JP2001022826A - Client relation learning system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a highly effective service corresponding to the purchase history or attribute of each client. SOLUTION: A purchase cycle extraction system 106 extracts a purchase article list and a cycle information on a selected client. A purchase history time sequential accumulation system 107 generates the time sequential summary information of purchase activity for every client. A purchase candidate list generation system 108 outputs the list of purchase candidate articles purchased by a similar client, the information of the selected client and the similar client. On the basis of the information generated by the above individual systems and service history information, a proposal service determination system 113 automatically determines the service to be provided to every client.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] The present invention relates to purchase history information,
The present invention relates to a system for automatically determining the type, timing, and proposal method of a service proposed to a customer from customer attribute information and service history information.

[0002]

2. Description of the Related Art In recent years, database marketing,
As typified by the phrase one-to-one marketing, technology has been developed that analyzes the purchase history of customers and uses it for marketing.

AprioriS which extracts a customer's time-series purchase pattern from data including a customer ID, a processing time, and a purchase item
There is a technique called ome or AprioriAll. Proc. Of
theInt'l Conference on Data Engineering (ICDE), Ta
"Mining Sequential P" in ipei, Taiwan, March 1995
Atterns "has a commentary by R. Agrawal and R. Srikant (prior art 1).

Also, "ONEtoONE Corporate Strategy" (Don Peppers, Martha Rogers, Toshiaki Iseki, One to Two)
Translated by One Marketing Council, Mari Kuramochi, Fujitsu iM
In the i-net translation), the following cases related to learning relations and cases using community knowledge are introduced.

[0005] According to a shopping list that is updated weekly, a merchandise delivery company obtains information on which customers are consuming which products and how much, and provides products and services according to individual customers' requests. , And store the result (Prior Art 2).

[0006] Firefly Network, Inc.
The site has been launched, where users are presented with CD titles and other information. Visitors first rank what they think of different types of songs and musicians. With this ranking, people with similar preferences are searched for from among the entire users, and the CDs they prefer are referenced and presented (Prior Art 3).

[0007]

The prior art 1 is a technique for extracting a purchase pattern of the entire target customer. It does not state how to use this result.

The prior art 2 merely provides products and services according to each customer's request and purchase history, and provides new products and services by referring to the history and attribute information of other customers. There is no mechanism to do this.

The prior art 3 selects similar customers from information indicating customer preferences gathered by a questionnaire,
The purchased product of the similar customer is presented. However, there is no mechanism for learning the products to be presented and the timing of presentation from the customer's own purchase history.

In addition, none of the prior arts 1, 2, and 3 can use the history of services provided in the past to determine a proposed service.

In addition, none of the prior arts 1, 2, and 3 can use the expected effect of each service in determining a proposed service.

In view of the problems of the prior arts 1, 2, and 3,
A first object of the present invention is to automatically propose a service proposed to a customer using purchase pattern information obtained from purchase history information.

Further, when deciding a service to be proposed to each customer, not only the purchase history and service history of the same customer but also the purchase history and service history of other similar customers are utilized.
A second object is to propose services that have not been used by customers.

It is a third object of the present invention to learn the contents of a service provided to each customer, the method of providing the service, and the time of providing the service from the purchase history of the customer.

Further, in deciding a proposed service,
A fourth object is to use a history of services provided in the past and not provide services that are unlikely to be used.

It is a fifth object of the present invention to manage purchase activities of each customer in chronological order, evaluate expected effects of each service, and determine a proposed service.

[0017]

In order to achieve the first object, in the customer relationship learning system of the present invention, a proposed service is automatically determined by a proposed service determination system based on a customer's purchase pattern. .

In order to achieve the second object, in the customer relationship learning system of the present invention, the purchase candidate list generation system and the proposed service determination system refer to the purchase history and service history of the customer having similar attributes, Automatically determine the services to be provided to each customer.

In order to achieve the third object, the customer relationship learning system according to the present invention includes a purchase cycle extracting system,
The proposed service determination system automatically determines the content and time of the proposed service according to the customer's own purchase pattern.

