JP2017073114A - Recommendation item output method, recommendation item output program, and recommendation item output device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To present an appropriate information of recommendation item.SOLUTION: A server 10 has a user evaluation acquisition section 12 that acquires a piece of information on a search item (characteristics item) selected by a user for searching an item on a search I/F which is used for searching items, and stores the same in an evaluation value table 22. A user classification section 14 identifies users similar to the user based on the evaluation value table 22. An information presentation section 16 outputs a piece of information on an item to which the similar users have referred previously as a piece of information of an item to be recommended to the user.SELECTED DRAWING: Figure 3

Description

  The present invention relates to a recommended item output method, a recommended item output program, and a recommended item output device.

  Conventionally, a collaborative filtering technique is known as a technique for identifying an item that is highly likely to be of interest to a user searching for an item on the Web or a database and presenting it as a recommended item. As a collaborative filtering technique, for example, a similar user is identified based on user information or an item accessed by the user, and an item accessed by the similar user is presented as a recommended item. As a collaborative filtering method, there is a method of specifying an item similar to an item accessed by the user based on the similarity of the item itself and presenting it as a recommended item.

  Patent Documents 1 to 3 and the like disclose a technique for recommending information related to content viewed by a user.

JP 2006-48320 A JP 2011-135183 A JP 2012-22365 A

  However, in the above collaborative filtering method, since the recommended item is not presented in consideration of what kind of feature the user is interested in, there is a possibility that an appropriate item cannot be presented as the recommended item. .

  In one aspect, an object of the present invention is to provide a recommended item output method, a recommended item output program, and a recommended item output device capable of outputting information of appropriate recommended items.

  In one aspect, the recommended item output method obtains information on a search item selected when a user searches for an item on an item search screen, and information on search items selected by a plurality of users in the past. The information of the search item acquired in the acquisition process is stored in the storage unit stored in the item, the user similar to the user is identified with reference to the storage unit, and the item referred to in the past by the similar user This is a recommended item output method in which the computer executes the process of outputting the information as item information recommended for the user.

  Appropriate recommended item information can be output.

It is a figure showing roughly the composition of the information service system concerning one embodiment. It is a figure which shows the hardware constitutions of a server. It is a functional block diagram of a server. It is a figure which shows an example of search I / F. FIG. 5A is a diagram showing the data structure of the characteristic item table, FIG. 5B is a diagram showing the data structure of the evaluation value table, and FIG. 5C is the data of the work table. It is a figure which shows a structure. 6A is a diagram showing the data structure of the group table, FIG. 6B is a diagram showing the data structure of the access history table, and FIG. 6C is the data structure of the timer variable table. FIG. It is a flowchart which shows a user evaluation acquisition process. It is a flowchart which shows a timer management process. It is a flowchart which shows a recommended item presentation process. It is a flowchart which shows a user classification | category process. It is a figure which shows an example of a self-organization map. It is FIG. (1) for demonstrating a modification. It is FIG. (2) for demonstrating a modification. It is FIG. (3) for demonstrating a modification.

  Hereinafter, an embodiment of an information providing system will be described in detail with reference to FIGS.

  FIG. 1 schematically shows a configuration of an information providing system 100 according to an embodiment. The information providing system 100 according to the present embodiment provides information on items (for example, items such as hard disks, SSDs (Solid State Drives), and diodes) to the user based on search conditions input by the user. Further, the information providing system identifies an item to be recommended to the user (recommended item), and provides the user with information on the identified recommended item.

  As shown in FIG. 1, the information providing system 100 includes a server 10 as a recommended item output device and a user terminal 70. The server 10 and the user terminal 70 are connected to a network 80 such as the Internet.

