JPWO2013088682A1 - Recommendation condition correcting device, recommended condition correcting method, and recommended condition correcting program - Google Patents

Abstract

  An apparatus capable of effectively correcting recommended conditions is provided. The apparatus stores a recommendation condition for designating a range of attribute values of the user's context, and each attribute value of the target user's context and the target user are predetermined for each response of the target user to which the recommendation information is output. The recommendation history including success / failure information indicating whether or not the action is taken is stored, and the value of each attribute of the context of the successful user is stored in correspondence with each successful user. Furthermore, the device has a recommendation history ratio indicating that a predetermined action is taken out of the recommendation history in which the attribute of the context is the value corresponding to each value of each attribute included in the context of all the recommendation histories. For each attribute in the recommendation condition, the ratio of the success history in which the value of the attribute in the context matches the recommendation condition is calculated from the success history, and the recommendation condition is corrected based on both ratios.

Description

The present invention relates to a recommended condition correcting device, a recommended condition correcting method, and a recommended condition correcting program.

  There is a recommendation system for sending recommendation information, for example, an advertisement, which is information for encouraging people to perform a predetermined action such as product purchase. The recommendation system selects people who are likely to perform a predetermined action and tries to send recommendation information. The recommendation system uses a recommendation condition to select a recommendation target person.

  There are two recommended performance indicators of the recommendation system: “recommendation success rate” and “recommendation success number”. The recommendation success rate is a ratio of users who have taken a predetermined action with respect to all users who have received recommendation information. The recommendation success rate is, for example, a ratio of users who have purchased the product with respect to all users who have received the advertisement information of the product. The number of successful recommendations is the number of users who have taken a predetermined action after receiving recommendation information.

  Non-Patent Document 1 discloses a recommendation system that measures improvement in “recommendation success rate” and “recommendation success number”. The recommendation system described in Non-Patent Document 1 makes recommendations to a larger number of users by relaxing the original recommendation conditions in order to improve the “number of successful recommendations”. The system then records a recommendation history consisting of user information at the time of recommendation and success / failure information indicating whether or not a predetermined action was taken, and using the recommendation history, a recommendation condition for improving the “recommendation success rate” Implement the correction.

  Patent Document 1 discloses an apparatus for optimizing an advertisement according to an advertisement result while performing an advertisement. When the sales amount has not reached the target, this device expands the target audience of the advertisement.

  Patent Document 2 discloses a business analysis table creation system for recognizing the effect of an advertising medium. This business analysis table displays questionnaire survey responses in a matrix table. This business analysis table includes the number of people who answered that the target product was advertised with a flyer but wanted to purchase the product. Patent Literature 3 discloses a recommendation effect measuring device for an online shopping server.

JP 2010-39095 JP 2004-258996 A JP 2006-260135 A

Feedback Information Collection Architecture Based on Scenario in Action Recommendation System, Multimedia, Distributed, Collaboration and Mobile (DICOMO2011) Symposium, pp.1461-1466, 2011-07

  None of the techniques disclosed in the above documents disclose a method for effectively correcting the recommendation condition. In the recommendation system of Non-Patent Document 1, if the relaxation is not properly performed once in order to relax the recommendation conditions, there is a possibility that the “recommendation success rate” may once decrease until the recommendation history is sufficiently collected. . The device of Patent Document 1 does not disclose a specific method for expanding the target audience of advertisements.

  An object of the present invention is to provide a recommended condition correction device, a recommended condition correction method, and a recommended condition correction program that solve the above-described problems.

The recommendation condition correction device according to an embodiment of the present invention includes a condition storage unit that stores a recommendation condition that specifies a range of one or more values of the attribute that is user information including a plurality of attributes,
When the context satisfies the recommendation condition, each attribute value of the target user and the target user correspond to each target user who is a user who has output recommendation information for encouraging a predetermined action. A recommendation history storage means for storing a recommendation history including success / failure information indicating whether or not a predetermined action has been taken;
A success history storage means for storing the value of each attribute of the context of the successful user and a recommendation history storage means for each successful user who is a user who has taken the predetermined action other than the target user. The value indicating that the attribute of the context is the value corresponding to each value of the attribute included in the context of all of the recommendation histories being performed indicates that the predetermined action has been taken in the recommendation history A first contribution ratio indicating a ratio of recommendation history is calculated, and a ratio of the success history in which the value of the attribute in the context matches the recommendation condition in the success history for each attribute in the recommendation condition And a recommendation condition correction means for correcting the recommendation condition based on the first contribution ratio and the second contribution ratio.

The recommendation condition correction method according to an embodiment of the present invention stores a recommendation condition that specifies a range of one or more values of the attribute that is user information including a plurality of attributes in a condition storage unit,
When the context satisfies the recommendation condition, each attribute value of the target user and the target user correspond to each target user who is a user who has output recommendation information for encouraging a predetermined action. Storing a recommendation history including success / failure information indicating whether or not a predetermined action has been taken in the recommendation history storage means;
In response to each successful user who is the user who has taken the predetermined action other than the target user, the value of each attribute of the context of the successful user is stored in a success history storage unit, and stored in the recommendation history storage unit The value indicating that the attribute of the context is the value corresponding to each value of the attribute included in the context of all of the recommendation histories being performed indicates that the predetermined action has been taken in the recommendation history A first contribution ratio indicating a ratio of recommendation history is calculated, and a ratio of the success history in which the value of the attribute in the context matches the recommendation condition in the success history for each attribute in the recommendation condition A second contribution ratio indicating the second contribution ratio is calculated, and the recommendation condition is corrected based on the first contribution ratio and the second contribution ratio.

A recommendation condition correction program according to an embodiment of the present invention includes a condition storage process for storing a recommendation condition that specifies a range of one or more values of a context that is user information, including a plurality of attributes, as condition storage means. ,
When the context satisfies the recommendation condition, each attribute value of the target user and the target user correspond to each target user who is a user who has output recommendation information for encouraging a predetermined action. A recommendation history storage process for storing a recommendation history including success / failure information indicating whether or not a predetermined action has been taken in the recommendation history storage means;
Processing for storing each attribute value of the context of the successful user in a success history storage unit in response to each successful user who is the user who has taken the predetermined action other than the target user, and the recommendation history storage unit In response to each value of each of the attributes included in the context of all the recommendation histories stored in the context, the predetermined action is taken out of the recommendation history in which the attribute of the context is the value Calculating a first contribution ratio indicating a ratio of the recommendation history to be shown, and corresponding to each attribute in the recommendation condition, the success history in which the value of the attribute of the context matches the recommendation condition in the success history And calculating a second contribution ratio indicating a ratio of the first contribution ratio and correcting the recommendation condition based on the first contribution ratio and the second contribution ratio on a computer. To.

  ADVANTAGE OF THE INVENTION According to this invention, the apparatus which can correct recommendation conditions effectively is realizable. For example, the device can improve the “number of successful recommendations” while suppressing a decrease in the “recommendation success rate”.