In order to achieve the fourth object, in the customer relationship learning system of the present invention, the service to be proposed is determined by referring to the effect of the service by the purchase history time series totaling system and the proposed service determination system. Can decide.

[0021]

Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 is a configuration diagram of a customer relationship learning system to which the present invention is applied. This system automatically determines the type, timing, and method of service proposed to each customer.

In FIG. 1, reference numeral 106 denotes customer selection information 101 for specifying a customer such as a customer ID; a purchased product determination parameter 102 provided from outside such as a survey period;
A purchase history information 103 including a set of information specifying a purchased product such as D and time information such as purchase date and time is input, and a selected customer purchase including at least information specifying a product purchased by the customer selected by the customer selection information 101 A purchase cycle extraction system, 107, for selecting, from the product list and the purchased product list, products whose periodicity is recognized for purchase, and outputting purchase cycle information to which the periodicity information is added; 1
03, a purchase history time-series totaling system 108 that outputs purchase activity time-series summary information in which purchase activities of each customer are totaled in a time-series manner. A customer selection information 101, a purchase item determination parameter 102, a purchase history Information 103 and customer attribute information 104 indicating the characteristics of the customer are input, and the product I
D, a purchase candidate list generation system that outputs a purchase candidate list 112 including at least information specifying a product likely to be purchased by the customer specified by the customer selection information 101. , Selected customer purchase item list 110, purchase activity time-series summary information 111,
This is a proposal service determination system that inputs a purchase candidate list 112 and outputs proposal service information 114 including at least a set of information specifying a customer such as a customer ID and information specifying a service such as a service ID.

Next, each system of FIG. 1 will be described in detail. First, the purchase cycle extraction system 106 will be described. FIG. 2 shows a system configuration diagram of the purchase cycle extraction system 106.

In FIG. 2, the purchase cycle extraction system 10
Reference numeral 6 denotes a purchase in which the customer selection information 101, the purchase item determination parameter 102, and the purchase history information 103 are input, and the selected customer purchase history information 202, which is purchase history information on the customer selected by the customer selection information 101, is output. The history extraction device 201, the selected customer purchase history information 202, and the purchased product determination parameter 102 are input, and a selected customer purchased product list 110, which is a list of products purchased by the customer selected by the customer selection information 101, is output. Purchase product list extraction device and selected customer selection customer purchase product list 110
And a purchase cycle extraction device 204 that receives the purchase product determination parameter 102 as input and outputs purchase cycle information 109.

FIG. 3 shows an example of the customer selection information 101.
In this example, only the customer ID of each customer is listed. However, the format is not limited to this format as long as information that can identify the customer is included.

FIG. 4 shows an example of the purchase commodity decision parameter 102. The target period indicates a period during which the purchase history information 103 is investigated. In this example, since January 1, 1999, 19
This indicates that only the purchase history up to February 28, 1999 is targeted. The purchase item selection criterion is a criterion for selecting a purchase item. In this example, it is shown that a product whose purchase count is four or more is selected.

FIG. 5 shows an example of the purchase history information 103.
This is time-series data indicating a purchase history. At a minimum, the date and time, information for specifying a purchased customer such as a customer ID, and information for specifying a product such as a product ID are included. FIG. 5 shows data in a spreadsheet format, but the format is not limited to this. In a preferred embodiment of the present invention, the processing ID
And information indicating whether or not the price of each product and the proposed service have been used.

FIG. 6 shows an example of the selected customer purchase history information 202. FIG. 6 shows spread format data similar to FIG. 5, but the format is not limited to this.

The purchase history extracting device 201 extracts only the customer history information indicated in the purchase history information 103 from the purchase history information 103 and stores it in the selected customer purchase history information 202.

FIG. 7 shows an example of the selected customer purchase item list 110. The format is not limited to this as long as information for specifying a product such as a product ID is included at a minimum. When a plurality of customers are selected in the customer selection information 101, information for specifying the customer such as a customer ID is also required.