  The server 10 searches for an item based on the search condition input at the user terminal 70 and outputs the search result to the user terminal 70. Further, the server 10 specifies an item recommended to the user based on the search condition input at the user terminal 70, and outputs the specified result to the user terminal 70. FIG. 2 shows the hardware configuration of the server 10. As shown in FIG. 2, the server 10 includes a CPU (Central Processing Unit) 90, a ROM (Read Only Memory) 92, a RAM (Random Access Memory) 94, a storage unit (HDD (Hard Disk Drive)) 96, a network. An interface 97 and a portable storage medium drive 99 are provided. Each component of the server 10 is connected to a bus 98. In the server 10, a program (including a recommended item output program) stored in the ROM 92 or the HDD 96 or a program (including a recommended item output program) read from the portable storage medium 91 by the portable storage medium drive 99 is stored in the CPU 90. 3, the user evaluation acquisition unit 12 as the acquisition unit and the storage processing unit, the user classification unit 14 as the specification unit and the deletion unit, the information presentation unit 16 as the output unit, and the timer management unit illustrated in FIG. 18 is realized. FIG. 3 also shows various tables stored in the HDD 96 or the like. The various tables include a characteristic item table 21, an evaluation value table 22 as a storage unit, a work table 23, a group table 24, an access history table 25, and a timer variable table 26.

  The user evaluation acquisition unit 12 monitors a user operation in a search I / F (Interface) (see FIG. 4) as a search screen displayed on the user terminal 70, and updates the evaluation value table 22 based on the operation. To do. At the time of this update, the user evaluation acquisition unit 12 refers to the characteristic item table 21. The search I / F in FIG. 4 inputs item information (original article number and classification code) and searches for an alternative (completely compatible part) of the item, but no fully compatible part is searched. The search I / F is assumed to be used in some cases. However, the present invention is not limited to this, and an alternative part of an item is searched. However, when a detailed search is performed by further updating the condition, a search I / F displayed by designation from the user may be used. Specifically, the search I / F in FIG. 4 inputs search conditions (handling classification, package shape, number of terminals, etc.) of items (for example, diodes) corresponding to the original article number and classification code, and sets the search conditions. It is a screen for searching for a corresponding item.

  Here, the characteristic item table 21 has fields of “index” and “characteristic item” as shown in FIG. Identification information assigned to each characteristic item is stored in the “index” field, and the name of the characteristic item of the item selected as the search item in the search I / F is stored in the “characteristic item” field. The The characteristic item is an item included in a column in which “capacity value” or “rated voltage 1” indicated by an arrow in the search I / F of FIG. 4 is displayed. Type (characteristic species). A specific value (such as 16V) of “capacity value” or “rated voltage 1” means “search value”.

  The evaluation value table 22 is a table that records, for each item to be searched, how many times the user has selected each characteristic item in the search I / F, and has a data structure as shown in FIG. Specifically, the evaluation value table 22 includes fields of “user”, “synchronization flag”, and “characteristic item”. User identification information is stored in the “user” field. The “synchronization flag” stores a flag indicating whether or not synchronization processing is currently being performed. In the present embodiment, “FALSE” is stored as the synchronization flag when the synchronization process is performed, and “TRUE” is stored as the synchronization flag when the synchronization process is not performed. In the “characteristic item” field, a value indicating how many times each characteristic item has been selected by each user is stored. Note that the user evaluation acquisition unit 12 increments the value (evaluation value) stored in the field of the characteristic item every time the user selects it as a search item, but if the search item is not selected for a predetermined time, the user evaluation acquisition unit 12 sets the value. Reset (return to 0). Note that the “evaluation value” of the characteristic item is also referred to as “number of selections” below. The evaluation value table 22 is prepared for each item to be searched.

  Returning to FIG. 3, the user classification unit 14 creates a work table 23 based on the evaluation value table 22, divides similar users into groups based on the work table 23, and stores information on each group in the group table 24. Store. Here, as shown in FIG. 5C, the work table 23 is a table in which information of users whose synchronization flag is “TRUE” in the evaluation value table 22 is gathered. The group table 24 is a table that stores information on groups in which similar users are grouped, and has a data structure as shown in FIG. Specifically, as shown in FIG. 6A, the group table 24 includes fields of “group” and “belonging user”. In the “group” field, group identification information is stored, and in the “affiliation user” field, identification information of users belonging to each group is stored.