It is a block diagram of recommendation condition correction system 90 of a 1st embodiment. It is a figure which shows an example of the scenario 12 of 1st Embodiment. It is a figure which shows an example of the context 13 of 1st Embodiment. It is a figure which shows an example of the condition information 41 of 1st Embodiment. It is a figure which shows an example of the some condition information 41 memorize | stored in the condition memory | storage part 40 of 1st Embodiment. It is a figure which shows an example of the some recommendation log | history 51 memorize | stored in the recommendation log | history memory | storage part 50 of 1st Embodiment. It is a figure which shows an example of the several success log | history 31 memorize | stored in the success log | history memory | storage part 30 of 1st Embodiment. It is a figure which shows an example of the recommendation condition correction request | requirement 14 of 1st Embodiment. It is a figure which shows an example of the 1st success contribution rate computed from the recommendation log | history 51 in 1st Embodiment. In 1st Embodiment, it is a figure which shows an example of the 2nd success contribution rate computed from the success log | history 31. FIG. It is a scenario registration flowchart which shows operation | movement of the system concerning 1st Embodiment. It is a recommendation condition correction flowchart (1/2) which shows operation | movement of the system concerning 1st Embodiment. It is a recommendation condition correction flowchart (2/2) which shows operation | movement of the system concerning 1st Embodiment. It is a processing flowchart of deletion attribute information judgment of a recommendation condition of a 1st embodiment. It is a processing flowchart of additional attribute information determination of recommendation conditions of a 1st embodiment. It is a block diagram of the recommendation condition correction system 90 of 2nd Embodiment. It is a figure which shows an example of the recommendation log | history 51 of 2nd Embodiment. It is a figure which shows an example of the success log | history 31 of 2nd Embodiment. It is a context process flowchart which shows operation | movement of the system concerning 2nd Embodiment. It is a success history process flowchart which shows operation | movement of the system concerning 2nd Embodiment. It is a block diagram of the recommendation condition correction system 90 of 3rd Embodiment. It is a figure which shows an example of the scenario 12 of 3rd Embodiment. It is a figure which shows an example of the several condition information 41 memorize | stored in the condition memory | storage part 40 of 3rd Embodiment. It is a figure which shows an example of the context 13 of 3rd Embodiment. It is a figure which shows an example of the context 13 of 3rd Embodiment. It is a figure which shows an example of the condition information 41 of 3rd Embodiment. It is a figure which shows an example of the context 13 of 3rd Embodiment. It is a figure which shows an example of the success log | history 31 of 3rd Embodiment. It is a context process flowchart which shows operation | movement of the system concerning 3rd Embodiment. It is a block diagram of the recommendation condition correction system 90 of 3rd Embodiment.

<First Embodiment>
Embodiments for carrying out the present invention will be described in detail with reference to the drawings.

  FIG. 1 is a block diagram of a recommended condition correction system 90 according to the first embodiment of this invention. A recommendation condition correction system 90 shown in FIG. 1 includes a recommendation condition correction device 10, an administrator terminal 11, a recommendation execution device 17, and a user terminal 15.

  The recommended condition correction device 10 of the present embodiment includes a registration unit 60, a condition storage unit 40, a recommended condition correction unit 20, a recommendation history storage unit 50, and a success history, each including a logic circuit, a storage device, and the like. A storage unit 30 is included. The recommended condition correction apparatus 10 may be a computer that operates under program control. In this case, the registration unit 60, the condition storage unit 40, the recommended condition correction unit 20, the recommendation history storage unit 50, and the success history storage unit 30 are programs stored in the storage device by the processing device included in the computer. It may be realized by reading and executing. The recommendation history storage unit 50, the condition storage unit 40, and the success history storage unit 30 may include a disk device included in the computer.

  The registration unit 60 receives the scenario 12 from the administrator terminal 11 and registers the condition information 41 in the condition storage unit 40. FIG. 2 shows an example of the scenario 12. The scenario 12 includes a scenario ID, recommendation conditions, and recommendation information. The scenario ID is information for distinguishing the scenario 12. The recommendation condition is information including a recommendation condition ID and a conditional expression. The recommendation condition ID is information for distinguishing the recommendation condition. The conditional expression expresses a condition for selecting a user to be recommended using part or all of the user context 13.

  Here, the context 13 is information indicating the state of the user and includes a plurality of attribute information. The attribute information includes a pair of attribute name and attribute value. The attribute name indicates the property of the attribute, and is, for example, a user ID, a position, and an age. The attribute value indicates the value of the attribute, for example, Ichiro Tanaka, XX station, 30 years old.

  FIG. 3 shows an example of the context 13. The contexts 13 shown in FIG. 3 are six such as “user ID = user 1”, “position = station”, “age = 50s”, “sex = male”, “step count = 1000”, and “health state = good”. This is a context 13 including the attribute information. This context 13 indicates that “User 1 who is a man in his fifties is currently at the station, has 1000 steps, and is in good health”.

  The conditional expression represents the condition of the context 13 that should be satisfied by the user to be recommended by one or more attribute information, that is, one or more attribute name / attribute value pairs. The conditional expression shown in FIG. 2 indicates that a user having a context 13 with “position = station”, “age = 50s”, and “health condition = good” is a recommendation target. The attribute value may be a range of values. For example, the attribute information in the conditional expression may be “age = 50 to 54” or the like.

  The recommendation information is information related to the recommendation. For example, the recommendation information may be a recommendation information ID indicating recommendation content, an address of the recommendation execution device 17 that performs the recommendation, an advertisement that is presented to a user that matches the conditional expression of the recommendation condition, and a program that is executed. FIG. 2 shows a scenario 12 when the recommendation information is a recommendation information ID.

  The condition information 41 is information including a condition ID, a scenario ID, a conditional expression, and association information. FIG. 4 shows an example of the condition information 41. The registration unit 60 creates the condition information 41 from the scenario 12. The registration unit 60 acquires a recommendation condition ID, a scenario ID, a conditional expression, and recommendation information from the scenario 12, and sets the information as the condition ID, the scenario ID, the conditional expression, and the association information of the condition information 41.

  The condition storage unit 40 stores one or more condition information 41. FIG. 5 shows an example of a plurality of condition information 41 stored in the condition storage unit 40. FIG. 5 shows an example in which the condition storage unit 40 stores two condition information 41 having a condition ID of recommendation condition 1 and recommendation condition 2.

  The recommendation history storage unit 50 includes the context 13 at the time of recommendation of the user recommended by the recommendation execution device 17 and whether the user has taken the expected behavior described later (success) or not yet (unsuccessful). One or more recommendation histories 51 including recommendation success / failure information indicating “” are stored.

  FIG. 6 shows an example of a plurality of recommendation histories 51 stored in the recommendation history storage unit 50. FIG. 6 shows a recommendation history 51 for the recommendation condition 1. The example of FIG. 6 shows that the recommendation execution device 17 made the recommendation four times, and the effect expected only once was obtained. In addition, this example shows that the context 13 when the user who received the expected effect received the recommendation was “position = station, age = 50s, gender = female, number of steps = 2000”. .

  For example, the recommendation execution device 17 receives the user context 13 from the user terminal 15, refers to the condition storage unit 40 included in the recommendation condition correction device 10, transmits recommendation information to the user terminal 15, and recommends The history 51 is written into the recommendation history storage unit 50. The recommendation condition correction apparatus 10 may acquire the recommendation history 51 collectively from the recommendation execution apparatus 17 or the like by other means, for example, a data migration tool or the like.

  The success history storage unit 30 stores a plurality of success histories 31 including the context 13 at the time of the expected behavior of the user who is not recommended by the recommendation execution device 17.

  The expected behavior is an effect or reaction expected from the user when the system operator who registered the scenario 12 in the recommended condition correction system 90 receives a recommendation from the scenario 12. For example, when an advertisement is sent to a user of a specific age group and gender and a purchase of a product is recommended, the expected behavior is that the user purchases the product or the user stops at a store that sells the product. is there.

  In the recommendation service, it is often the case that a user who does not make a recommendation takes an expected action. For example, the recommendation execution device 17 may have sent an advertisement for recommending a product to a woman in her 20s, but a woman in her 30s who did not send an advertisement bought the product.

  The success history storage unit 30 accumulates the context 13 at the time of the expected behavior of the user who has taken the expected behavior even though such recommendation is not performed. FIG. 7 shows an example of a plurality of success histories 31 stored in the success history storage unit 30. FIG. 7 shows a success history 31 for the recommendation condition 1. FIG. 7 shows that 10 users have taken expected actions despite not receiving recommendations.

  The success history 31 is transmitted, for example, from the sales terminal of the store to the recommended condition correction device 10. The success history 31 may be collectively transferred from the sales log of the sales terminal of the store by other means such as a data migration tool. The recommendation condition correction apparatus 10 can also acquire the recommendation success / failure information of the recommendation history 51 in the same manner.