The purchase item list extracting device 203 extracts only items that satisfy the condition specified by the parameter 102 from the selected customer purchase history information 202, and outputs information for specifying the item to the selected customer purchase item list 110. To be stored.

FIG. 8 shows an example of the purchase cycle information 109. The average value and standard deviation of the purchase cycle are described for each product. At least, the format is not limited as long as it includes information for specifying a product such as a product ID and information about a purchase cycle such as an average purchase cycle.

The purchase cycle extraction device 204 receives the selected customer purchase item list 110 as input, calculates information representing the purchase cycle such as an average purchase cycle for each product, and stores the information in the purchase cycle information 109.

FIG. 9 is a flowchart of a process for calculating the purchase cycle. Each process in FIG. 9 will be described.

Reference numeral 901 denotes a product selection process for sequentially selecting products from the selected customer purchase product list 110.

Reference numeral 902 denotes a selection status confirmation process for confirming whether or not the process for all the commodities has been executed.

Reference numeral 903 indicates that from the purchase history information 103,
This is a history extraction process for extracting a history regarding a currently selected product.

Reference numeral 904 denotes an average purchase interval calculation process for calculating an average value of purchase intervals in the history information extracted in 903.

Reference numeral 905 denotes a purchase interval standard deviation calculation process for calculating the standard deviation of the purchase interval in the history information extracted in 903.

An information storage process 906 stores the values calculated in 904 and 905 in the purchase cycle information 109.

Next, the purchase history time series totaling system 107
Will be described. FIG. 10 shows a configuration diagram of the purchase history time series totaling system. In FIG. 10, the purchase history time series totaling system 107 includes a purchased product determination parameter 102.
And the purchase history information 103, and outputs all customer purchase history information 1001, which is purchase history information of all customers. All customer purchase history extraction device 1001, all customer purchase history information 1002, and purchase item determination. The system includes a purchase history totaling device 1003 that receives the parameter 102 as input and outputs purchase activity time-series summary information 111.

FIG. 11 shows an example of the purchase commodity decision parameter 102. The purchase history time-series totaling system 107 refers to a totaling target period and a totaling item from the purchase commodity determination parameters 102. Here, the tallying target period is a period to be extracted when information for tallying is extracted from the purchase history information 103. The last day of the tallying period is the tallying date. That is, in the example of FIG.
February 8, 1999 is the target date. In FIG.
Although the description is in the range of year, month and day, it is possible to add time if necessary. The aggregation item is an item name to be aggregated in the purchase history aggregation device. The calculation method of each item is defined in the purchase history totaling device. Here, each tally item will be described.

The total purchase price per month is the total value of the purchase amount of each customer in the past month or 30 days from the target date.

The total purchase price per week is the total value of the purchase amount of each customer in the past week, that is, seven days, from the target date.

The number of store visits per month is the total value of the number of visits each customer has made in the past month or 30 days from the target date.

The number of visits per week is the total value of the number of visits made by a customer to the past one week, that is, seven days, from the target date.

[0048] The monthly average customer unit price is defined as:
This is the average value of purchases made during one visit.

The weekly average customer unit price is the average value of the purchase amount of each customer at a single visit to the store during the past week, that is, seven days, from the target date.

The number of purchased items per month is the total value of the number of items purchased by each customer in the past month or 30 days from the tallying date.

The number of purchased items per week is the total value of the items purchased by each customer in the past week, that is, seven days, from the tallying date.

The purchase price on the day is the total value of the purchases on the target date.

The number of items purchased on the day is the total value of the number of items purchased on the counting target day.

The format of all customer purchase history information 1002 is the same as the example of purchase history information 103 shown in FIG. However, the all customer purchase history information 1002 includes only history information included in the tallying target period specified by the purchase product determination parameter 102.

The all-customer purchase history extraction device 1001 selects the purchase commodity determination parameter 10 from the purchase history information 103.
Only the information included in the tally period specified in 2 is extracted and stored in all customer purchase history information 1002.