  The information presentation unit 16 refers to the group table 24 and the access history table 25 to determine an item to be recommended to the user and outputs it to the user terminal 70. Here, the access history table 25 is a table for managing the history of items accessed by the user, and has a data structure as shown in FIG. Specifically, the access history table 25 has fields of “user” and “item” as shown in FIG. User identification information is stored in the “user” field, and item identification information accessed by each user is stored in the “item” field.

  When the timer variable set for each characteristic item becomes a predetermined value based on the timer variable table 26, the timer management unit 18 notifies the user classification unit 14 of the fact. Here, the timer variable table 26 is a table for managing timer variables set in each characteristic item of each user, and has a data structure as shown in FIG. Specifically, the timer variable table 26 has fields of “user”, “characteristic item”, and “timer variable”. User identification information is stored in the “user” field, and identification information of the characteristic item selected by the user is stored in the “characteristic item” field. The “timer variable” field stores a timer variable used to determine whether or not it is time to reset the evaluation value table 22.

  Returning to FIG. 1, the user terminal 70 is a terminal such as a PC (Personal Computer), displays the search I / F of FIG. 4 on the display unit, and stores the search condition input by the user via the input unit as a server. 10 to send. In addition, the user terminal 70 receives the search result based on the search condition in the server 10 and displays it on the display unit. Further, the user terminal 70 receives information on recommended items from the server 10 and displays it on the display unit.

(Information provision system processing)
Next, the processing of the information providing system 100 will be described in detail with reference to the other drawings as appropriate along the flowcharts of FIGS.

(User evaluation acquisition process)
First, the user evaluation acquisition process by the server 10 will be described with reference to the flowchart of FIG. The user evaluation acquisition process of FIG. 7 is a process of totalizing which characteristic item the user has selected in the search I / F.

  In the process of FIG. 7, first, in step S <b> 10, the user evaluation acquisition unit 12 waits until the search I / F of FIG. 4 is displayed on any user terminal 70. The search I / F shown in FIG. 4 is displayed as described above, although the item information (original article number and classification code) is input to search for an alternative to the item. Is not searched, or when a substitute for the item is found, a detailed search with further modified conditions is desired. The information presentation unit 16 transmits the search I / F screen in FIG. 4 to the user terminal 70.

  After the search I / F is displayed on the user terminal 70, the process proceeds to step S12, and the user evaluation acquisition unit 12 sets the synchronization flag of the user who uses the user terminal 70 on which the search I / F is displayed to “ Set to FALSE. In this case, the user evaluation acquisition unit 12 sets the user's synchronization flag to “FALSE” in the evaluation value table 22 shown in FIG.

  Next, in step S14, the user evaluation acquisition unit 12 determines whether or not the characteristic item has been changed. Note that the column indicated by the arrow of the search I / F in FIG. 4 is initially blank. Therefore, when the user selects one characteristic item as a search item in the field, or when the selected characteristic item is changed, the determination in step S14 is affirmed.

  If the determination in step S14 is negative, the process proceeds to step S20. If the determination in step S14 is positive, the process proceeds to step S16.

  In step S <b> 16, the user evaluation acquisition unit 12 increments the evaluation value (number of selections) of the changed characteristic item in the evaluation value table 22 by one. For example, when the user U1 selects the characteristic item of index = B0, the user evaluation acquisition unit 12 changes the value of the characteristic item B0 of the user U1 from 2 to 3 in the evaluation value table 22 of FIG. change.

  Next, in step S18, the user evaluation acquisition unit 12 sets the value of the timer variable in the timer variable table 26 to zero. If there is no changed characteristic item among the characteristic items associated with the searched user in the timer variable table 26, the changed characteristic item is added and the value of the timer variable is set. Is set to 0. Thereafter, the user evaluation acquisition unit 12 proceeds to step S20.

  If transfering it to step S20, the user evaluation acquisition part 12 will judge whether the search was performed. That is, the user evaluation acquisition unit 12 determines whether or not the “search” button has been pressed by the user in the search I / F of FIG. If the determination in step S20 is negative, the process returns to step S14. If the determination is positive, the process proceeds to step S22.