  When the recommendation condition correction unit 20 receives the recommendation condition correction request 14 from the administrator terminal 11, the recommendation condition correction unit 20 corrects the conditional expression of the recommendation condition of the recommendation condition ID specified by the recommendation condition correction request 14. The recommendation condition correction unit 20 acquires the success history 31 and the recommendation history 51 from the success history storage unit 30 and the recommendation history storage unit 50, acquires the conditional expression from the condition storage unit 40, and corrects it.

  The recommendation condition correction request 14 includes a recommendation condition ID, an attribute information deletion threshold value, and an attribute information addition threshold value. FIG. 8 shows an example of the recommended condition correction request 14. The attribute information addition threshold is a lower limit value for adding attribute information when the first success contribution rate of each attribute information calculated from the recommendation history 51 is calculated. The recommendation condition correction unit 20 adds attribute information whose first success contribution rate is above the attribute information addition threshold.

  The attribute information deletion threshold is a lower limit value that does not delete the attribute information when the second success contribution rate of each attribute information calculated from the success history 31 is calculated. The recommendation condition correction unit 20 deletes attribute information whose second success contribution rate is equal to or less than the attribute information deletion threshold.

  When receiving the recommendation condition modification request 14, the recommendation condition modification unit 20 creates an acquisition request for the condition information 41, an acquisition request for the success history 31, and an acquisition request for the recommendation history 51. The acquisition request for the condition information 41, the acquisition request for the success history 31, and the acquisition request for the recommendation history 51 include a condition ID.

  The recommended condition correction unit 20 transmits an acquisition request for the condition information 41 to the condition storage unit 40, and acquires the condition information 41 as a response. Also, the recommendation condition correction unit 20 transmits an acquisition request for the success history 31 and an acquisition request for the recommendation history 51 to the success history storage unit 30 and the recommendation history storage unit 50, and the success history 31 and the recommendation history 51 that are responses thereto. Receive.

  Then, the recommendation condition correction unit 20 calculates the first and second success contribution ratios of the attribute information using the success history 31 and the recommendation history 51, and adds the attribute information to the conditional expression of the condition information 41 and Perform deletion.

  An example of calculating the first success contribution rate will be described. The first success contribution rate is used for adding attribute information.

  The recommendation condition correction unit 20 extracts all different attribute information, that is, attribute name / attribute value pairs, as a list from all the recommendation histories 51 stored in the recommendation history storage unit 50. For each piece of attribute information in the list, the recommendation condition correcting unit 20 out of the recommendation history 51 including the attribute information in the context 13, the number of recommended histories 51 (success number) with a success or failure of recommendation and “unsuccessful”. Is obtained (the number of unsuccessful attempts). Then, the recommendation condition correction unit 20 calculates the number of successes / (number of successes + number of unsuccessful) for each attribute information in the list and sets it as the first success contribution rate.

  FIG. 9 shows the first success contribution rate calculated from the four recommendation histories 51 shown in FIG. When the recommendation condition correction unit 20 receives the recommendation condition correction request 14 shown in FIG. 8, the recommendation condition correction unit 20 adds attribute information “step count = 2000” exceeding the attribute information addition threshold of 80% to the recommendation condition.

    Next, an example of calculating the second success contribution rate will be described. The second success contribution rate is used for deleting attribute information.

  The recommendation condition correction unit 20 acquires a conditional expression corresponding to the recommendation condition ID included in the recommendation condition correction request 14 from the condition storage unit 40, and includes the attribute information in the context 13 for each attribute information included in the conditional expression. The number of success histories 31 (number of matches) is obtained. The recommendation condition correction unit 20 calculates the number of matches / total number of success histories 31 for each attribute information, and sets it as the second success contribution rate.

  FIG. 10 shows the second success contribution rate calculated from the ten success histories 31 shown in FIG. When the recommendation condition correction unit 20 receives the recommendation condition correction request 14 illustrated in FIG. 8, the recommendation condition correction unit 20 deletes the attribute information “age = 50s” lower than the attribute information deletion threshold 20% from the recommendation condition.

  The administrator terminal 11 is an information processing apparatus including an input device such as a keyboard and a mouse, an output device such as a liquid crystal display, a processing device that operates by program control, and a storage device including a memory. The administrator terminal 11 creates a scenario 12 and transmits it to the recommended condition correction device 10. Further, the administrator terminal 11 creates a recommended condition correction request 14 and transmits it to the recommended condition correction apparatus 10.

  The recommendation execution device 17 is an information processing device including an input device such as a keyboard and a mouse, an output device such as a liquid crystal display, a processing device operated by program control, and a storage device including a memory. The recommendation execution device 17 receives the information including the recommendation condition and the recommendation content, and transmits the recommendation information to the user.

  Next, the overall operation of the present exemplary embodiment will be described in detail with reference to a flowchart. The operation of the system according to the present embodiment can be divided into a scenario registration flow for registering the scenario 12 and a recommendation condition correction flow for correcting the recommendation conditions.

  FIG. 11 is a scenario registration flowchart showing the operation of the system according to the first embodiment.

  In S-101S, the registration unit 60 receives the scenario 12 and proceeds to S-102S. In S-102S, the registration unit 60 acquires a scenario ID, a recommendation condition ID, a conditional expression, and recommendation information from the scenario 12, and sets the condition information 41 using the respective information as a scenario ID, a condition ID, a conditional expression, and association information. Create and go to S-103S.

  In S-103S, the registration unit 60 transmits the condition information 41 to the condition storage unit 40, and the condition storage unit 40 stores the condition information 41 and ends the scenario registration flow. With this process, the linking information is stored in the linking information storage unit.

  12A and 12B are recommended condition correction flowcharts showing the operation of the system according to the first embodiment.

  In S-1010, the recommended condition correction unit 20 receives the recommended condition correction request 14 from the administrator terminal 11, and proceeds to S-1020. The recommendation condition correction request 14 includes a recommendation condition ID, an attribute information deletion threshold value, and an attribute information addition threshold value.

  In S-1020, the recommended condition modification unit 20 creates a condition acquisition request, and proceeds to S-1030. The condition acquisition request includes a recommended condition ID.

  In S-1030, the recommended condition correction unit 20 transmits a condition acquisition request to the condition storage unit 40. When the condition storage unit 40 receives the condition acquisition request, the condition storage unit 40 returns condition information 41 having a condition ID that matches the recommended condition ID included in the condition acquisition request to the recommended condition correction unit 20. If there is no condition information 41 having a matching condition ID, the condition storage unit 40 returns “no match”. The recommended condition correction unit 20 receives the condition information 41 or “no match” from the condition storage unit 40, and proceeds to S-1040.

  In S-1040, the recommended condition correction unit 20 proceeds to S-1050 if the condition information 41 is received from the condition storage unit 40 in S-1030, and ends the process if “no match” is received.

  In S-1050, the recommendation condition correction unit 20 creates an acquisition request for the success history 31, and proceeds to S-1060. The acquisition request for the success history 31 includes a recommendation condition ID.

  In S-1060, the recommendation condition correction unit 20 transmits a success history 31 acquisition request to the success history storage unit 30. When the success history storage unit 30 receives the acquisition request for the success history 31, the success history storage unit 30 returns a success history 31 having a condition ID that matches the recommendation condition ID included in the acquisition request for the success history 31 to the recommendation condition correction unit 20. If there is no success history 31 having a matching condition ID, the success history storage unit 30 returns “no match”. The recommendation condition correction unit 20 receives the success history 31 or “no match” from the success history storage unit 30, and proceeds to S-1070.

  In S-1070, the recommendation condition correction unit 20 creates an acquisition request for the recommendation history 51, and proceeds to S-1080. The acquisition request for the recommendation history 51 includes a recommendation condition ID.

