JP2020170534A - Information processing device, success assisting method, and success assisting program - Google Patents

Abstract

To provide an information processing device, a success assisting method, and a success assisting program that make more effective matching service possible.SOLUTION: The information processing device includes acquisition means for acquiring the attribute information of each of matching objects, and presentation means for presenting the content of change in the attribute information of at least one of the matching objects and a success score of the matching objects the matching of which is successful. The success score is estimated on the basis of the attribute information that includes the change content of each of the matching objects.SELECTED DRAWING: Figure 1

Description

The present invention relates to an information processing device, a success support method, and a success support program that support an improvement in the success rate for a predetermined purpose between people or things.

For example, at a dating agency, a matching service that introduces members who meet the desired conditions entered by the members is widely used.

Most of these matching services use two types of information, such as personal information such as name, age, and educational background entered by the member, and the desired conditions for the desired person, and select the person who meets the desired conditions from among the members. Search and extract.

For example, Patent Document 1 describes an example of a matching service using its own attribute and a condition indicating an attribute required of the other party. The matching service described in Patent Document 1 extracts a user with an attribute that matches the condition indicating the attribute that one user wants from the other party, and further, the condition that the extracted user indicates the attribute requested from the other party is the above-mentioned one user. Determine if the attributes of are matched. Then, when it is determined that they match, the extraction user is notified of the attribute of one user.

Further, for example, Patent Document 2 describes a matching service that searches for a marriage partner based on scoring data obtained by scoring user-specific data.

Japanese Unexamined Patent Publication No. 2011-11546 Japanese Unexamined Patent Publication No. 2007-095011

However, even if both wishes are met, the marriage rate is not always high. In addition, the marriage rate may be improved by changing some of the attributes and conditions of the person. However, the current situation is that introductions that deviate from such desired conditions and advice on changes to one's own attributes and conditions are made based on human experience. Since subjective advice such as human experience is poorly grounded, it is preferable to give advice based on objective data in order to provide a more effective matching service.

In view of the above, it is an object of the present invention to provide an information processing device, a success support method, and a success support program capable of more effective matching services.

The information processing apparatus according to the present invention presents an acquisition means for acquiring each attribute information of the matching target, a change content in at least one attribute information of the matching target, and a success score for successful matching of the matching target. The success score, including the means, is estimated based on the attribute information including each change of the matching target.

Further, in the success support method according to the present invention, the computer acquires each attribute information of the matching target, and presents the change contents in at least one attribute information of the matching target and the success score for successful matching of the matching target. However, the success score is estimated based on the attribute information including each change of the matching target.

Further, the success support program according to the present invention includes a process of acquiring each attribute information of the matching target on a computer, changes in at least one attribute information of the matching target, a success score for successful matching of the matching target, and the like. The success score is estimated based on the attribute information including the changed contents of each matching target.

According to the present invention, a more effective matching service can be realized.

It is a block diagram which shows the structural example of the success support system of 1st Embodiment. It is a block diagram which shows the more detailed configuration example of the proposed means 12. It is a flowchart which shows an example of the operation of the success support system of 1st Embodiment. It is a flowchart which shows an example of the more detailed operation of a proposed step. It is a block diagram which shows the other configuration example of the success support system 10. It is a block diagram which shows the other configuration example of the success support system 10. It is a block diagram which shows the structural example of the marriage simulation system 100 of 2nd Embodiment. It is explanatory drawing which shows an example of the data structure of a desired condition. It is explanatory drawing which shows an example of the data structure of user information. It is explanatory drawing which shows an example of the data structure of a matching record. It is explanatory drawing which shows an example of the data structure of the 1st estimation result (matching rate prediction result). It is explanatory drawing which shows an example of the data structure of the marriage record. It is explanatory drawing which shows an example of the data structure of the 2nd estimation result (marriage rate prediction result). It is explanatory drawing which shows the example of the display method of the simulation result. It is a flowchart which shows an example of the operation of the matching learning step of the marriage simulation system 100. It is a flowchart which shows an example of the operation of the marriage learning step of the marriage simulation system 100. It is a flowchart which shows an example of the operation of the matching prediction step of the marriage simulation system 100. It is a flowchart which shows an example of the operation of the marriage prediction step of the marriage simulation system 100. It is a flowchart which shows an example of the operation of the marriage simulation step of the marriage simulation system 100. It is explanatory drawing which shows another example of the display method of a simulation result. It is explanatory drawing which shows another example of the display method of a simulation result. It is explanatory drawing which shows another example of the display method of a simulation result.

Embodiment 1.
Hereinafter, embodiments of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram showing a configuration example of a success support system according to the first embodiment of the present invention. The success support system 10 shown in FIG. 1 includes a relational information storage means 11 and a proposal means 12.

The relationship information storage means 11 stores information indicating the relationship between the success rate of a person or thing for a predetermined purpose and the attribute information possessed by each target person or thing.

The success rate may be, for example, the marriage rate. The success rate is not limited to the marriage rate, and may be, for example, the success rate for a predetermined purpose in the process leading to marriage such as dating, or the success rate of an organized project in work, etc., not limited to marriage. There may be. Further, the target for measuring the success rate is not limited to humans, and may be, for example, animals and plants, microorganisms, chemical substances, and the like. In the present embodiment, the success rate covers all items that indicate the possibility of success for matters whose success or failure depends on so-called compatibility.

In addition, the attribute information is not limited to unique information such as properties, characteristics, and titles of the target person or thing, and for example, the thing (including documents) generated by the target person or thing. It may contain information.

The proposing means 12 proposes a modification of at least a part of the attribute information possessed by at least one person or thing in an unknown combination based on the information stored in the relational information storage means 11.

The proposed means 12 is, for example, a changed content estimated based on the changed contents of at least one item included in the attribute information of at least one person or thing and the information stored in the relational information storage means 11. You may propose changes by presenting the success rate of.

Further, the proposal means 12 is a first user and a first user when the success rate is due to the users registered in advance, such as the success rate for a predetermined purpose by the users registered as members. Is a change for improving the success rate in combination with another user who satisfies the condition specified by, and may propose a change of at least a part of the attribute information possessed by the first user. Further, as another example of proposal for the same purpose, the proposal means 12 changes, for example, at least one item included in the attribute information possessed by another user and satisfying the condition specified by the first user. , May suggest.