  If transfering it to step S22, the user evaluation acquisition part 12 will return a user's synchronization flag to "TRUE" in the evaluation value table 22. FIG. After step S22, the process returns to step S10, and the user evaluation acquisition unit 12 repeatedly executes the above-described processing.

(Timer management processing)
Next, timer management processing by the timer management unit 18 will be described with reference to the flowchart of FIG. The process in FIG. 8 is a process for managing timer variables in the timer variable table 26. Note that the process of FIG. 8 is a process executed in parallel with the process of FIG.

  In the process of FIG. 8, first, in step S30, the timer management unit 18 waits until a predetermined time elapses. In the present embodiment, the predetermined time is 1 second. When the determination in step S30 is affirmed and the process proceeds to step S32, the timer management unit 18 increases all timer variables by a predetermined time. When the predetermined time is 1 second, the timer management unit 18 increments the values of all timer variables existing in the timer variable table 26 (FIG. 6C) by one.

  Next, in step S34, the timer management unit 18 determines whether all users have been designated. Here, since no one has been designated yet, the process proceeds to the next step S36, and the timer management unit 18 designates one undesignated user.

  Next, in step S38, the timer management unit 18 determines whether there is an unspecified timer variable value among the specified user timer variables. If the determination in step S38 is negative, the process returns to step S34. If the determination is positive, the process proceeds to step S40.

  In step S40, the timer management unit 18 designates one of unspecified timer variables. Next, in step S42, the timer management unit 18 determines whether or not the designated timer variable is larger than the threshold value. If the determination is negative, the process returns to step S38. If the determination is positive, the process proceeds to step S44.

Here, it is assumed that the threshold is determined based on the unit price (for example, average) of the item that the user is searching for. For example, when the predetermined fixed time interval is GLOBAL_UNIT_FLUSHTIME and the unit price of the item is ITEM_COST, the threshold value CONST_TIME can be expressed by the following equation (1) as an example.
CONST_TIME = GLOBAL_UNIT_FLUSHTIME × ITEM_COST (1)

  As shown in the above equation (1), the threshold value is decreased as the unit price of the item is lower, and the threshold value is increased as the unit price of the item is higher. it can. For example, an item with a low unit price has a high frequency of design replacement and component convergence, so it is necessary to make a determination in a short time about the number of selections of the characteristic item of interest. For this reason, a short reset time is set for an item with a low unit price. In addition, since an item with a high unit price has no change such as replacement for a long period of time, the number of selections of a characteristic item to be focused on may be determined in a long time. For this reason, a long reset time is set for an item with a high unit price.

  When the determination in step S42 is affirmed and the process proceeds to step S44, the timer management unit 18 sets the evaluation value (number of selections) of the specified characteristic item of the specified user in the evaluation value table 22 to 0. 14 is instructed. In response to the instruction, the user classifying unit 14 sets (resets) the evaluation value of the designated characteristic item to 0. In addition, the timer management unit 18 deletes the corresponding characteristic item of the designated user from the timer variable table 26.

  Thereafter, the process returns to step S38, and if there is an unspecified timer variable value (S38: affirmative), it is specified (S40), and processing based on the result of comparison with the threshold value is executed (S42, S44). After designating all timer variables of the designated user (S38: No), the process returns to step S34, designates the next undesignated user (S36), and executes the same processing (S38 to S44) as described above. To do.

  In this way, after all users have been specified and all timer variables have been specified, the determination in step S34 is affirmed, and all the processes in FIG. 8 are terminated. Note that the process of FIG. 8 is repeatedly executed. That is, the processing after step S32 is repeatedly executed every predetermined time (1 second in the present embodiment).

(Recommended item presentation process)
Next, the recommended item presentation processing by the user classification unit 14 and the information presentation unit 16 will be described along the flowcharts of FIGS. 9 and 10. 9 and 10 is a process of presenting searched item information and recommended item information to the user. This process is also performed in parallel with the above-described user evaluation acquisition process (FIG. 7) and timer management process (FIG. 8).