  In S-1080, the recommendation condition correction unit 20 transmits a recommendation history 51 acquisition request to the recommendation history storage unit 50. Upon receiving the recommendation history 51 acquisition request, the recommendation history storage unit 50 returns a recommendation history 51 having a condition ID that matches the recommendation condition ID included in the recommendation history 51 acquisition request to the recommendation condition correction unit 20. If there is no recommendation history 51 having a matching condition ID, the recommendation history storage unit 50 returns “no match”. The recommendation condition correction unit 20 receives the recommendation history 51 or “no match” from the recommendation history storage unit 50, and proceeds to S-1090.

  In S-1090, the recommendation condition correction unit 20 proceeds to S-1100 if the success history 31 is received from the success history storage unit 30 in S-1060, and proceeds to S-1110 if “no match” is received.

  In S-1100, the recommendation condition correction unit 20 performs deletion attribute information determination, and proceeds to S-1110. Detailed operation of attribute information deletion will be described later with reference to FIG.

  In S-1110, the recommendation condition correction unit 20 proceeds to S-1120 if the recommendation history 51 is received from the recommendation history storage unit 50 in S-1080, and proceeds to S-1130 if “no match” is received.

  In S-1120, the recommendation condition correction unit 20 performs additional attribute information determination, and proceeds to S-1130. The detailed operation for adding attribute information will be described later with reference to FIG.

  In S-1130, the recommendation condition correction unit 20 deletes the attribute information determined as the deletion attribute information, adds the attribute information determined as the additional attribute information, and ends the process.

  Next, the details of the deletion attribute information determination and the additional attribute information determination performed in S-1100 and S-1120 of the recommendation condition correction flow will be described using a flowchart.

  FIG. 13 is a flowchart of a process for determining deletion attribute information of a recommendation condition.

  In S-101D, the recommended condition correction unit 20 extracts a conditional expression from the condition information 41, and proceeds to S-102D. In S-102D, the recommendation condition correction unit 20 sets 0 to the counter i, and proceeds to S-103D.

  In S-103D, the recommended condition correction unit 20 calculates the second success contribution rate of the i-th attribute information of the conditional expression from the success history 31, and proceeds to S104. The recommendation condition correction unit 20 counts the number of success histories 31 including the i-th attribute information in the context 13 in the success history 31 for all the success histories 31 stored in the success history storage unit 30. The second success contribution rate is obtained by calculating the number of counts / success history 31.

  In S-104D, the recommended condition correction unit 20 proceeds to S-106D if the calculation of the second success contribution rate has been completed for the attribute information of all the conditional expressions, and proceeds to S-105D if it has not been completed.

  In S-105D, the recommended condition correction unit 20 adds 1 to the counter i and proceeds to S-103D.

  In S-106D, the recommendation condition correction unit 20 determines that attribute information below the attribute information deletion threshold among the second success contribution rates calculated in S-103D is deletion attribute information, and ends the process.

  FIG. 14 is a flowchart of processing for determining additional attribute information of recommendation conditions.

  In S-101A, the recommended condition correction unit 20 sets 0 to the counter i, and proceeds to S-102A.

  In S-102A, the recommendation condition correction unit 20 acquires the i-th recommendation history 51 from the recommendation history 51 acquired from the recommendation history storage unit 50, and proceeds to S-103A.

  In S-103A, the recommended condition correcting unit 20 sets 0 to the counter j, and proceeds to S-104A.

  In S-104A, the recommendation condition correction unit 20 proceeds to S-105A if the success / failure information of the i-th recommendation history 51 acquired in S-102A is “success”, and proceeds to S-108A if “unsuccessful”. .

  In S-105A, the recommendation condition correction unit 20 acquires the j-th attribute information of the i-th recommendation history 51, adds 1 to the success number of the attribute information, and proceeds to S-106A.

  In S-106A, the recommendation condition correcting unit 20 proceeds to S-111A if the processing of S-105A is completed for all the attribute information of the i-th recommendation history 51, and proceeds to S-107A if not completed.

  In S-107A, the recommended condition correction unit 20 adds 1 to the counter j and proceeds to S-105A.

  In S-108A, the recommendation condition correction unit 20 acquires the j-th attribute information of the i-th recommendation history 51, adds 1 to the unsuccessful number of attribute information, and proceeds to S-109A.

  In S-109A, the recommendation condition correction unit 20 proceeds to S-111A if the processing of S-108A is completed for all the attribute information of the i-th recommendation history 51, and proceeds to S-110A if not completed.

  In S-110A, the recommended condition correction unit 20 adds 1 to the counter j and proceeds to S-108A.

  In S-111A, the recommendation condition correction unit 20 proceeds to A-113A if the processing has been completed for all the recommendation histories 51, and proceeds to S-112A if it has not been completed.

  In S-112A, the recommended condition correction unit 20 adds 1 to the counter i and proceeds to S-102A.

  In S-113A, the recommendation condition correction unit 20 calculates the first success contribution rate for each attribute information, and proceeds to S114A. The recommendation condition correcting unit 20 calculates the first success contribution rate by calculating the number of successes / (the number of successes + the number of unsuccessful) for each attribute.

  In S-114A, the recommendation condition correction unit 20 determines that attribute information in which the first success contribution rate exceeds the attribute information addition threshold among the attribute information is additional attribute information, and ends the processing.

  The recommendation condition correction apparatus 10 according to the present embodiment can perform deletion attribute information determination without using the recommendation history 51 while assuming feedback correction of the recommendation condition based on the recommendation history 51. As a result, the recommendation condition correction system 90 increases the number of recommendations to improve the “number of successful recommendations”, while reducing the “recommendation success rate” until the recommendation history 51 is accumulated and fed back. Can be suppressed. The reason is that the recommendation condition correction apparatus 10 performs the deletion attribute information determination using the success history 31 accumulated in the success history storage unit 30.

<Modification of First Embodiment>
The recommended condition correcting apparatus 10 according to the first embodiment is based on the assumption that there can be a plurality of recommended conditions. The recommended condition correction apparatus 10 may be implemented on the assumption that there is one recommended condition. In this case, for example, the recommendation history 51 and the success history 31 do not need to include the condition ID, and the recommendation condition correction unit 20 or the like does not need to perform processing for checking the condition ID.

  In addition, the recommendation condition correction apparatus 10 may perform the deletion attribute information determination in consideration of the recommendation success rate calculated from the recommendation history 51. Here, the recommendation success rate is a ratio of the number of recommendation histories 51 whose recommendation success / failure information indicates “success” in the recommendation histories 51 stored in the recommendation history storage unit 50. Specifically, the recommendation condition correction unit 20 may calculate a recommendation success rate, and may perform deletion attribute information determination similar to that of the first embodiment only when this value is equal to or less than a predetermined threshold.

  With this mechanism, the recommendation condition correcting apparatus 10 can avoid the risk of correcting the recommendation condition and lowering the recommendation success rate when the recommendation success rate is sufficiently high.

  Furthermore, the recommendation condition correction apparatus 10 may add that the second contribution ratio of the attribute information to be added is a predetermined value or more to the attribute addition condition in the additional attribute information determination. With this mechanism, the recommendation condition correction apparatus 10 can carefully add attribute information.

<Second Embodiment>
Next, a second embodiment of the present invention will be described in detail with reference to the drawings.

  FIG. 15 is a block diagram of a recommendation condition correction system 90 according to the second embodiment of this invention. A recommendation condition correction system 90 illustrated in FIG. 15 includes a recommendation condition correction device 10, an administrator terminal 11, a user terminal 15, a recommendation execution device 17, and a success history input terminal 16.