The above changes may be made only for items designated as items that can be changed.

FIG. 2 is a block diagram showing a more detailed configuration example of the proposed means 12. As shown in FIG. 2, the proposed means 12 may include a success rate estimating means 121, a simulation means 122, and an analysis means 123.

When the set of attribute information is input, the success rate estimating means 121 is determined by the set of attribute information based on the input set of attribute information and the information stored in the relational information storage means 11. Estimate the success rate for the purpose.

When the simulation means 122 accepts the designation of a combination of arbitrary people or things, it changes at least a part of the attribute information of at least one of the designated combinations, and simulates the change in the success rate accompanying the change.

Based on the result of the simulation, the analysis means 123 changes the part of the attribute information set by the combination to improve the success rate for a predetermined purpose by the combination of arbitrary people or things, the content of the change in the part, and the change contents. Identify the contribution to the change.

In the configuration shown in FIG. 2, the proposing means 12 proposes a change by displaying, for example, the changed part specified by the analysis means 123, the content of the change in the part, the contribution rate to the change, and the success rate after the change. You may.

FIG. 3 is a flowchart showing an example of the operation of the present embodiment. In the example shown in FIG. 3, for example, the proposal means 12 has a success rate for a predetermined purpose by people or things stored in the relational information storage means 11 when an arbitrary combination whose success rate is unknown is specified. And read out the information indicating the relationship between the target person or the attribute information possessed by each thing (step S11). At this time, the proposing means 12 may also read out the attribute information possessed by each person or each thing that is the target of the designated unknown combination.

Then, the proposing means 12 is information that is stored in the relational information storage means 11 and shows the relationship between the success rate for a predetermined purpose by people or things and the attribute information possessed by each target person or thing. Based on, at least a part of the attribute information possessed by at least one person or thing in the specified unknown combination is proposed (step S12). In step S12, the proposing means 12 changes at least a part of the attribute information of at least one of the specified unknown combinations, analyzes the change in the success rate with the change of the attribute information, and is identified as a result of the success. The change may be proposed by displaying the changed part for improving the rate, the content of the change in the part, and the contribution rate to the change.

FIG. 4 is a flowchart showing an example of a more detailed operation of the proposed step in step S12 of FIG. In the example shown in FIG. 4, when an unknown combination is specified, the proposing means 12 (for example, the simulation means 122) sets a set of attribute information including the attribute information of each person or each thing targeted by the combination. Set (step S121).

Next, the proposal means 12 (for example, the simulation means 122) predicts (estimates) the success rate based on the set set of attribute information by using the success rate estimation means 121 (step S122). For example, the simulation means 122 may first estimate the success rate of a set of attribute information consisting of the current attribute information of each person or thing that is the target of the combination.

Next, the proposing means 12 (for example, the simulating means 122) changes at least a part of at least one of the attribute information and simulates the change in the success rate accompanying the change in the attribute information (returns to step S123 and step S122). .. The simulation means 122 proceeds to step S124 after making the above changes for, for example, a predetermined number of times or all the items that can be changed.

In step S124, the proposing means 12 (for example, the analysis means 123) analyzes the simulation result, and changes, for example, a set of attribute information for improving the success rate for a predetermined purpose by the specified combination. Or identify the contribution to the change.

Finally, the proposed means 12 (for example, the analysis means 123) displays the analysis result (step S125). In step S125, the analysis means 123 may display, for example, the changed part, the changed content, or the degree of contribution to the change with respect to the specified set of attribute information.

According to such a configuration, not only the combination of the current attribute information but also the combination of the attribute information after changing some of the attribute information is based on the objective data of the past performance rather than the experience knowledge of the person. Since proposals can be made in consideration of the success rate, more effective matching services can be provided.

Further, FIG. 5 is a block diagram showing another configuration example of the success support system 10. The success support system 10 selects candidates to narrow down the other people or things to be simulated for the success rate based on the desired conditions for the attribute information of the other person or thing specified by the first person or thing, for example. Means 13 may be further provided.

Further, the success support system 10 uses, for example, a set of attribute information possessed by each person or each thing in a combination of people or things that can determine success or failure for a predetermined purpose as learning data, and any person or thing. Further, a learning means 14 for machine learning the relationship between the success rate of each other for a predetermined purpose and the attribute information possessed by each person or each thing to be combined may be provided. By providing such a learning means 14, learning becomes possible even if the number of dimensions of the input data is enormous. The number of dimensions of the input data may be, for example, 1000 or more, or 10000 or more.

Further, FIG. 6 is a block diagram showing another configuration example of the success support system 10. As shown in FIG. 6, the success support system 10 may include a display means 15 for displaying the result of the simulation instead of the analysis means 123. It should be noted that the display means 15 is a modification to the set of attribute information specified based on the result of the simulation, like the analysis means 123, in order to improve the success rate for a predetermined purpose by the set of designated attribute information. The location, the content of the change in the location, and the degree of contribution to the change may be displayed.

Further, the candidate selection means 13 and the learning means 14 may be added to the configuration shown in FIG.

In the present embodiment, the relational information storage means 11 is realized by, for example, a storage device. Further, the proposal means 12, the candidate selection means 13, and the learning means 14 are realized by, for example, an information processing device that operates according to a program. Further, the display means 15 is realized by, for example, an information processing device that operates according to a program and a display device such as a display.

Embodiment 2.
Next, a second embodiment of the present invention will be described. In the following, the case where the success rate is the marriage rate will be described as an example, but the success rate is not limited to the marriage rate.

FIG. 7 is a block diagram showing a configuration example of the marriage simulation system 100 of the present embodiment. The marriage simulation system 100 corresponds to another example of the success support system 10 of the first embodiment. The marriage simulation system 100 shown in FIG. 7 matches the desired condition storage unit 21, the first data processing unit 22, the matching performance storage unit 23, the first learning unit 24, and the matching learning model storage unit 25. Score estimation unit 26, first estimation result storage unit 27, user information storage unit 28, second data processing unit 29, marriage record storage unit 30, second learning unit 31, and marriage learning. It includes a model storage unit 32, a marriage score estimation unit 33, a second estimation result storage unit 34, and a simulation unit 35.