  In the process of FIG. 9, first, in step S50, the user classification unit 14 waits until a search is executed by the user. That is, when the “search” button of the search I / F in FIG. 4 is pressed by the user, the user classifying unit 14 proceeds to step S52.

  In step S52, the user classification unit 14 executes a user classification processing subroutine. In step S52, the process according to the flowchart of FIG. 10 is executed.

(User classification process (S52))
In the process of FIG. 10, first, in step S <b> 70, the user classification unit 14 determines whether all users have been designated. If the determination in step S70 is negative, the process proceeds to step S72, and the user classifying unit 14 designates one unspecified user.

  Next, in step S <b> 74, the user classifying unit 14 acquires the synchronization flag of the designated user from the evaluation value table 22. Next, in step S76, the user classification unit 14 determines whether or not the acquired synchronization flag is “TRUE”. If the determination in step S76 is negative, the process returns to step S70. If the determination is positive, the process proceeds to step S78.

  In step S78, the user classification unit 14 copies the user information designated in the work table 23. Specifically, the user classifying unit 14 copies the specified user information excluding the synchronization flag in the evaluation value table 22 to the work table 23. Thereafter, the process returns to step S70.

  When returning to step S70, the above-described processing (S70 to S78) is repeatedly executed, and when all the users have been designated, the determination in step S70 is affirmed, and the user classifying unit 14 proceeds to step S80. The work table 23 is obtained by extracting data with the synchronization flag “TRUE” from the evaluation value table 22 as described above. That is, it can be said that the work table 23 is a part of the evaluation value table 22.

  In step S80, the user classifying unit 14 acquires a score (characteristic item evaluation value (number of selections)) from the work table 23. Next, in step S82, the user classification unit 14 executes user classification calculation. In this case, as an example, the user classifying unit 14 classifies the users based on the number of selections of the characteristic items of each user by using a self-organizing map (SOM) algorithm, and identifies similar users. . Note that the user classifying unit 14 may classify users by SVM (Support Vector Machine).

  Hereinafter, the SOM algorithm executed by the user classifying unit 14 will be described in detail. The SOM algorithm is a type of neural network, and is a model configured by a dual structure of an input layer and a competitive layer, and expressing the similarity of given input information by the distance between neurons on the competitive layer. The neurons (nodes) in the competitive layer are arranged in a two-dimensional lattice.

(A) First, the user classifying unit 14 initializes reference vectors m _i (the number of selections of characteristic items in this embodiment) of all the competitive layer nodes i (users in this embodiment) to random values.

(B) Next, the user classifying unit 14 sets data randomly selected from the input data (the number of selections of characteristic items by the randomly selected user) as the input vector x.

(C) Then, the user classification unit 14, the following equation (2), determine the Euclidean distance between the reference vectors m _i of the node i of competitive layer and the input vector x, find the winning node c of the distance is minimum . In other words, the node having the number of selections of the characteristic item most similar to the number of selections of the characteristic item by the randomly selected user is defined as the winner node c. In the following formula (2), i is i = 1, 2, 3,.
| X−m _c | = min | x−m _i | (2)

Here, for example, the distance obtained from the number of selections of the characteristic items A0 to E0 of the user U0 and the number of selections of the user U1 in FIG. 5B can be expressed as the following equation (3).
{(3-0) ² + (0-2) ² + (7-5) ² + (0-3) ² + (4-5) ² } ^1/2
≒ 5.2 (3)

Further, for example, the distance obtained from the number of selections of the characteristic items A0 to E0 of the user U0 and the number of selections of the user U2 in FIG. 5B can be expressed as the following equation (4).
{(3-5) ² + (0-0) ² + (7-3) ² + (0-0) ² + (4-2) ² } ^1/2
≒ 3.5 ... (4)

When obtaining the Euclidean distance, the weighted Euclidean distance can be obtained by weighting each characteristic item. For example, when weighting based on the selection frequency of characteristic items is applied (when the selection frequency is high, the weight is increased), as shown in the following equation (5), selection of the characteristic item j by all users as the weight w _j A value obtained by dividing the number of times N _j by the number of times of selection N _all of the characteristic items can be used.
W _j = N _j / N _all (5)

By using such a weight w _j , the influence on the distance of the characteristic item selected by many users becomes large, and the influence on the distance of the characteristic item hardly selected by the user becomes small.