  The recommended condition correction apparatus 10 of the present embodiment includes a registration unit 60, a condition storage unit 40, a search unit 70, a recommendation history storage unit 50, and a success history registration unit, which are configured with logic circuits, storage devices, and the like. 80, a success history storage unit 30, and a recommendation condition correction unit 20. The recommended condition correction apparatus 10 may be a computer that operates under program control. In this case, the registration unit 60, the condition storage unit 40, the search unit 70, the recommendation history storage unit 50, the success history registration unit 80, the success history storage unit 30, and the recommendation condition correction unit 20 are provided in the computer. The central processing unit may be realized by reading and executing a program stored in the storage device. The recommendation history storage unit 50, the condition storage unit 40, and the success history storage unit 30 may include a disk device included in the computer.

  The registration unit 60, condition storage unit 40, recommendation history storage unit 50, success history storage unit 30, and recommended condition modification unit 20 in the second embodiment of the present invention are the same as those in the first embodiment. Hereinafter, the search unit 70 and the success history registration unit 80, which are differences from the first embodiment, will be described.

  The search unit 70 receives the user context 13 from the user terminal 15, acquires the condition information 41 that matches the context 13 from the condition storage unit 40, creates a recommendation history 51, and uses the condition information 41 as a recommendation execution device. 17 to send.

  For example, when a plurality of condition information 41 shown in FIG. 5 is stored in the condition storage unit 40, when the search unit 70 receives the context 13 shown in FIG. 3, the context 13 is “location = station, age = 50s”. , Condition information 41 matching “sex = male, number of steps = 1000, health condition = good” is acquired.

  The acquired condition information 41 is condition information 41 with a condition ID of recommendation condition 1, a scenario ID of scenario 1, and a conditional expression of “position = station & age = 50s & health condition = good”. It is.

  The search unit 70 creates a recommendation history 51 from the context 13 and the condition information 41 and stores it in the recommendation history storage unit 50. Here, the search unit 70 sets the condition ID obtained from the condition information 41 and the context 13 received from the user terminal 15 to the condition ID and context 13 of the recommendation history 51 to be created. The search unit 70 sets “unsuccessful” information in the recommendation success / failure information of the recommendation history 51. For example, when the context 13 shown in FIG. 3 and the condition information 41 shown in FIG. 4 are obtained, the search unit 70 generates the recommendation history 51 shown in FIG.

  Further, the search unit 70 transmits the condition information 41 to the recommendation execution device 17. The recommendation execution device 17 executes recommendation for the user terminal 15.

  The success history registration unit 80 receives the success history 31 from the success history input terminal 16 and determines whether the recommendation history 51 corresponding to the success history 31 exists in the recommendation history storage unit 50.

  The success history 31 includes a recommendation condition ID and an expected action context 13 and is received when the user takes an action expected by the system operator.

  FIG. 17 shows an example of the success history 31. The success history 31 shown in FIG. 17 is the success history 31 of the user 1 for the scenario 12 having the recommendation condition indicated by the condition ID of the recommendation condition 1. The success history 31 indicates that the context 13 at the time of expected behavior is “position = park, age = 50s, gender = male, number of steps = 3000, health condition = good”.

  When the recommendation history storage unit 50 stores the recommendation history 51 shown in FIG. 6 and the success history registration unit 80 receives the success history 31 shown in FIG. 17, the recommendation history 51 that matches the condition ID and the user ID. Is found. When the recommendation history 51 corresponding to the success history 31 is found, the success history registration unit 80 updates the recommendation success / failure information of the recommendation history 51 of the recommendation history storage unit 50 corresponding to the success history 31 to “success”.

  For example, when the recommendation history storage unit 50 stores the recommendation history 51 illustrated in FIG. 6 and the success history registration unit 80 receives the success history 31 illustrated in FIG. 17, the success history registration unit 80 includes the success history 31. The recommendation success / failure information of the recommendation history 51 that matches “condition ID = recommendation condition 1, user ID = user 1” is updated from unsuccessful to successful. This recommendation history 51 includes a recommendation that “condition ID is recommendation condition 1, recommendation context 13 is user ID = user 1, position = station, age = 50s, gender = male, number of steps = 1000, health condition = good”. This is a history 51.

  When the recommendation history 51 corresponding to the success history 31 is not found, the success history registration unit 80 stores the success history 31 in the success history storage unit 30.

  The user terminal 15 is an information processing apparatus including an input device such as a button, a GPS (Global Positioning System) sensor, an acceleration sensor, and a microphone, an output device such as a liquid crystal display, a processing device, and a storage device such as a memory. is there.

  The user terminal 15 transmits the context 13 to the recommended condition correction device 10 every time there is a change in the value of the input device or at regular intervals. The context 13 is a position acquired by the sensor, a health state input from the input device, or a gender value stored in the storage device. Note that a part of the context 13 such as gender may be stored in the search unit 70 or the like as user profile information in association with the user ID or the like. In this case, for example, the search unit 70 adds the stored attribute information to the attribute information received from the user terminal 15 to complete the context 13.

  The success history input terminal 16 is an information processing device including an input device such as a keyboard and a mouse, an output device such as a liquid crystal display, a processing device operated by program control, and a storage device including a memory. 31 is created and transmitted to the recommended condition correction apparatus 10.

  The success history input terminal 16 is, for example, a POS terminal (Point of Sales) installed in a store that sells products recommended by the recommendation information. The success history input terminal 16 can receive the context 13 from the user terminal 15 possessed by the user who visited the store, for example, by proximity communication. In addition, the success history input terminal 16 stores, for example, correspondence information between sold product codes and recommendation condition IDs, and may create the success history 31 when inputting the sales of products.

  Next, the overall operation of the present exemplary embodiment will be described in detail with reference to a flowchart. The operation of the system according to the present embodiment includes a scenario registration flow for registering a scenario 12, a context processing flow performed every time a user context 13 is received, a success history processing flow performed every time a success history 31 is received, and a recommendation It can be divided into a recommended condition correction flow for correcting conditions.

  The scenario registration flow and the recommendation condition correction flow are the same as those in the first embodiment. Hereinafter, the user context processing flow and the success history processing flow will be described.

  FIG. 18 is a context processing flowchart showing the operation of the system according to the second embodiment.

  In S-201U, the search unit 70 receives the context 13 from the user terminal 15, and proceeds to S-202U.

  In S-202U, the search unit 70 searches the condition storage unit 40 for the condition information 41 that matches the context 13 received in S-201U, and proceeds to S-203U. The condition information 41 that matches the context 13 means the condition information 41 that satisfies the conditional expression that the context 13 includes.

  In S-203U, if there is matching condition information 41 as a result of the search in S-202U, the search unit 70 proceeds to S-204U, and if not, ends the processing.

  In S-204U, the search unit 70 creates a recommendation history 51 from the context 13 and the condition information 41, and proceeds to S-205U. Here, the search unit 70 sets the condition ID obtained from the condition information 41 and the context 13 received from the user terminal 15 to the condition ID and context 13 of the recommendation history 51 to be created. The search unit 70 sets “unsuccessful” information in the recommendation success / failure information of the recommendation history 51.

  In S-205U, the search unit 70 stores the recommendation history 51 generated in S-204U in the recommendation history storage unit 50, and proceeds to S-206U.

  In S-206U, the search unit 70 transmits the condition information 41 to the recommendation execution device 17 and ends the process.

  FIG. 19 is a success history process flowchart showing the operation of the system according to the second embodiment.

  In S-201s, the success history registration unit 80 receives the success history 31 from the success history input terminal 16, and proceeds to S-202s.

  In S-202s, the success history registration unit 80 creates a search request for the recommendation history 51 from the success history 31, and proceeds to S-203s. The search request for the recommendation history 51 includes a recommendation condition ID and a user ID included in the success history 31.

  In S-203s, the success history registration unit 80 searches the recommendation history storage unit 50 for the recommendation history 51 including the same recommendation condition ID and user ID as the search request for the recommendation history 51, and proceeds to S-204s.

  In S-204s, the success history registration unit 80 proceeds to S-205s if there is a matching recommendation history 51 as a result of S-203s, and proceeds to -206s if not.