In addition, the first data processing unit 22 includes a desired condition preprocessing unit 221 and a desired condition feature extraction unit 222. Further, the first learning unit 24 includes a matching result preprocessing unit 241 and a matching result learning unit 242. In addition, the second data processing unit 29 includes a user information preprocessing unit 291 and a user information feature extraction unit 292. In addition, the second learning unit 31 includes a marriage record preprocessing unit 311 and a marriage record learning unit 312.

The marriage simulation system 100 is realized by, for example, an information processing device such as an information processing terminal such as a personal computer or a server device, and a storage device group such as a database system accessible to the information processing device. At this time, the first data processing unit 22, the first learning unit 24, the matching score estimation unit 26, the second data processing unit 29, the second learning unit 31, the marriage score estimation unit 33, and the simulation unit 35 For example, it may be realized by a CPU included in the information processing device. In that case, the CPU realizes the function of each processing unit by reading a program describing the operation of each processing unit stored in a predetermined storage device and operating according to the program. Further, the desired condition storage unit 21, the matching record storage unit 23, the matching learning model storage unit 25, the first estimation result storage unit 27, the user information storage unit 28, the marriage record storage unit 30, the marriage learning model storage unit 32, The second estimation result storage unit 34 may be realized by, for example, a storage device group accessible to the information processing device. The number of storage devices may be one or a plurality.

The desired condition storage unit 21 holds a desired condition, which is a condition desired by a user who receives a matching service (hereinafter, simply referred to as a user). Examples of desired conditions include age, height, weight, personality, etc., which are retained as part of the attribute information of each user.

FIG. 8 is an explanatory diagram showing an example of a data structure of desired conditions held by the desired condition storage unit 21. As shown in FIG. 8, the desired condition storage unit 21 associates with the member ID as the user's identification information, and at least a part of the attribute information such as age, height, weight, and annual income desired by the other party. Information that specifies the content of the item may be retained.

The user information storage unit 28 holds user information which is attribute information of the user. Examples of user information include personal information such as name, age, gender, and educational background. Further, the user information may include information or the like generated by the user by using SNS (Social Networking Service) or the like.

FIG. 9 is an explanatory diagram showing an example of a data structure of user information held by the user information storage unit 28. As shown in FIG. 9, the user information storage unit 28, for example, associates the user information with the member ID, which is the identification information of the user, and associates the user's name, age, gender, and height with the user information. Alternatively, you may retain profile data in text format. The user information may include information regarding the desired conditions of the user.

The first data processing unit 22 processes the designated desired condition into data in a format that can be handled by the first learning unit 24.

In the present embodiment, the desired condition preprocessing unit 221 reads, for example, a record including the desired condition of the designated user from the desired condition storage unit 21 based on the command from the desired condition feature extraction unit 222, and obtains the desired condition. You may generate a vector. Here, the desired condition vector is a representation of the desired condition as a multidimensional vector composed of numerical values.

Further, the desired condition feature extraction unit 222 performs feature extraction from the desired condition vector generated by the desired condition preprocessing unit 221 and generates a desired condition feature vector, for example. Here, the desired condition feature vector may be a numerical vector having a number of dimensions smaller than the number of dimensions of the desired condition vector.

The second data processing unit 29 processes the designated user information into data in a format that can be handled by the second learning unit 31.

In the present embodiment, the user information preprocessing unit 291 reads, for example, a record including the designated user information from the user information storage unit 28 based on a command from the user information feature extraction unit 292, and uses the record. The person information condition vector may be generated. Here, the user vector is a representation of user information as a multidimensional vector composed of numerical values.

Further, the user information feature extraction unit 292 performs feature extraction from the user vector generated by the user information preprocessing unit 291 to generate a user feature vector. Here, the user feature vector may be a numerical vector having a number of dimensions less than the number of dimensions of the user vector.

The matching record storage unit 23 holds the matching record. Examples of the information indicating the matching results include information in which the desired conditions of a certain user are associated with the user information of the user who matches the desired conditions of the user. Further, as another example of the information indicating the matching record, there is information in which the desired condition of a certain user is associated with the user information of the user who does not match the desired condition of the user. In addition, as another example of the information indicating the matching record, the information in which the desired condition of a certain user is associated with the user information of another user and the information indicating whether or not the desired condition is met is provided. Can be mentioned.

FIG. 10 is an explanatory diagram showing an example of a matching record data structure held by the matching record storage unit 23. As shown in FIG. 10, the matching record storage unit 23 includes, for example, a member ID (desired condition specific information) of a user who has issued a desired condition for a pair for which a matching determination has been made in the past as a matching record. Information that associates the member ID (target person identification information) of the user who is a matching candidate of the user with the information indicating the matching determination result may be retained.

The first learning unit 24 is based on the information indicating the matching record stored in the matching record storage unit 23, between the desired conditions of any user and the user information of another arbitrary user. Learn the relationship between the presence or absence of matching (more specifically, the matching score, which is the degree to which the user information matches the desired conditions).

In the present embodiment, whether or not the first learning unit 24 matches the desired condition feature vector and the user feature vector generated based on the information indicating the matching record stored in the matching record storage unit 23. Using the information including the label of, a matching learning model showing the relationship of the matching score is generated for the combination of the arbitrary desired condition and the user information by machine learning. In addition, the first learning unit 24 not only combines the desired conditions desired by a certain user with the user information of the user who matches the desired conditions, but also the desired conditions and the user information of each other. You may generate a matching learning model showing the relationship of matching scores with respect to the combination of. This may be, for example, generating a matching learning model showing the relationship of matching scores for a combination of user information including desired conditions.

More specifically, the matching record preprocessing unit 241 reads information indicating the matching record from the matching record storage unit 23 in response to a command from the matching record learning unit 242, and for example, label information <member ID (desired condition). (Specific information), member ID (target person specific information), success / failure label indicating whether or not they match> are generated.