On the other hand, for example, when the weight is increased as the selection frequency is lower, a value represented by the following equation (6) can be used as the weight W _j .
W _j = {1− (N _j / N _all )} (6)
The characteristic items selected by many users are naturally reflected in the distance, but the characteristic items that are hardly selected are not so reflected in the distance calculation. However, in practice, the difference in the characteristic items by the user should be most meaningful in the distance calculation. Therefore, by using the weight w _j as in the above equation (6), it is possible to increase the influence of the difference in the number of times each user has selected a characteristic item having a low selection frequency on the Euclidean distance.

For example, the weight of the characteristic item A0 in FIG. 5B is 0.3, the weight of B0 is 0.2, the weight of C0 is 0.1, the weight of D0 is 0.3, and the weight of E0 is 0.1. Suppose there was. In this case, the distance obtained from the number of selections of the characteristic items A0 to E0 by the user U0 in FIG. 5B and the number of selections of the characteristic items A0 to E0 by the user U1 can be expressed as the following equation (7). it can.
{0.3 × (3-0) ² + 0.2 × (0-2) ² + 0.1 × (7-5) ² + 0.3 × (0-3) ² + 0.1 × (4-5) ² } ^1/2 ≒ 2.6
... (7)

Further, for example, the distance obtained from the number of selections of the characteristic items A0 to E0 by the user U0 in FIG. 5B and the number of selections of the characteristic items A0 to E0 by the user U2 is expressed as the following equation (8). Can do.
{0.3 × (3-5) ² + 0.2 × (0-0) ² + 0.1 × (7-3) ² + 0.3 × (0-0) ² + 0.1 × (4-2) ² } ^1/2 ≒ 1.4
(8)

  As can be seen by comparing the magnitudes of the values of the above formulas (5) and (6) with the magnitudes of the values of the above formulas (7) and (8), the distance between when the weight is used and when it is not used. There is no change in the magnitude relationship.

(D) Then, the user classification unit 14, in the competitive layer is updated based nodes of the reference vectors m _i of winning node c and the vicinity thereof (around) the following equation (9). In the case of updating the reference vectors m _i in the vicinity of the following equation (9), as the distance is farther, the more the number of updates increases, changes in the reference vector m _i by updating is reduced.
m _i (t + 1) = m _i + h _ci (t) m _i [x (t) −m _i (t)] (9)
Note that h _ci is called a neighborhood function and is defined by a Gaussian function such as Equation (10).

  By repeating the above (a) to (d), self-organization is realized.

  FIG. 11 shows an example of the result of user classification using SOM. In FIG. 11, each point represents a user. A set of users is displayed by shifting each user from the center point for convenience of expressing the set. In the case of FIG. 11, since the users are classified into 8 × 8 grids, the users included in the grid are defined as the same group for each grid, and the users in the same group are defined as similar users. Also, the closer the grid distance is, the closer the similarity between user groups is. In FIG. 11, users with the same mark indicate users who belong to the same department. In the self-organizing map, the initial weight of each user is often randomly assigned in the initial stage, but may be determined by reflecting characteristic items with high update frequency for each user.

  Next, in step S84, the user classification unit 14 generates a group table 24. Specifically, the user classification unit 14 assigns identification information to each group including similar users, and stores the identification information of users belonging to the group in the group table 24 for each group identification information. . When the above processing is performed, the entire processing in FIG. 10 is terminated, and the process proceeds to step S54 in FIG.