  In S-205s, the success history registration unit 80 updates the recommendation success / failure information of the recommendation history 51 of the recommendation history storage unit 50 including the same recommendation condition ID and user ID as the received success history 31 to “success”, and ends the processing. To do.

  In S-206s, the success history registration unit 8060 stores the received success history 31 in the success history storage unit 30 and ends the process.

  In the recommendation condition correction apparatus 10 of the second embodiment, the apparatus itself can automatically accumulate the recommendation history 51 and the success history 31. As a result, the degree of freedom in modifying the recommendation conditions is improved, and the burden of creating and maintaining the manager's recommendation history 51 and success history 31 is reduced.

  The reason is that the recommendation condition correction apparatus 10 receives the scenario 12 from the administrator terminal 11, the context 13 from the user terminal 15, and the success history 31 from the success history input terminal 16, creates a recommendation history 51, and recommends the recommendation history 51. This is because the success history 31 is accumulated.

<Third Embodiment>
Next, a third embodiment of the present invention will be described in detail with reference to the drawings.

  FIG. 20 is a block diagram of a recommendation condition correction system 90 according to the third embodiment of this invention. A recommendation condition correction system 90 illustrated in FIG. 20 includes a recommendation condition correction apparatus 10, an administrator terminal 11, a user terminal 15, a recommendation condition correction apparatus 10, and a recommendation execution apparatus 17.

  The recommended condition correction apparatus 10 of the present embodiment includes a registration unit 60, a condition storage unit 40, a search unit 70, a recommendation history storage unit 50, and a success history registration unit, which are configured with logic circuits, storage devices, and the like. 80, a success history storage unit 30, and a recommendation condition correction unit 20. The recommended condition correction apparatus 10 may be a computer that operates under program control. In this case, the registration unit 60, the condition storage unit 40, the search unit 70, the recommendation history storage unit 50, the success history registration unit 80, the success history storage unit 30, and the recommendation condition correction unit 20 are provided in the computer. The central processing unit may be realized by reading and executing a program stored in the storage device. The recommendation history storage unit 50, the condition storage unit 40, and the success history storage unit 30 may include a disk device included in the computer.

  The condition storage unit 40, the search unit 70, the recommendation history storage unit 50, the success history registration unit 80, the success history storage unit 30, and the recommendation condition correction unit 20 in the third embodiment of the present invention are the same as those in the second embodiment. It is the same. Hereinafter, the registration unit 60, the search unit 70, and the success history registration unit 80, which are differences from the second embodiment, will be described.

  The registration unit 60 receives the scenario 12 and registers the condition information 41 in the condition storage unit 40. FIG. 21 shows an example of the scenario 12. The scenario 12 includes a scenario ID, recommendation conditions, recommendation information, and expected conditions.

  The scenario ID, recommendation conditions, and recommendation information are the same information as in the first embodiment.

  The expected condition is information including an expected condition ID and a conditional expression. The expected condition ID is information for distinguishing the expected condition. The conditional expression expresses a state expected of the user as a result of the recommendation using a part or all of the context 13. The recommended condition correcting apparatus 10 determines that the user whose context 13 matches the expected condition has taken the expected behavior.

  For example, in the case of scenario 12 shown in FIG. 21, the recommended letter training correct system 90 is indicated by recommendation information 1 for a user whose context 13 matches “position = station & age = 50s & health condition = good”. Make a recommendation. As a result, the user whose context 13 is “position = park” is determined to have taken the expected behavior. That is, the recommendation information 1 in the scenario 12 of FIG. 21 is information for guiding the user to the park.

  The condition information 41 is information including a scenario ID, a condition ID, a conditional expression, and a condition type. In the first embodiment, the condition information 41 exists corresponding to the recommended condition. However, in the third embodiment, the condition information 41 is created corresponding to both the recommended condition and the expected condition. The The condition type indicates whether the condition information 41 corresponds to a recommended condition or an expected condition. In addition, the recommendation condition ID of the scenario 12 is stored as the association information. FIG. 22 shows an example of the condition information 41.

  The registration unit 60 acquires a scenario ID, a recommendation condition ID, a recommendation condition conditional expression, recommendation information, an expected condition ID, and an expected condition conditional expression from the scenario 12. Then, the registration unit 60 sets the condition information 41 as “scenario ID = scenario ID, condition ID = recommended condition ID, conditional expression = conditional condition of recommended condition, condition type = recommended condition, association information = recommended information ID”. Create Furthermore, the registration unit 60 also includes condition information 41 with “scenario ID = scenario ID, condition ID = expected condition ID, conditional expression = conditional condition of expected condition, condition type = expected condition, association information = recommended condition ID”. create.

  The condition storage unit 40 stores one or more condition information 41. FIG. 22 shows an example of a plurality of condition information 41 stored in the condition storage unit 40. FIG. 22 shows an example in which the condition storage unit 40 stores four condition information 41 including recommendation condition 1, recommendation condition 2, expectation condition 1, and expectation condition 2.

  The search unit 70 receives the context 13 from the user terminal 15 and acquires the condition information 41 that matches the context 13 from the condition storage unit 40. When the condition type of the matching condition information 41 is the recommendation condition, the search unit 70 creates the recommendation history 51 and transmits the condition information 41 to the recommendation execution device 17. When the condition type is an expected condition, the search unit 70 transmits the context 13 and the condition information 41 to the success history registration unit 80.

  For example, when a plurality of condition information 41 shown in FIG. 22 is stored in the condition storage unit 40, when the search unit 70 receives the context 13 shown in FIG. 23A, the context 13 is “location = station, age = 50s”. , Condition information 41 matching “sex = male, number of steps = 1000, health condition = good” is acquired.

  In this case, the acquired condition information 41 is that the condition ID is the recommended condition 1, the scenario ID is the scenario 1, the conditional expression is “position = station & age = 50s & health = good”, and the association information is the recommended information 1 Condition information 41. Since the condition type of the condition information 41 is a recommendation condition, the search unit 70 creates the recommendation history 51 and transmits the condition information 41 to the recommendation execution device 17. The recommendation execution device 17 executes recommendation.

  When the search unit 70 receives the context 13 shown in FIG. 23B, the search unit 70 acquires condition information 41 that matches the “position = park, age = 50s, gender = male, step count = 3000, health = good”. . In this case, the acquired condition information 41 is the condition information 41 where the condition ID is the expected condition 1, the scenario ID is the scenario 1, and the conditional expression is “position = park”. Since the condition type of the condition information 41 is an expected condition, the condition information 41 and the context 13 are transmitted to the success history registration unit 80.

  The success history registration unit 80 receives the condition information 41 and the context 13 and generates a success history 31 from the condition information 41 and the context 13. The success history registration unit 80 generates the success history 31 from the association information (recommended condition ID) extracted from the condition information 41 and the context 13.

  When the condition information 41 shown in FIG. 24 and the context 13 shown in FIG. 25 are received, the success history registration unit 80 extracts the recommendation condition 1 of the association information (recommended condition ID) from the condition information 41 and matches it with the context 13. 26 creates the success history 31 shown in FIG.

  Next, the overall operation of the present exemplary embodiment will be described in detail with reference to a flowchart. The operation of the system according to the present embodiment can be divided into a scenario registration flow for registering a scenario 12, a user context processing flow performed every time a user context 13 is received, and a recommended condition correction flow for correcting a recommendation condition. .

  The scenario registration flow and the recommendation condition correction flow are the same as those in the first embodiment. Hereinafter, the user context processing flow will be described.

  FIG. 27 is a context process flowchart showing the operation of the system according to the third embodiment.

  In S-301U, the search unit 70 receives the user context 13, and proceeds to S-302U.

  In S-302U, the search unit 70 searches the condition storage unit 40 for the condition information 41 that matches the context 13 received in S-301U, and proceeds to S-303U.

  In S-303U, as a result of the search in S-302U, the search unit 70 proceeds to S-304U if the matching condition type of the condition information 41 is a recommended condition, and otherwise proceeds to S-307U.