The matching achievement learning unit 242 uses the desired condition feature vector of the member ID (desired condition specific information) generated by the desired condition feature extraction unit 222 based on the label information generated by the matching achievement preprocessing unit 241 and the user information feature. Learning data <desired condition feature vector, user feature vector, success / failure label> is generated using the user feature vector of the member ID (target person identification information) generated by the extraction unit 292, and matching learning is performed by machine learning. Generate a model.

The matching learning model storage unit 25 holds the learning result by the first learning unit 24 (more specifically, the matching performance learning unit 242), and more specifically, the information indicating the matching learning model. The information indicating the matching learning model may be, for example, information indicating the relationship between the pair of the desired condition feature vector and the user feature vector and the matching score.

The matching score estimation unit 26 reads out the learning result (information indicating the matching learning model) stored in the matching learning model storage unit 25 by the first learning unit 24, and the unknown member ID set indicates the unknown. The matching score is estimated (calculated) for the combination of the desired conditions and the user information. The matching score estimation unit 26 may calculate a matching score for a combination of desired conditions between candidate users and user information. For example, the matching score estimation unit 26 may calculate the matching score by using a learning model that has learned the relationship of the matching score with respect to the combination of the desired conditions of the candidate users and the user information. Further, for example, the matching score estimation unit 26 calculates a first matching score for a combination of a designated user's desired condition and user information of another user, and further, the other use. A second matching score may be calculated for the combination of the desired condition of the person and the user information of the designated user, and the sum of these may be used as the final matching score.

As the first estimation result, the first estimation result storage unit 27 holds the matching score calculated by the matching score estimation unit 26 together with the set of member IDs to be calculated.

FIG. 11 is an explanatory diagram showing an example of a data structure of the first estimation result (matching rate prediction result) held by the first estimation result storage unit 27. As shown in FIG. 11, the first estimation result storage unit 27 refers to, for example, a member ID that identifies a user who has given a desired condition for a combination of users for which a matching score has been calculated, and the desired condition. Information that associates the member ID that identifies the user whose matching score is calculated with the matching score may be retained.

The marriage record storage unit 30 holds the marriage record. Examples of the information indicating the marriage record include information in which the member ID of one member and the member ID of the other member are associated with each other regarding the combination of members who have been married in the past. In addition, as another example of the information showing the marriage record, there is information in which the member ID of one member and the member ID of the other member are associated with each other for the combination of members who have not been married in the past. .. In addition, as another example of the information showing the marriage record, regarding the combination of the members introduced in the past, the member ID of one member, the member ID of the other member, and whether or not the marriage has been reached are shown. Information that is associated with information can be mentioned. In addition to this information, the information showing the marriage record is also the date and time when the marriage was introduced, the date and time when the marriage was reached, and the date and time when the decision was made when the marriage was not reached. It may include information and the like.

FIG. 12 is an explanatory diagram showing an example of a data structure of the marriage record held by the marriage record storage unit 30. As shown in FIG. 12, the marriage record storage unit 30 may hold information including the member IDs of both members and the date and time when the marriage was established, regarding the combination of the members who have reached the marriage.

The second learning unit 31 is based on the information indicating the marriage record stored in the marriage record storage unit 30, whether or not there is a marriage between any user and any other user (more specifically). Learn the relationship (marriage score), which indicates the possibility of getting married.

In the present embodiment, the second learning unit 31 includes the first user feature vector and the second user vector generated based on the information indicating the marriage record stored in the marriage record storage unit 30. Using the information including the success / failure label indicating whether or not the person has been married, machine learning is used to generate a marriage learning model showing the relationship between the marriage scores for any combination of user information.

More specifically, the marriage record preprocessing unit 311 reads the information indicating the marriage record from the marriage record storage unit 30 in response to the command from the marriage record learning unit 312, and the label information <member ID (first target person). (Specific information), member ID (second target person specific information), success / failure label indicating whether or not the person was married> is generated.

Based on the label information generated by the user information feature extraction unit 292, the marriage record learning unit 312 has a user feature vector of one user who has a record of marriage in the past (user feature vector of the first target person). ) And the user feature vector of the other user (user feature vector of the second target user), and the learning data <user feature vector of the first target user, user feature vector of the second target user. , Success / failure label> is generated, and a marriage learning model is generated by machine learning. At this time, the marriage performance learning unit 312 may learn the weight given to the marriage score by each item of the user information and include the learning result in the marriage learning model.

The marriage learning model storage unit 32 holds the learning result of the marriage achievement learning unit 312, more specifically, information indicating the marriage learning model. The information indicating the marriage learning model may be, for example, information indicating the relationship between any set of user feature vectors and the marriage score.

The marriage score estimation unit 33 reads out the learning result (information indicating the marriage learning model) by the second learning unit 31 stored in the marriage learning model storage unit 32, and is unknown, which is indicated by the designated member ID set. The marriage score is estimated (calculated) for the combination of the user information of the two users.

In the present embodiment, the marriage score estimation unit 33 reads the first estimation result from the first estimation result storage unit 27, and meets or meets the desired conditions of the designated user (third target person). Other users (fourth target person) who have a high degree of doing so are candidates for estimating the marriage score. More specifically, the marriage score estimation unit 33 is another user (fourth target person) whose matching score is equal to or higher than a predetermined value with respect to the desired conditions of the designated user (third target person). Is a candidate. Then, the marriage score estimation unit 33 generates a user feature vector of the third target person and a user feature vector of the fourth target person. Then, the marriage score estimation unit 33 marries a set of the user characteristic vector of the third target person and the user characteristic vector of one of the fourth target people by using the marriage learning model. Estimate (calculate) the score.

As the second estimation result, the second estimation result storage unit 34 holds the marriage score calculated by the marriage score estimation unit 33 together with the set of member IDs to be calculated.

FIG. 13 is an explanatory diagram showing an example of a data structure of the second estimation result (marriage rate prediction result) held by the second estimation result storage unit 34. As shown in FIG. 13, the second estimation result storage unit 34 has, for example, a member ID that identifies the third target person and a member ID that identifies the fourth target person for the combination of users who have calculated the marriage score. And the information associated with the marriage score may be retained.