  If transfering it to step S54, the information presentation part 16 will perform the process which displays a search result and a recommended item. Specifically, the information presentation unit 16 searches an item DB (not shown) that stores detailed information of an item, and uses information on an item that satisfies a search condition input on the search I / F as a search result. It transmits to the user terminal 70 of the searched user. In addition, the information presentation unit 16 extracts a user similar to the searching user from the group table 24 and acquires information on an item accessed by the similar user from the access history table 25. And the information presentation part 16 transmits the information of the item which is not contained in a search result among the acquired item information to the user terminal 70 which the searched user uses as information of a recommended item.

  In addition, the information presentation unit 16 recommends item information that matches the search item (characteristic item) selected by the user on the search I / F and the input search value among items of information accessed by similar users. May be output as the information.

  Next, in step S56, the information presentation unit 16 determines whether or not a detailed display instruction has been input. If the determination in step S56 is negative, the process proceeds to step S60. On the other hand, when either the search result or the recommended item is selected by the user and an instruction to display the detailed information of the item is issued, the determination in step S56 is affirmed, and the process proceeds to step S58.

  If transfering it to step S58, the information presentation part 16 will acquire the detailed information of the selected item from item DB not shown, and will transmit to the user terminal 70. FIG. The user terminal 70 displays the received information on the display unit. Further, the information presentation unit 16 records an item on which detailed information is displayed in the access history table 25. After the process of step S58, the process proceeds to step S60.

  In step S60, the user classifying unit 14 determines whether the search is finished. If the determination in step S60 is negative, the process returns to step S56. If the determination is positive, all the processes in FIG. 9 are terminated.

(Concrete example)
Here, a specific example will be described.
As an example, a case where a replacement candidate (substitution candidate) of a Zener diode is searched as an item will be described. When searching for a replacement candidate for a Zener diode, the user selects a characteristic item (number of pins, package, failure rate, life, Zener voltage, etc.) as a search item on the search I / F, and specifies the specific characteristic item. Enter a value (search value).

  In the server 10, the user classifying unit 14 acquires a characteristic item (for example, lifetime) selected by the user in the search I / F, and selects a user group using a self-organizing map or the like based on the selected characteristic item. Generate (S80-S84). Then, the server 10 extracts Zener diodes and the like accessed in the past by a user similar to the searched user, and transmits it to the user terminal 70 used by the searched user (S54). In this case, for example, when the user is searching with the search condition “lifetime is 3 years or more”, among the Zener diodes or the like accessed by similar users in the past, the search condition is that the lifespan is 3 years or more. Only information on the corresponding Zener diode or the like may be transmitted to the user terminal 70.

  Since the Zener diode is frequently replaced, the threshold value of the timer variable used in the timer management process (FIG. 8) is set to a smaller value (see step S42).

  As described above, in this example, information such as Zener diodes that have been accessed in the past by a user who has a similar search tendency (the tendency of the characteristic item to be selected) is provided to a user who is searching for a Zener diode. This makes it possible to provide appropriate information as recommended item information. Actually, as a result of operating the information providing system 100 by employees of a plurality of departments, a self-organizing map as shown in FIG. 11 was obtained. From this result, the same characteristic item as the characteristic item selected by the user is obtained. It can be seen that the employees who have selected are not necessarily employees of the same department. For this reason, by using information on items accessed by similar users based on the self-organizing map as recommended items, it is possible to provide users with useful information that matches the user's interests regardless of the user's affiliation, etc. The possibility is considered high.

  As described above in detail, according to the present embodiment, when the user searches for an item in the search I / F (FIG. 4) used when searching for an item, the user evaluation acquisition unit 12 of the server 10 searches for the item. Information on the selected search item (characteristic item) is acquired and stored in the evaluation value table 22 (S16). And the user classification | category part 14 specifies the user similar to a user based on the evaluation value table 22 (S80-S84). Furthermore, the information presentation unit 16 outputs information on items that have been referred to in the past by similar users as information on items recommended to the user (S54). As a result, it is possible to present information of items that have been referred to in the past by users with similar search trends (the tendency of characteristic items to be selected) to the user as recommended items. It is possible to present information on items that have been adapted.