FIG. 12 shows an example of the purchase activity time-series summary information. FIG. 12 shows a spreadsheet format, but the format is not limited to this. The meaning of each item described in the first line of FIG. 12 is as described above in the parameter information example of FIG.

The purchase history totaling device 1003 extracts purchase history information of each customer from all customer purchase history information, and calculates the value of the total item specified by the purchase commodity determination parameter 102 for each customer. FIG. 13 shows a processing flow.
Each process in FIG. 13 will be described.

Reference numeral 1301 denotes purchase activity time-series summary information 1
This is a customer selection process for sequentially selecting calculation target customers for which 11 is to be generated.

Reference numeral 1302 denotes customer selection status determination processing for determining whether or not processing has been performed for all customers. If all the customers have been processed, the process is terminated. Otherwise, the next history extraction process 1303 is executed.

Reference numeral 1303 denotes all customer purchase history information 1002
Is a history extraction process for extracting calculation target customer history information, which is history information on the calculation target customer selected by the customer selection process 1301 from among the above.

Reference numeral 1304 denotes a total purchase price per month calculation process for summing up all price values only for records in the past month from the calculation target customer history information to the count target date.

Reference numeral 1305 denotes a weekly total purchase price calculation process for summing up all price values only for records in the past week from the calculation target customer history information.

Reference numeral 1306 denotes a number-of-stores-per-month calculation process for summing up the numbers of different process IDs only for records in the past month from the calculation target customer history information to the counting target date.

Reference numeral 1307 denotes a number of visits per week that calculates the number of visits per week by summing up the numbers of different processing IDs only for records in the past week from the target date of calculation from the customer history information to be calculated. This is a number calculation process.

In step 1308, the prices are summed up for each processing ID only for the records for the past month from the calculation target customer history information, and the average value of the total value is calculated. This is a monthly average customer unit price calculation process.

1309 is a week for calculating the average value of the total value for each processing ID only for records of the past one week from the calculation target customer history information, going back to the calculation target date. This is an average customer unit price calculation process.

Reference numeral 1310 denotes a monthly average purchased item number calculation process for summing up the number of records only for records in the past month from the calculation target customer history information to the counting target date.

Reference numeral 1311 denotes a weekly average purchased item number calculation process for summing up the number of records only for records in the past week from the calculation target customer history information to the counting target date.

Reference numeral 1312 denotes a day-of-purchase amount calculation process for summing up the price values only for the record of the calculation target date from the calculation target customer history information.

Reference numeral 1313 denotes a day-of-purchase item number calculation process for summing up the number of records only for the record on the calculation target date from the calculation target customer history information.

Reference numeral 1314 denotes information storage processing for storing the values obtained in the processing 1304 to 1313 in the purchase activity time-series summary information 111 together with the customer ID and the calculation target date.

Next, the purchase candidate list generation system 108
Will be described. FIG. 14 shows a configuration diagram of the purchase candidate list generation system. In FIG. 14, a purchase candidate list generation system 108 receives customer selection information 101, a purchased product determination parameter 102, and customer attribute information 104 as input, and specifies a customer similar to the customer specified by the customer selection information 101. Similar customer information 1 including ID etc.
A similar customer selection device 1401 that outputs 402, purchased product determination parameters 102, similar customer information 1402,
With the purchase history information 103 as input, similar customer information 1402
Is the purchase history information of one or more similar customers identified in,
The system includes a purchase history extraction device 202 that outputs similar customer purchase history information 1403, and a similar customer purchase item list extraction device 1404 that receives the similar customer purchase history information 1403 and outputs a purchase candidate item list 112.

FIG. 15 shows a purchase candidate list generation system 1
08, the purchased goods determination parameter 102 used
Here is an example of the element. In FIG. 15, a similar customer selection method specifies a method for selecting similar customers. In this example,
A commonly known inference method called MBR (Memory Based Reasoning) is used. For MBR, see "C
raig Stanfill, David Waltz, TOWAR MEMORY-BASED REA
SONING, Communications of the ACM, Dec 1986, Vol.2
9, Number 29, pp. 1213-1228 ". The MBR is a method of searching for similar data and performing prediction from the value of the similar data. In the present invention, a similar data search portion in the MBR is used. The target period and the purchase selection criterion are as described above.