The simulation unit 35 reads the second estimation result from the second estimation result storage unit 34, and simulates the possibility that one user will marry another user with a higher matching score.

The simulation unit 35 is, for example, the current user information of a certain user (third target person) and the current user of another user (fourth target person) having a higher matching score with respect to the user. The marriage score based on the combination of information may be displayed as a prediction result of the current marriage rate with the partner. Then, the simulation unit 35 changes at least one user information in each combination based on the prediction result, and analyzes the state of the change in the marriage score accompanying the change. At this time, the simulation unit 35 searches for user information of the third target person such that the marriage score with a predetermined partner is maximized or the total marriage score with the partner with a higher matching score is maximized. , The difference between the result and the current user information may be output.

Further, the simulation unit 35 may present the changed part of the changed user information and the changed content as a goal setting of the user who is the third target person. Further, the simulation unit 35 searches for a combination of user information that maximizes the total marriage score with a partner who satisfies a part of the desired conditions regardless of the matching score, and the result and the current present. The difference from the combination of user information may be output. At this time, the changed part of the user information of the fourth target user and the changed content may be presented as a change proposal of the desired conditions of the user who is the third target user.

FIG. 14 is an explanatory diagram showing an example of a method of displaying a simulation result. FIG. 14 shows an example of difference display of user information as an example of information displayed as a result of simulation. As shown in FIG. 14, the simulation unit 35 is a member of the user (fourth target person) on the other side who performed the simulation as a proposal for change to the user (third target person) who gave the desired condition. In association with the ID, the current value of the user information of itself (third target person) and the target setting value (changed value) for at least some items specified as a result of the simulation may be presented. .. The simulation unit 35 may present the weight of the item (the degree of influence on the marriage score) together with the target setting value. The weight may be, for example, extracted by machine learning using past achievements performed by the marriage achievement learning unit 312, or may be given based on the result of simulation.

The matching result storage unit 23, the first learning unit 24, the matching learning model storage unit 25, the matching score estimation unit 26, and the first estimation result storage unit 27 of the present embodiment are candidates for the first embodiment. It corresponds to the selection means 13. Further, the marriage learning model storage unit 32 corresponds to the relationship information storage means 11 of the first embodiment. Further, the second learning unit 31 corresponds to the learning means 14 of the first embodiment. Further, the marriage score estimation unit 33 and the simulation unit 35 correspond to the proposal means 12 of the first embodiment.

Next, the operation of this embodiment will be described. The operation of the marriage simulation system 100 of the present embodiment includes five steps of a matching learning step, a marriage learning step, a matching prediction step, a marriage prediction step, and a marriage simulation step.

In the matching learning step, the relationship of the matching score for the combination of the desired condition and the user information is machine-learned mainly based on the past matching results, and a matching learning model is generated as the learning result.

In the marriage learning step, the relationship of the marriage score to the combination of user information is machine-learned mainly based on the past marriage record, and the marriage learning model is generated as the learning result.

In the matching prediction step, the learning result in the matching learning step is read out, and the matching score between members is calculated using the combination of the specified desired condition and the user information.

In the marriage prediction step, the calculation result of the matching score in the matching prediction step and the learning result in the marriage learning step are read out, and the user information of the specified user and the matching score are higher for the user. The marriage score based on the current user information of the members is calculated by using the combination of the user information of the users of.

In the marriage simulation step, the marriage score calculated in the marriage prediction step based on the current user information between members is read out, and the combination of user information that increases the marriage score is searched by simulation, and the result of the search is obtained. Based on the difference in user information, we propose changes to the designated users to increase the marriage rate.

FIG. 15 is a flowchart showing an example of the operation of the matching learning step of the marriage simulation system 100. In the example shown in FIG. 15, first, the matching record preprocessing unit 241 reads the matching record, which is the history of the matching result between the desired conditions and the user information by the past members, from the matching record storage unit 23, and the label information < A member ID (desired condition specific information), a member ID (target person specific information), and a success / failure label> are generated (step S101).

Next, the matching result learning unit 242 of the first data processing unit 22, the second data processing unit 29, and the first learning unit 24 repeats the operations of steps S103 to S108 for the number of matching results (step). S102, step S109).

In step S103, the desired condition preprocessing unit 221 reads the user's desired condition indicated by the label information generated in step S101 from the desired condition storage unit 21 and generates a desired condition vector. Here, at least, the user's desired condition indicated by the member ID as the desired condition specific information is read. In addition, when considering each other's desired conditions, the user's desired conditions indicated by the member ID as the target person identification information may also be read. The desired condition preprocessing unit 221 reads, for example, a record matching the member ID as the desired condition specific information from the desired condition storage unit 21, converts the record into a vector format, and generates a desired condition vector.

Vectorization (quantification) of desired conditions is carried out, for example, as follows. First, the desired condition preprocessing unit 221 decomposes items described in text such as a profile into words by using morphological analysis, and vectorizes whether or not the words appear by a predetermined numerical value (for example, 0 or 1). To be. Items other than text (height, weight, educational background, etc.) are not decomposed into words, and whether or not they are included in a predetermined range classified in advance is vectorized with a predetermined numerical value in the same manner as above. ..

In step S104, the desired condition feature extraction unit 222 reads out the desired condition vector generated in step S103, extracts features from the read desired condition vector, and generates a desired condition feature vector.

In step S105, the user information preprocessing unit 291 reads the user information of the user indicated by the label information generated in step S101 from the user information storage unit 28, and generates a user vector. Here, at least, the user information of the user indicated by the member ID as the target person specific information is read. In addition, when considering each other's desired conditions, the user information of the user indicated by the member ID as the target person identification information may also be read. The user information preprocessing unit 291 reads, for example, a record having a matching member ID from the user information storage unit 28, converts the record into a vector format, and generates a user vector. The vectorization (quantification) of the user information may be performed by the same method as the method for generating the desired condition vector shown in step S103.

In step S106, the user information feature extraction unit 292 reads out the user vector generated in step S105, extracts features from the read user vector, and generates a user feature vector.

