer, or other programmable data processing apparatus, either locally or over a wide area network (WAN), such as a local area network (LAN), the Internet, or the like. ) and may be executed to create means for performing the operations specified in the flowcharts or block diagrams. Examples of processors include computer processors, processing units, microprocessors, digital signal processors, controllers, microcontrollers, and the like.

FIG. 17 illustrates an example computer 2200 upon which aspects of the invention may be implemented in whole or in part. Programs installed on the computer 2200 may cause the computer 2200 to function as one or more sections of an operation or apparatus associated with an apparatus according to embodiments of the invention, or may Sections may be executed and/or computer 2200 may be caused to execute processes or steps of such processes according to embodiments of the present invention. Such programs may be executed by CPU 2212 to cause computer 2200 to perform certain operations associated with some or all of the blocks in the flowcharts and block diagrams described herein.

Computer 2200 according to this embodiment includes CPU 2212 , RAM 2214 , graphics controller 2216 , and display device 2218 , which are interconnected by host controller 2210 . Computer 2200 also includes input/output units such as communication interface 2222, hard disk drive 2224, DVD-ROM drive 2226, and IC card drive, which are connected to host controller 2210 via input/output controller 2220. there is The computer also includes legacy input/output units such as ROM 2230 and keyboard 2242 , which are connected to input/output controller 2220 through input/output chip 2240 .

CPU 2212 operates according to programs stored in ROM 2230 and RAM 2214, thereby controlling each unit. A graphics controller 2216 retrieves image data generated by CPU 2212 into itself, such as a frame buffer provided in RAM 2214 , and the image data is displayed on display device 2218 .

Communication interface 2222 communicates with other electronic devices over a network. Hard disk drive 2224 stores programs and data used by CPU 2212 within computer 2200 . DVD-ROM drive 2226 reads programs or data from DVD-ROM 2201 and provides programs or data to hard disk drive 2224 via RAM 2214 . The IC card drive reads programs and data from IC cards and/or writes programs and data to IC cards.

ROM 2230 stores therein programs that are dependent on the hardware of computer 2200, such as a boot program that is executed by computer 2200 upon activation. Input/output chip 2240 may also connect various input/output units to input/output controller 2220 via parallel ports, serial ports, keyboard ports, mouse ports, and the like.

A program is provided by a computer-readable medium such as a DVD-ROM 2201 or an IC card. The program is read from a computer-readable medium, installed in hard disk drive 2224 , RAM 2214 , or ROM 2230 , which are also examples of computer-readable medium, and executed by CPU 2212 . The information processing described within these programs is read by computer 2200 to provide coordination between the programs and the various types of hardware resources described above. An apparatus or method may be configured by implementing the manipulation or processing of information in accordance with the use of computer 2200 .

For example, when communication is performed between the computer 2200 and an external device, the CPU 2212 executes a communication program loaded in the RAM 2214 and sends communication processing to the communication interface 2222 based on the processing described in the communication program. you can command. The communication interface 2222 reads transmission data stored in a transmission buffer processing area provided in a recording medium such as the RAM 2214, the hard disk drive 2224, the DVD-ROM 2201, or an IC card under the control of the CPU 2212, and transmits the read transmission data. Data is transmitted to the network, or received data received from the network is written to a receive buffer processing area or the like provided on the recording medium.

In addition, the CPU 2212 causes the RAM 2214 to read all or necessary portions of files or databases stored in external recording media such as a hard disk drive 2224, a DVD-ROM drive 2226 (DVD-ROM 2201), an IC card, etc. Various types of processing may be performed on the data in RAM 2214 . CPU 2212 then writes back the processed data to the external recording medium.

Various types of information, such as various types of programs, data, tables, and databases, may be stored on recording media and subjected to information processing. CPU 2212 performs various types of operations on data read from RAM 2214, information processing, conditional decision making, conditional branching, unconditional branching, and information retrieval, as specified throughout this disclosure and by instruction sequences of programs. Various types of processing may be performed, including /replace, etc., and the results written back to RAM 2214 . In addition, the CPU 2212 may search for information in a file in a recording medium, a database, or the like. For example, if a plurality of entries each having an attribute value of a first attribute associated with an attribute value of a second attribute are stored in the recording medium, the CPU 2212 determines that the attribute value of the first attribute is specified. search the plurality of entries for an entry that matches the condition, read the attribute value of the second attribute stored in the entry, and thereby associate it with the first attribute that satisfies the predetermined condition. an attribute value of the second attribute obtained.

The programs or software modules described above may be stored in a computer readable medium on or near computer 2200 . Also, a recording medium such as a hard disk or RAM provided in a server system connected to a dedicated communication network or the Internet can be used as a computer-readable medium, thereby providing the program to the computer 2200 via the network. do.

Although the present invention has been described above using the embodiments, the technical scope of the present invention is not limited to the scope described in the above embodiments. It is obvious to those skilled in the art that various modifications and improvements can be made to the above embodiments. It is clear from the description of the scope of the claims that forms with such modifications or improvements can also be included in the technical scope of the present invention.

The execution order of each process such as actions, procedures, steps, and stages in devices, systems, programs, and methods shown in claims, specifications, and drawings is etc., and it should be noted that they can be implemented in any order unless the output of a previous process is used in a later process. Regarding the operation flow in the claims, specification, and drawings, even if explanations are made using "first," "next," etc. for the sake of convenience, it means that it is essential to carry out in this order. isn't it.

10, 12, 14 insurance product presentation device, 20 terminal, 100 integrated information storage unit, 110 reading unit, 112 estimation unit, 114, 116 presentation unit, 120, 122 machine learning model, 130 policyholder information DB, 132 complaint information DB , 134 accident information DB, 136 recruitment information DB, 150 commentary storage unit

Claims (4)

a commentary text storage unit that associates and stores a feature quantity of a machine learning model that has learned the relationship between information on a contractor and a contracted financial product, and a commentary commentary that explains the feature quantity;
a presentation unit that presents financial-related products to the target person by using the machine learning model for information about the target person;
When the number of the feature values used to present the financial product is equal to or greater than a predetermined number, the presentation unit stores the commentary text storage unit corresponding to the predetermined number of the feature values. A financial-related product presentation device that further presents the stored commentary text. 2. The presentation unit according to claim 1, wherein the presentation unit limits the feature values to the predetermined number in descending order of importance for presenting the financial product from the feature values, and presents the commentary. financial product presentation device. 3. The financial product according to claim 1 or 2, wherein the commentary text for each of the plurality of feature quantities is stored such that the total number of characters of the commentary text corresponding to the feature quantity is within a predetermined number of characters. presentation device. to the computer,
a commentary text storage function that associates and stores a feature quantity of a machine learning model that has learned the relationship between information about a contractor and a contracted financial product, and a commentary commentary that explains the feature quantity;
Realizing a presentation function of presenting financial-related products to the target person by using the machine learning model for information about the target person,
The presentation function is the commentary text storage function corresponding to the predetermined number of the feature amounts used to present the financial product when the number of the feature amounts used to present the financial product is equal to or greater than a predetermined number. The program further comprising presenting said stored commentary text.

Images