FIG. 16 shows an example of the customer attribute information 104. The customer attribute information 104 includes age, gender,
Stores attribute values such as occupation. In FIG. 16, it is represented by a spreadsheet format table, but the format is not limited to this.

If the MBR is used, for example, similar customer information 1402 as shown in FIG. 17 can be obtained. As shown in the example of FIG. 15, the three most similar customers are selected. In FIG. 18, customer selection information 101 is obtained from customer attribute information 104.
The customer whose customer ID is designated as 000345 and the customer ID indicated in FIG.
An example is shown in which attribute information of a similar customer is extracted. The three customers are the same age or one year older, have the same gender and occupation, and have the same or different annual income ranks (the annual income is A
To E).

FIG. 19 shows an example of similar customer purchase history information. FIG. 19 shows purchase history information on the customer specified by the similar customer information 1402 shown in FIG.

Similar customer purchase item list extraction device 1404
Performs a function similar to that of the purchase list extraction device in the purchase cycle extraction system 106. The difference from the purchase list extraction device is that the product I is extracted after the purchase list is extracted.
The point is that the number of similar customers is counted for each ID specifying a product, such as D, and the ratio to the number of all similar customers is calculated. FIG. 20 shows an example of a purchase candidate product list.

Next, the proposed service determination system will be described. FIG. 21 shows a configuration diagram of the proposed service determination system. As shown in FIG. 21, the proposed service determination system 113 receives the selected customer purchase item list 110 and the purchase candidate item list 112 as input, and generates a service candidate list 2102 that is a candidate for the proposed service information 114 and outputs a candidate service list 2102. A service history verification device 2103 that receives a device 2101, a candidate service list 2102, and service history information 105 and outputs a verified candidate service list 2104 in which the effect of a service is verified based on a history of services provided in the past. A verified candidate service list 2104, service history information 105,
A service evaluation device 2105 that receives the purchase activity time-series summary information 111 as input and outputs service characteristic information 2106 that stores changes due to the provision of services, such as changes in the purchase amount per month and the number of visits, and service characteristic information 2106.
And a service selection device 2107 that outputs proposed service information 114 storing the type and time of the proposed service.

Next, each device of FIG. 21 will be described in detail.

First, the service candidate list generation device 210
1 will be described. FIG. 22 shows a configuration diagram of the service candidate list generation device. As shown in FIG. 22, the service candidate list generation device 2101 includes purchase cycle information 109,
A selected customer candidate service selection that inputs a selected customer purchase product list 110, service determination parameter information 115, and a product service correspondence table 2102 defining services to be provided to customers who have purchased each product, and outputs a candidate service list 2102 Apparatus 2101 and purchase candidate list 11
2, service determination parameter information 115, and a product service correspondence table 2102, and a candidate service list 2
Similar customer candidate service selection device 210 that outputs 102
3 FIG. 23 shows a product service correspondence table 21.
02 is shown. In FIG. 23, it is represented by a spreadsheet format table, but the format is not limited to this. Two or more services may correspond to one product, or two or more products may correspond to one service.

Selected customer candidate service selection device 2201
Is the purchase cycle information 109 and the selected customer purchase item list 110
Is selected from the product service correspondence table 2202, and the candidate service list 2
Stored in 102. Whether only the product ID in the purchase cycle information 109 or all the product IDs in the selected customer purchase product list is used is described in the service determination parameter information.

The similar customer service selection device 2203 selects a service corresponding to the product ID described in the purchase candidate product list 112 from the product service correspondence table 2202, and stores the service in the candidate service list 2102. That is, the candidate service list 2102 includes the selected customer candidate service selection device 2201 and the similar customer candidate service selection device 22
03 are merged. FIG. 23 shows an example of the candidate service list 2102.