In step S107 and step S108, the matching achievement learning unit 242 uses the desired condition feature vector generated in step S104, the user feature vector generated in step S106, and the success / failure label acquired in step S101. , Adjust the model parameters of the matching learning model.

In the present embodiment, the matching performance learning unit 242 first calculates the cosine similarity between the desired condition feature vector and the user feature vector (step S107). Next, the matching performance learning unit 242 updates the model parameters using the calculated cosine similarity and the success / failure label. The learning method is not limited to the above.

When the above process is repeated for the number of matching achievements, the matching achievement learning unit 242 writes the finally adjusted model parameters to the matching learning model storage unit 25 (step S110), and ends the matching learning step.

FIG. 16 is a flowchart showing an example of the operation of the marriage learning step of the marriage simulation system 100.

In the example shown in FIG. 16, first, the marriage record preprocessing unit 311 reads the marriage record, which is the history of the marriage results of the past members, from the marriage record storage unit 30, and label information <member ID (first target person identification). Information), member ID (second target person identification information), success / failure label> are generated (step S201).

Next, the matching result learning unit 242 of the second data processing unit 29 and the second learning unit 31 repeats the operations of steps S203 to S208 for the number of matching results (step S202, step S209).

In step S203, the user information preprocessing unit 291 reads the user information of the first target person indicated by the label information generated in step S201 from the user information storage unit 28, and generates a user vector. The user information preprocessing unit 291 reads, for example, a record matching the member ID as the first target person specific information from the user information storage unit 28, converts the record into a vector format, and generates a user vector. .. The vectorization (quantification) of the user information may be performed by the same method as the method for generating the user vector shown in step S105.

In step S204, the user information feature extraction unit 292 reads out the user vector generated in step S203, extracts features from the read user vector, and generates a user feature vector.

In step S205, the user information preprocessing unit 291 reads the user information of the second target person indicated by the label information generated in step S201 from the user information storage unit 28, and generates a user vector. The user information preprocessing unit 291 reads, for example, a record matching the member ID as the second target person specific information from the user information storage unit 28, converts the record into a vector format, and generates a user vector. ..

In step S206, the user information feature extraction unit 292 reads out the user vector generated in step S205, extracts features from the read user vector, and generates a user feature vector.

In step S207 and step S208, the marriage record learning unit 312 uses the user feature vector generated in step S204, the user feature vector generated in step S206, and the success / failure label acquired in step S201. , Adjust the model parameters of the marriage learning model.

In the present embodiment, the marriage record learning unit 312 first calculates the cosine similarity between the two user feature vectors (step S207). Next, the marriage record learning unit 312 updates the model parameters using the calculated cosine similarity and the success / failure label. The learning method is not limited to the above.

When the above process is repeated for the number of marriage records, the marriage record learning unit 312 writes the finally adjusted model parameters in the marriage learning model storage unit 32 (step S210), and ends the marriage learning step.

FIG. 17 is a flowchart showing an example of the operation of the matching prediction step of the marriage simulation system 100.

In the example shown in FIG. 17, first, the matching score estimation unit 26 reads the adjusted model parameters written in step S110 from the matching learning model storage unit 25 (step S301).

Next, the matching score estimation unit 26 requests the first data processing unit 22 to generate a desired condition feature vector from the desired conditions of the user (third target person) who simulates marriage (in step S302). move on). Here, the third target person is designated, for example, by an advisor or a user who uses this system.

In step S302, the desired condition preprocessing unit 221 reads the desired condition of the designated third target person from the desired condition storage unit 21 and generates a desired condition vector.

Next, in step S303, the desired condition feature extraction unit 222 performs feature extraction on the desired condition vector generated in step S302 to generate the desired condition feature vector.

Next, the second data processing unit 29 and the matching score estimation unit 26 repeat the operations of steps S305 to S308 for the number of users who are candidates for the third target person (steps S304 and S309). .. Here, the user who is the partner candidate of the third target person may be, for example, all users of the opposite sex who are currently registered as members. For example, when the third target person is designated, the matching score estimation unit 26 may generate a list of member IDs of users who are candidates for the third target person.

In step S305, the user information preprocessing unit 291 extracts the member IDs included in the list in order, reads the user information of the user indicated by the extracted member IDs from the user information storage unit 28, and reads the user vector. To generate.

In step S306, the user information feature extraction unit 292 reads out the user vector generated in step S304, extracts features from the read user vector, and generates a user feature vector.

In step S307, the matching score estimation unit 26 uses the model parameters read in step S301, the desired condition feature vector generated in step S304, and the user feature vector generated in step S306, and is used by the third target person. The matching score between the user and the user who is the candidate for the other party is calculated.

In step S308, the matching score estimation unit 26 writes the calculation result in step S307 into the first estimation result storage unit 27. The matching score estimation unit 26 may write the calculation result to the first estimation result storage unit 27 in the form of <member ID (desired condition specific information), member ID (target person information specific information), matching score>, for example. Good.

When the above process is repeated for the number of opponent candidates, the matching score estimation unit 26 ends the matching prediction step.

FIG. 18 is a flowchart showing an example of the operation of the marriage prediction step of the marriage simulation system 100.

In the example shown in FIG. 18, first, the marriage score estimation unit 33 reads the adjusted model parameters written in step S210 from the marriage learning model storage unit 32 (step S401).

Next, the marriage score estimation unit 33 reads out the top N records from the estimation result (first estimation result) of the matching score with the designated third target person from the first estimation result storage unit 27. (Step S402).

Next, the marriage score estimation unit 33 requests the second data processing unit 29 to generate the user feature vector of the third target person (proceed to step S403).

In step S403, the user information preprocessing unit 291 reads the user information of the designated third target user from the user information storage unit 28 and generates a user vector.

Next, in step S404, the user information feature extraction unit 292 performs feature extraction on the user vector generated in step S403 to generate a user feature vector.

Next, the second data processing unit 29 and the marriage score estimation unit 33 are in steps S406 to S409 for the number of opponent candidates (fourth target person) who estimate the marriage score for the third target person. The operation is repeated (step S405, step S410). Here, the fourth target person is, for example, a user who is a candidate for the other party of the third target person in the top N cases of the matching score read out in step S402, and more specifically, a member as user specific information. It may be a user indicated by an ID. For example, the marriage score estimation unit 33 may generate a list of member IDs of users who are the fourth target person when the third target person is designated. By adjusting the number of N cases, it is possible to adjust the number of members to be shown to the user as a candidate.