  In S-304U, the search unit 70 creates a recommendation history 51 from the context 13 and the condition information 41, and proceeds to S-305U. Here, the search unit 70 sets the condition ID obtained from the condition information 41 and the context 13 received from the user terminal 15 to the condition ID and context 13 of the recommendation history 51 to be created. The search unit 70 sets “unsuccessful” information in the recommendation success / failure information of the recommendation history 51.

  In S-305U, the search unit 70 stores the recommendation history 51 generated in S-304U in the recommendation history storage unit 50, and proceeds to S-306U.

  In S-306U, the search unit 70 transmits the condition information 41 to the recommendation execution device 17 and ends the process.

  In S-307U, if the condition type of the matching condition information 41 is an expected condition as a result of the search in S-302U, the search unit 70 transmits the context 13 and the condition information 41 to the success history registration unit 80, and the S-308U If there is no matching condition information 41, the process is terminated.

  In S-308U, the success history registration unit 80 generates a success history 31 from the received context 13 and condition information 41, and proceeds to S-309s. The success history registration unit 80 generates the success history 31 from the association information extracted from the condition information 41 and the context 13.

  In S-309U, the success history registration unit 80 creates a recommendation history search request from the success history 31, and proceeds to S-310s. The recommendation history search request includes a recommendation condition ID and a user ID included in the success history 31.

  In S-310U, the success history registration unit 80 searches the recommendation history storage unit 50 for condition information 41 including the same recommendation condition ID and user ID as the recommendation history search request, and proceeds to S-311U.

  In S-311U, the success history registration unit 80 proceeds to S-312U if there is matching condition information 41 as a result of S-310U, and proceeds to S-313U if not.

  In S-312U, the success history registration unit 80 updates the recommendation success / failure information of the recommendation history 51 of the recommendation history storage unit 50 including the same recommendation condition ID and user ID as the success history 31 to “success”, and ends the process.

  In S-313U, the success history registration unit 80 stores the success history 31 in the success history storage unit 30 and ends the process. The recommendation condition correction apparatus 10 according to the third embodiment can automatically accumulate the recommendation history 51 and the success history 31 by itself. As a result, the degree of freedom in modifying the recommendation conditions is improved, and the burden of creating / maintaining the recommendation history 51 and the success history 31 of the manager or the like is reduced. Furthermore, it is not necessary to input the success history 31 from the success history input terminal 16, and the burden of creating and maintaining the recommendation history 51 and the success history 31 for the manager and the like is further reduced.

  The reason is that the recommendation condition correction apparatus 10 receives the scenario 12 from the administrator terminal 11 and the context 13 from the user terminal 15, creates the recommendation history 51, and accumulates the recommendation history 51 and the success history 31. It is.

<Fourth embodiment>
Next, a fourth embodiment of the present invention will be described in detail with reference to the drawings.

  FIG. 28 is a block diagram of the recommended condition correcting apparatus 10 according to the fourth embodiment of this invention. The recommended condition correction apparatus 10 of the present embodiment includes a condition storage unit 40, a recommendation history storage unit 50, a success history storage unit 30, and a recommendation condition correction unit 20.

  The condition storage unit 40 stores a recommendation condition that specifies a range of one or more attribute values of the context 13 that is user information including a plurality of attributes. The recommendation history storage unit 50 responds to each of the target users who are the users to whom the recommendation information for encouraging a predetermined action is output when the context 13 satisfies the recommendation condition. A recommendation history 51 including success / failure information indicating whether or not the target user has taken a predetermined action is stored. The success history storage unit 30 stores the value of each attribute of the context 13 of the successful user in response to each successful user who is a user who has taken a predetermined action other than the target user.

  The recommendation condition modification unit 20 corresponds to each value of each attribute included in the context 13 of all the recommendation histories 51 stored in the recommendation history storage unit 50, and the recommendation history 51 of the recommendation history 51 in which the attribute of the context 13 is the value. Of these, a first contribution ratio indicating the ratio of the recommendation history 51 indicating that a predetermined action has been taken is calculated. Further, the recommendation condition correction unit 20 calculates a second contribution ratio indicating the ratio of the success history 31 in which the value of the attribute of the context 13 matches the recommendation condition in the success history 31 for each attribute in the recommendation condition. Then, the recommendation condition is corrected based on the first contribution rate and the second contribution rate.

  According to the present invention, it is possible to realize the recommended condition correcting apparatus 10 that can effectively correct the recommended conditions. For example, the device can improve the “number of successful recommendations” while suppressing a decrease in the “recommendation success rate”. The reason is that the recommendation condition correction apparatus 10 performs the deletion attribute information determination using the success history 31 accumulated in the success history storage unit 30.

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

  The present invention can be applied to a recommendation condition correction system 90 represented by an advertisement distribution system, a control system, a notification system, an expert system, a navigation system, and the like.

  This application claims the priority on the basis of Japanese application Japanese Patent Application No. 2011-274793 for which it applied on December 15, 2011, and takes in those the indications of all here.

DESCRIPTION OF SYMBOLS 10 Recommendation condition correction apparatus 11 Administrator terminal 15 User terminal 16 Success history input terminal 17 Recommendation execution apparatus 20 Recommendation condition correction part 30 Success history memory | storage part 31 Success history 40 Condition memory | storage part 41 Condition information 50 Recommendation history memory | storage part 51 Recommendation history 60 Registration unit 70 Search unit 80 Success history registration unit 90 Recommended condition correction system

Claims (18)