The service history verification device 2103 refers to the service history information 105 for the service candidates listed in the candidate service list 2102, verifies the effects of the services performed in the past, and selects the services that have no effect. Remove from service list and output as verified candidate service list. FIG. 25 shows an example of the service history information.

Next, the service history verification device 2103 will be described. FIG. 26 shows the service history verification device 210
3 is a configuration diagram. As shown in FIG. 26, the service history verification device 2103 receives the target customer list 116 and the service history information 105 as input, and performs target customer service history information 260 as service history information of the selected customer and similar customers.
2, target customer service history information 2602, and candidate service list 2
102, and a service selection device 2603 that receives the service determination parameter information 115 as an input and outputs a verified candidate service list 2104.

The service selection device 2603 counts, for each service indicated in the candidate service list, the number of times proposed to the selected customer, the number of times used by the selected customer, the number of times proposed to similar customers, and the number of times used by similar customers. . Then, the usage rate of the selected customer and the usage rate of similar customers are calculated. Then, only the service that satisfies the constraint indicated in the service determination parameter information is output as a verified service list.

FIG. 27 shows an example of a verified candidate service list. FIG. 28 shows an example of the service determination parameter information.

Next, the service evaluation device 2105 will be described. The service evaluation device 2105 obtains, for each service, a difference between purchase activities before and after using the service from the purchase activity time-series summary information 111. FIG.
FIG. 9 shows a flowchart of a process executed by the service evaluation device 2105. Each process in FIG. 29 will be described.

Reference numeral 2901 denotes a service selection process for sequentially selecting one service from the verified candidate service list 2104.

Reference numeral 2902 denotes service selection status confirmation processing for confirming whether or not all services in the verified candidate service list 2104 have been processed.

Reference numeral 2903 denotes service history information for sequentially selecting one from the service history information of the service currently being processed from the service history information 105. However,
The range in which the service history information is searched is determined by the service determination parameter information 115 to be within 30 days from the date of using the service.

[0091] Reference numeral 2904 confirms whether or not the service history information 105 has searched all the history information within a predetermined period for the service being processed selected from the verified candidate service list 2104. This is a service history selection status confirmation process.

[0092] Reference numeral 2905 denotes a service characteristic calculation process for summarizing the purchase activity summary information before and after use of the service for each service. For example, the difference between the average value of the total purchase amount per week for 30 days before use of the service and the average value of the total purchase amount per week for 30 days after use of the service is obtained. This value is calculated for all customers who have used the service, and the results are added together, thereby making it possible to calculate whether the total purchase amount per week has increased or decreased with this service. The same processing can be performed for the total purchase amount per month and the number of visits per week.

FIG. 30 shows an example of the service characteristic information 2106.

Next, the service selection device 2107 will be described. The service selection device selects a service to be proposed from the service characteristic information 2106 in accordance with the conditions defined by the service determination parameter information. The condition determined by the service determination parameter is, for example, “change in total purchase price per week> 100” as shown in FIG. In addition, the service selection device 2107 determines whether a service related to a product included in the purchase cycle information is
In accordance with the service determination parameter 115, a delivery date is added and the result is output as the proposed service information 114. FIG.
The example of the proposal service information 114 is shown in FIG.

FIG. 32 shows an example of a parameter information setting screen. Although this example is an example of a setting screen of the service determination parameter information 115, the setting of the purchase item determination purchase item determination parameter 102 can be similarly performed. These settings can be input on a display of the computer by operating a keyboard and a mouse attached to the computer.

FIG. 33 shows a display screen example of the proposed service information 114. The proposed service information 114 can be stored on a hard disk or the like, but can also be displayed on a computer display as shown in FIG.
The user can use the mouse attached to the computer
By pressing the parameter reset button, it is possible to return to the parameter information setting screen shown in FIG. 32 again.

The above embodiment can be modified and implemented as follows.

First, purchase candidate list generation system 10
In step 8, similar customers are extracted using not only customer attribute information but also purchase history information. With this change, not only the customer attributes but also the customers with similar purchase histories can be selected as similar customers.