In the iterative process, first, the marriage score estimation unit 33 requests the second data processing unit 29 to sequentially specify users from the fourth target person list and generate a user feature vector (step S406). Proceed to).

In step S406, the user information preprocessing unit 291 reads the user information of the designated fourth target user from the user information storage unit 28 and generates a user vector.

In step S407, the user information feature extraction unit 292 extracts features from the user vector generated in step S406 and generates a user feature vector.

In step S408, the marriage score estimation unit 33 uses the model parameters read in step S401 and the user feature vectors generated in steps S404 and S407 to specify the third target person as the fourth target person. Calculate the marriage score between and.

In step S409, the marriage score estimation unit 33 writes the calculation result in step S408 into the second estimation result storage unit 34. The marriage score estimation unit 33 stores the calculation result in the form of <member ID (third target person identification information), member ID (fourth target person identification information), marriage score>, for example, in the second estimation result storage unit 34. You may write in.

FIG. 19 is a flowchart showing an example of the operation of the marriage simulation step of the marriage simulation system 100.

In the example shown in FIG. 19, first, the simulation unit 35 reads the marriage score written in step S409 from the second estimation result storage unit 34 (step S501).

Next, the simulation unit 35, the second data processing unit 29, and the marriage score estimation unit 33 are in step S503 for the number of opponent candidates (fourth target person) who estimate the estimated score for the third target person. -The operation of step S505 is repeated (step S502, step S506). Here, the number of repetitions may be the number of marriage scores read from the second estimation result storage unit 34 in step S501.

In step S503, the simulation unit 35 takes out one marriage score read in step S501, and changes the user information for the combination of users for which the marriage score is calculated by brute force. For example, even if the simulation unit 35 acquires the user feature vectors of each of the users whose marriage score is calculated, and changes the elements included in the pair of these user feature vectors in order. Good. Here, the target of one change is not limited to one element. Further, the target of the change may be limited to the elements that can be changed. When all the changes are completed, the process proceeds to step S505.

When the user information is changed, the simulation unit 35 requests the marriage score estimation unit 33 to carry out the above-mentioned marriage prediction step using the changed user information (step S504). The marriage score estimation unit 33 calculates the marriage score using the designated user information in response to the request from the simulation unit 35. However, the marriage score estimation unit 33 does not perform the operation of step S409 of the marriage prediction step, and outputs the calculated marriage score to the requester. Upon receiving the marriage score based on the changed user information, the simulation unit 35 compares the score with the previous score and retains the user information when the score is maximized.

When all the changes are completed (No in step S503), the user information with the maximum marriage score is compared with the user information before the change, and the difference is stored as an insufficient parameter (step S505).

When the above processing is completed for all the partner candidates, the simulation unit 35 ends the marriage simulation step.

FIG. 20 is an explanatory diagram showing another example of a method of displaying a simulation result. As a result of the simulation, if the marriage score improves by changing the user information of the user (third target person) who has given the desired conditions, the system will change the changed part and the changed contents as shown in FIG. It may be displayed to prompt the third subject to change his / her own parameters. That is, the simulation unit 35 may present to the user a profile in which the marriage rate with the desired partner is high after the change of the profile of the user so that the user can change his / her profile according to the partner. In the example shown in FIG. 20, as a result of simulation with a member of the opposite sex satisfying the desired condition, the profile when the marriage score becomes high is presented to the member together with the current profile as a profile to be improved. At this time, along with the changed desired condition, the information of the partner profile that matches the changed desired condition and the marriage score with the partner may be displayed together.

Further, FIG. 21 is an explanatory diagram showing another example of the method of displaying the simulation result. As a result of the simulation, when the marriage score is improved by changing the user information of the user (fourth target person) who is the other party of the third target person, the system changes the changed part and changes as shown in FIG. It is also possible to display the content and urge the third subject to change the desired conditions for the opposite sex. That is, in order for the user to have a high marriage rate even if he / she keeps his / her profile, the simulation unit 35 uses the profile of the other party whose marriage score becomes high due to the change of the desired conditions for the opposite sex after fixing his / her profile. May be presented to a person. In the example shown in FIG. 21, the desired condition when the marriage score becomes high as a result of the simulation with the opposite sex member by changing the desired condition is displayed. By doing so, members are urged to change their desired conditions in order to improve the marriage rate. At this time, along with the changed desired condition, the information of the partner profile that matches the changed desired condition and the marriage score with the partner may be displayed together.

Further, FIG. 22 is an explanatory diagram showing another example of the method of displaying the simulation result. 20 and 21 show an example of presenting the changed part and the changed content with respect to the current profile or the desired condition, but may further show the contribution of the changed part to the change. .. The contribution may be, for example, indicated by a model parameter, or is given by a method of giving a high contribution to a parameter having a higher marriage score as a result of the simulation unit 35 simulating the marriage score. It may be the one that has been done. In the example shown in FIG. 22, not only the value after the change of the profile proposed as the improvement point but also the degree of influence (contribution) of the improvement on the marriage score is presented to the user. By doing so, it is possible to prioritize one's own improvement points.

As described above, according to the present embodiment, the goal is set by presenting what the current member lacks for compatibility with the desired partner, or by presenting information such as specific changes and contributions. It can be performed. In addition, according to the present embodiment, when the marriage rate can be increased by changing a part of the desired conditions, it is possible to achieve a predetermined purpose (marriage) by proposing a change in such desired conditions. You can improve your sexuality.

Further, according to the present embodiment, matching and prediction of the marriage rate (calculation of a score based on actual results) can be objectively performed even by using a freely described text. In addition, in the method of selecting a candidate only by matching the desired conditions and the user information, the wishes of both parties may not always match, and there may be cases where the actual compatibility does not match. Most of the information to be input as desired conditions is in the selection format such as age, educational background, annual income, etc., and the person in charge visually checks the profile etc. described in text to determine whether the conditions are met. Was there. On the other hand, according to the present embodiment, it is possible to automatically select a partner candidate that reflects information on individuals and things using texts, and to perform matching (marriage rate) between information on individuals and things. Furthermore, the effect can be simulated and the result can be fed back to the user while changing at least a part of either one or both of the combined information.