Condition storage means for storing a recommendation condition that specifies a range of one or more values of the attribute that is user information including a plurality of attributes;
When the context satisfies the recommendation condition, each attribute value of the target user and the target user correspond to each target user who is a user who has output recommendation information for encouraging a predetermined action. A recommendation history storage means for storing a recommendation history including success / failure information indicating whether or not a predetermined action has been taken;
Success history storage means for storing the value of each attribute of the context of the successful user in response to each successful user who is a user who has taken the predetermined action other than the target user;
Corresponding to each value of each attribute included in the context of all the recommendation histories stored in the recommendation history storage unit, the predetermined action in the recommendation history in which the attribute of the context is the value Calculating a first contribution ratio indicating a ratio of the recommendation history indicating that
For each attribute in the recommendation condition, a second contribution ratio indicating a ratio of the success history in which the attribute value of the context matches the recommendation condition in the success history is calculated, and the first contribution A recommendation condition correction device comprising recommendation condition correction means for correcting the recommendation condition based on the rate and the second contribution rate.   The recommendation condition correcting means adds the value of the attribute having the first contribution rate equal to or higher than a first addition threshold to the recommendation condition, and sets the value of the attribute whose second contribution rate is equal to or lower than a first deletion threshold. 2. The recommended condition correcting apparatus according to claim 1, wherein the recommended condition correcting apparatus is deleted from the recommended condition. The recommendation condition correction means calculates a success rate that is a ratio of a recommendation history indicating that the predetermined action is taken out of the recommendation history, and when the success rate is equal to or less than a second deletion threshold, The recommendation condition correction apparatus according to claim 2, wherein a value of the attribute having a second contribution rate equal to or less than the first deletion threshold is deleted from the recommendation condition.   The recommendation condition correction means adds the value of the attribute having the first contribution rate equal to or higher than the first additional threshold value and the second contribution rate equal to or higher than a second additional threshold value to the recommendation condition. Item 2 or 3 recommendation condition correction device. The context includes a user identifier as one of the attributes,
When the context of the user who has received at least a part from the user terminal satisfies the recommendation condition, the context determined to satisfy the recommendation condition and the user does not take the predetermined action Search means for storing success / failure information as the recommendation history in the recommendation history storage means;
Obtain the context of a successful user from a success history input terminal, and search in the recommendation history storage unit whether the recommendation history including the user identifier that matches the user identifier included in the acquired context exists. And
If it exists, the success / failure information of the retrieved recommendation history is updated to indicate that the target user has taken the predetermined action, and if not, the context acquired from the success history input terminal The recommendation condition correcting device according to claim 1, further comprising a success history registration unit that stores the success history in the success history storage unit. In addition to the recommendation condition, the condition storage means includes an expectation condition including a condition to be satisfied by one or more attribute values selected from the context in order to determine that the user has taken the predetermined action Store and
The context includes a user identifier as one of the attributes,
When the user's context that has acquired at least a part from the user terminal satisfies the recommendation condition, the acquired context and the success / failure information indicating that the user has not taken the predetermined action are recommended. Search means for storing in the recommendation history storage means as a history;
Whether or not the recommendation history including the user identifier that matches the user identifier included in the acquired context exists when the context of the user acquired at least a part from the user terminal satisfies the expected condition If it exists, the success / failure information of the searched recommendation history is updated so as to indicate that the target user has taken the predetermined action. The recommendation condition correction device according to claim 1, further comprising a success history registration unit that stores the success history in the success history storage unit. Storing a recommendation condition that specifies a range of one or more values of the attribute that is user information including a plurality of attributes in the condition storage unit;
When the context satisfies the recommendation condition, each attribute value of the target user and the target user correspond to each target user who is a user who has output recommendation information for encouraging a predetermined action. Storing a recommendation history including success / failure information indicating whether or not a predetermined action has been taken in the recommendation history storage means;
In response to each successful user who is the user who has taken the predetermined action other than the target user, the value of each attribute of the context of the successful user is stored in a success history storage unit, and stored in the recommendation history storage unit The value indicating that the attribute of the context is the value corresponding to each value of the attribute included in the context of all of the recommendation histories being performed indicates that the predetermined action has been taken in the recommendation history Calculate the first contribution ratio indicating the ratio of recommendation history,
For each attribute in the recommendation condition, a second contribution ratio indicating a ratio of the success history in which the attribute value of the context matches the recommendation condition in the success history is calculated, and the first contribution A recommendation condition correction method for correcting the recommendation condition based on the rate and the second contribution rate.   A value of the attribute having the first contribution rate equal to or higher than a first addition threshold is added to the recommendation condition, and a value of the attribute having the second contribution rate equal to or lower than a first deletion threshold is deleted from the recommendation condition; The recommendation condition correction method according to claim 7.   A success rate that is a ratio of a recommendation history indicating that the predetermined action is taken out of the recommendation history is calculated, and when the success rate is equal to or less than a second deletion threshold, the second contribution rate is the first contribution rate. The recommendation condition correcting method according to claim 8, wherein a value of the attribute equal to or less than one deletion threshold is deleted from the recommendation condition.   The recommendation condition according to claim 7 or 8, wherein a value of the attribute having the first contribution rate equal to or higher than the first additional threshold value and the second contribution rate equal to or higher than a second additional threshold value is added to the recommendation condition. How to fix. The context includes a user identifier as one of the attributes,
When the context of the user who has received at least a part from the user terminal satisfies the recommendation condition, the context determined to satisfy the recommendation condition and the user does not take the predetermined action Storing success / failure information as the recommendation history in the recommendation history storage means;
Obtain the context of a successful user from a success history input terminal, and search in the recommendation history storage unit whether the recommendation history including the user identifier that matches the user identifier included in the acquired context exists. And
If it exists, the success / failure information of the retrieved recommendation history is updated to indicate that the target user has taken the predetermined action, and if not, the context acquired from the success history input terminal The recommendation condition correcting method according to claim 7, wherein: is stored in the success history storage unit. In the condition storage means, in addition to the recommendation condition, an expectation condition including a condition to be satisfied by one or more values of the attribute selected from the context in order to determine that the user has taken the predetermined action Store and
The context includes a user identifier as one of the attributes,
When the user's context that has acquired at least a part from the user terminal satisfies the recommendation condition, the acquired context and the success / failure information indicating that the user has not taken the predetermined action are recommended. Stored in the recommendation history storage means as a history,
Whether or not the recommendation history including the user identifier that matches the user identifier included in the acquired context exists when the context of the user acquired at least a part from the user terminal satisfies the expected condition If it exists, the success / failure information of the searched recommendation history is updated so as to indicate that the target user has taken the predetermined action. The recommendation condition correction method according to claim 7, wherein the recommendation condition correction method is stored in the success history storage unit. A condition storage process that stores a recommendation condition that specifies a range of one or more values of the attribute that is user information, including a plurality of attributes, as condition storage means;
When the context satisfies the recommendation condition, each attribute value of the target user and the target user correspond to each target user who is a user who has output recommendation information for encouraging a predetermined action. A recommendation history storage process for storing a recommendation history including success / failure information indicating whether or not a predetermined action has been taken in the recommendation history storage means;
A process of storing the value of each attribute of the context of the successful user in a success history storage unit in response to each successful user who is a user who has taken the predetermined action other than the target user;
Corresponding to each value of each attribute included in the context of all the recommendation histories stored in the recommendation history storage unit, the predetermined action in the recommendation history in which the attribute of the context is the value Calculating a first contribution ratio indicating a ratio of the recommendation history indicating that
For each attribute in the recommendation condition, a second contribution ratio indicating a ratio of the success history in which the attribute value of the context matches the recommendation condition in the success history is calculated, and the first contribution A recommendation condition correction program for causing a computer to execute a recommendation condition correction process for correcting the recommendation condition based on the rate and the second contribution rate.   The attribute value having the first contribution rate equal to or higher than a first addition threshold is added to the recommendation condition, and the attribute value having the second contribution rate equal to or less than a first deletion threshold is deleted from the recommendation condition. 14. The recommended condition correction program according to claim 13, which causes the computer to execute a recommended condition correction process.   A success rate that is a ratio of a recommendation history indicating that the predetermined action is taken out of the recommendation history is calculated, and when the success rate is equal to or less than a second deletion threshold, the second contribution rate is the first contribution rate. The recommendation condition correction program according to claim 14, wherein the computer executes the recommendation condition correction processing for deleting the attribute value equal to or less than one deletion threshold from the recommendation condition.   The recommended condition correction processing for adding the value of the attribute having the first contribution rate equal to or higher than the first additional threshold and the second contribution rate equal to or higher than the second additional threshold to the recommended condition. The recommended condition correction program according to claim 14 or 15, wherein the program is executed. The context includes a user identifier as one of the attributes,
When the context of the user who has received at least a part from the user terminal satisfies the recommendation condition, the context determined to satisfy the recommendation condition and the user does not take the predetermined action Search processing for storing success / failure information as the recommendation history in the recommendation history storage means;
Obtain the context of a successful user from a success history input terminal, and search in the recommendation history storage unit whether the recommendation history including the user identifier that matches the user identifier included in the acquired context exists. And
If it exists, the success / failure information of the retrieved recommendation history is updated to indicate that the target user has taken the predetermined action, and if not, the context acquired from the success history input terminal The recommendation condition correction program according to any one of claims 13 to 16, further causing the computer to execute a success history registration process for storing the success history in the success history storage unit. In addition to the recommended condition, an expected condition including a condition to be satisfied by one or more values of the attribute selected from the context in order to determine that the user has taken the predetermined action is stored in the condition storage unit The condition storage processing to
The context includes a user identifier as one of the attributes,
When the user's context that has acquired at least a part from the user terminal satisfies the recommendation condition, the acquired context and the success / failure information indicating that the user has not taken the predetermined action are recommended. A search process to be stored in the recommendation history storage means as a history;
Whether or not the recommendation history including the user identifier that matches the user identifier included in the acquired context exists when the context of the user acquired at least a part from the user terminal satisfies the expected condition If it exists, the success / failure information of the searched recommendation history is updated so as to indicate that the target user has taken the predetermined action. The recommended condition correction program according to any one of claims 13 to 16, further causing the computer to execute a success history registration process stored in a success history storage unit.

Images