Second, in the customer selection information, customer attribute information, purchase history information, and service history information, I indicating family members
Add D. With this change, the purchase activity can be managed on a family basis, and an appropriate service can be proposed on a family basis.

Thirdly, there are various service providing methods, such as cash register delivery and DM, and services having different providing methods give different service IDs. With this change, it is possible to evaluate the difference in the service used depending on the providing method.

[0101]

According to the present invention, there is an effect that an appropriate service can be proposed at an appropriate time according to the purchase history of each customer. In addition, there is an effect that a service that is not yet used by each customer but is likely to be used can be proposed by referring to the purchase history of another customer having similar attribute information and purchase history. Further, there is an effect that the probability of use is calculated based on the history of previously proposed services, and the proposed services can be selected. Further, by managing the purchase history of each customer on a daily basis and correlating with the service history information, the effect of each service can be evaluated, and the evaluation value can be referred to to select a proposed service.

[Brief description of the drawings]

FIG. 1 is a configuration diagram of a customer relationship learning system.

FIG. 2 is a configuration diagram of a purchase cycle extraction system.

FIG. 3 is a diagram showing an example of customer selection information.

FIG. 4 is a diagram illustrating an example of parameter information.

FIG. 5 is a diagram showing an example of purchase history information.

FIG. 6 is a diagram showing an example of selected customer purchase history information.

FIG. 7 is a diagram showing a selected customer purchase item list.

FIG. 8 is a diagram showing purchase note information.

FIG. 9 is a purchase cycle extraction flowchart.

FIG. 10 is a configuration diagram of a purchase history time-series totaling system.

FIG. 11 is a diagram illustrating a parameter information example 2;

FIG. 12 is a diagram showing purchase activity time-series summary information.

FIG. 13 is a purchase history totaling processing flow.

FIG. 14 is a configuration diagram of a purchase candidate list generation system.

FIG. 15 is a diagram illustrating a third example of parameter information;

FIG. 16 is a diagram illustrating an example of customer attribute information.

FIG. 17 is a diagram illustrating an example of similar customer information.

FIG. 18 is a diagram illustrating an example of attribute information of a selected customer and similar customers.

FIG. 19 is a diagram showing an example of similar customer purchase history information.

FIG. 20 is a diagram showing an example of a purchase candidate list.

FIG. 21 is a configuration diagram of a proposed service determination system.

FIG. 22 is a configuration diagram of a service candidate list generation device.

FIG. 23 is a diagram illustrating an example of a product service correspondence table.

FIG. 24 is a diagram showing an example of a candidate service list.

FIG. 25 is a diagram illustrating an example of service history information.

FIG. 26 is a configuration diagram of a service history verification device.

FIG. 27 is a diagram illustrating an example of a verified candidate service list.

FIG. 28 is a diagram illustrating an example of service determination parameter information.

FIG. 29 is a service evaluation processing flowchart.

FIG. 30 is a diagram showing an example of service characteristic information.

FIG. 31 is a diagram illustrating an example of proposed service information.

FIG. 32 is a diagram showing an example of a parameter information setting screen.

FIG. 33 is a diagram showing an example of a proposed service display screen.

 ──────────────────────────────────────────────────続 き Continuing from the front page (72) Inventor Yukiyasu Ito 5030 Totsuka-cho, Totsuka-ku, Yokohama-shi, Kanagawa Prefecture F-term in the Software Division of Hitachi, Ltd. 5B049 AA06 BB11 CC00 DD05 EE01 EE05 FF09

Claims (1)

[Claims] 1. Customer selection information specifying a customer such as a customer ID, purchase history information including a set of information specifying a purchased product such as a product ID and time information such as purchase date and time, and customer attributes indicating characteristics of the customer Customer relationship learning that automatically determines the type, timing, and method of service to be proposed to each customer based on the information, information identifying the customer such as a customer ID, and service history information that stores the usage status of the proposed service. system.