The above configuration is just an example. For example, when the other party to perform the simulation is not particularly selected, that is, when the simulation is performed by round-robin, the desired condition storage unit 21, the first data processing unit 22, and the first The learning unit 24, the matching result storage unit 23, the matching learning model storage unit 25, the matching score estimation unit 26, and the first estimation result storage unit 27 of 1 can be omitted.

It is also possible to use it for learning and prediction (estimation of score) as it is without extracting features from the user vector and the desired condition vector.

In addition, depending on the calculation result of the marriage score with the opposite sex member with the highest matching score for the user who entered the desired conditions, the calculation result based on the combination of the current user information is displayed without performing the simulation. May just be. For example, when the average marriage score with a member of the opposite sex with a higher matching score is equal to or higher than a predetermined value, the simulation step can be omitted.

Also, regarding the parameter change, in the above example, the system changed the user information by brute force and searched for the parameter that maximizes the marriage rate, but the user can specify the changeable part. You may. In this case, a parameter change GUI unit that accepts the specification of the item to be simulated for parameter change from the user may be added to the above configuration.

In addition, the matching rate (matching score) also changes due to the change of the parameter, but by displaying the matching order for a certain partner, it is possible to grasp the relative position of the member. As a result, the motivation for improvement of the members themselves can be increased. In this case, a matching rank calculation unit for calculating the matching rank and an output unit for outputting the matching rank may be added to the above configuration.

In general, a large amount of member data accumulated in a dating agency or the like includes not only personal information of members but also information such as conditions of a desired partner and past marriage records. By utilizing these data, it is possible to automatically select a partner who meets the desired conditions and to simulate the marriage rate with a partner who meets the conditions to some extent without human intervention.

In the above description, an example of matching between members using the marriage counseling business and simulating the marriage rate by changing the personal information (including desired conditions) of the target member is shown, but the present invention presents the present invention. , Not limited to individuals and things, but can also be applied to combinations such as individuals and organizations, and individuals and things. For example, by setting the field as job hunting and converting the target of combination to job seekers and companies, it is expected that the part that the current job seeker lacks or the current job seeker will be hired for the company that the job seeker desires. It is possible to present other companies with high qualifications. Also, for example, the field may be an organizational activity, and the target of the combination may be transformed into an employee and a boss, a project team or an organization. In that case, by extracting the skills and talents that the current employees and the organization side (boss, project team or organization) lack, it can be utilized for employee training, movement, and organizational reform.

Although the present invention has been described above with reference to the embodiments and examples, the present invention is not limited to the above embodiments and examples. Various changes that can be understood by those skilled in the art can be made within the scope of the present invention in terms of the structure and details of the present invention.

The present invention not only improves the success rate for a predetermined purpose due to compatibility between people or things, but also improves the composition rate for a predetermined purpose in any combination of people and things or people and organizations, and therefore. It can also be suitably applied to setting improvement goals and improving motivation.

10 Success support system 11 Related information storage means 12 Proposal means 121 Success rate estimation means 122 Simulation means 123 Analysis means 15 Display means 13 Candidate selection means 14 Learning means 100 Marriage simulation system 21 Desired condition storage unit 22 First data processing unit 221 Desired condition preprocessing unit 222 Desired condition feature extraction unit 23 Matching performance storage unit 24 First learning unit 241 Matching performance preprocessing unit 242 Matching performance learning unit 25 Matching learning model storage unit 26 Matching score estimation unit 27 First estimation Result memory unit 28 User information storage unit 29 Second data processing unit 291 User information preprocessing unit 292 User information feature extraction unit 30 Marriage record storage unit 31 Second learning unit 311 Marriage record preprocessing unit 312 Marriage record Learning Department 32 Marriage Learning Model Memory Department 33 Marriage Score Estimating Department 34 Second Estimating Result Memory Department 35 Simulation Department

Claims (10)

An acquisition method for acquiring each attribute information of the matching target,
A presenting means for presenting a change content in at least one of the attribute information of the matching target and a success score for successful matching of the matching target.
With
The success score is an information processing device estimated based on the attribute information including the changed contents of each of the matching targets. The information processing according to claim 1, wherein the success score is estimated by a model generated by machine learning using the attribute information of each person or thing and the success record of successful matching between the person or thing. apparatus. Claim 1 in which the presenting means presents the changed contents and the success score based on the plurality of the changed contents and the success score corresponding to each of the attribute information including the plurality of changed contents. Alternatively, the information processing apparatus according to claim 2. When the other satisfies the condition specified by one of the matching targets, the presenting means changes at least a part of the attribute information regarding the one to improve the success score for which the matching of the matching targets succeeds. The information processing apparatus according to any one of claims 1 to 3, which is presented as the content. The presenting means changes at least one item of the other attribute information of the matching target that matches the condition specified by one of the matching targets, and improves the success score for which the matching of the matching target succeeds. The information processing apparatus according to any one of claims 1 to 4, which is presented as a change. The presenting means changes at least one of the attribute information possessed by at least one of the matching targets, which is designated as changeable, to improve the success score for which the matching of the matching targets succeeds. The information processing apparatus according to any one of claims 1 to 5, which is presented as. The matching target is a user who receives a matching service.
The information processing device according to any one of claims 1 to 6, wherein the success score is a marriage score between users. The information processing device according to any one of claims 1 to 7, wherein the matching target is a job seeker and a company. The computer
Get the attribute information of each matching target,
The change contents in at least one of the attribute information of the matching target and the success score for which the matching of the matching target is successful are presented.
The success score is a success support method estimated based on the attribute information including the change contents of each of the matching targets. On the computer
The process of acquiring the attribute information of each matching target,
A process of presenting a change content in at least one of the attribute information of the matching target and a success score for successful matching of the matching target is executed.
The success score is a success support program estimated based on the attribute information including the changed contents of each of the matching targets.

Images