  Here, when specifying a similar user based on the user information and the item selection result of the user, (1) the first-time user cannot use the user because the user information or the item selection result is unknown, (2) the user accesses If the items are different, similar users cannot be extracted, and (3) the case where the user is interested in the feature amount (characteristic item) of the item, not the item itself, cannot be considered. In addition, when a recommended item is presented to a user based on the similarity between items, (1) a history of items accessed by the user in the past is required, and (2) which item features (characteristics) the user has There are circumstances such as not being able to consider whether you are interested in (item). On the other hand, in the present embodiment, it is possible to present information on appropriate recommended items even for a first-time user or the like in consideration of the user's interest.

  Further, in the present embodiment, among items that have been referred to by a similar user in the past, the item information identified from the characteristic item selected by the user through the search I / F and the search value is used as recommended item information. Can be output. Thereby, the information of the item similar to the item which the user is searching for can be presented as the recommended item information.

  In the present embodiment, the evaluation value (number of selections) of each characteristic item stored in the evaluation value table 22 is deleted based on a threshold corresponding to the price of the searched item. As a result, it is possible to reset the evaluation value at an appropriate timing based on the item replacement, the appearance frequency of the substitute product, and the like.

  In this embodiment, the synchronization flag of the user who is updating the evaluation value in the evaluation value table 22 is set to “FALSE”, so that the evaluation value being updated is not used when generating the user group. can do.

  In the above embodiment, in the access history table 25, the date and time when each user last accessed each item is recorded, and based on the date and time, it is determined whether or not to present it as recommended item information to the user. It is good to do. For example, only information on items whose period from the last accessed date is within a predetermined period is presented as recommended item information. Here, the predetermined period can be determined based on the item type. For example, the item with the highest unit price and the item with the lowest unit price among the items stored in the access history table 25 are acquired, and the average value is obtained. A long-term predetermined period is set for items whose unit price is higher than the average value, while a short-term predetermined period is set for items whose unit price is lower than the average value. In this way, for inexpensive parts such as Zener diodes (that is, items with a short product switching cycle), it is possible to provide the user with information on items accessed by similar users within a short past period. . On the other hand, for expensive parts (that is, items with a long product switching cycle), it is possible to provide the user with information on items accessed by similar users within a long past period. Thereby, it is possible to provide the user with information on appropriate recommended items in accordance with the product switching cycle. The predetermined period is not limited to the case where the predetermined period is set based on the unit price, and the predetermined period may be set based on other information (for example, a lifetime or a failure rate).

  In the above embodiment, in addition to the recommended items, what characteristic items tend to be selected by similar users may be presented to the user.

  In the above-described embodiment, the case where the timer management unit 18 counts up the timer variable in the timer variable table 26 (S32) has been described. However, the present invention is not limited to this. For example, when the characteristic item is changed, the timer variable in the timer variable table 26 is set to the same value as the threshold value determined for each item in step S18, and the timer variable is counted down in step S32. Also good. In this case, in step S42, it may be determined whether or not the value of the timer variable has become zero.

  In the above embodiment, the case where the user group is generated every time the user performs a search using the search I / F has been described. However, the present invention is not limited to this. For example, the process of generating a user group (S52) may be executed every predetermined time separately from the process of FIG.

  In the above-described embodiment, the case where the item to be searched is a component such as a Zener diode has been described. However, the present invention is not limited to this. For example, a commodity on a general EC site may be used as an item to be searched. In this case, the server 10 generates a user group based on the search item (characteristic item) selected when the user searches for the product, and is recommended to the user who searched for the product accessed by the user belonging to the user group. It may be presented as a product. If the product is a hard disk, the characteristic item is the number of revolutions, if the product is a curtain, the characteristic item is a pattern or the like, and if the product is a cushion, the characteristic item is a material or the like.

  For example, it is assumed that the purchase history of clothes of each user on the EC site is as shown in FIG. In this case, if the emphasis filtering method is used, focusing on the fact that Mr. A purchases new clothes, Mr. A has a high purchase frequency of the T-shirt (that he is interested in the T-shirt). The T-shirt purchased by another user who purchases the shirt (for example, Mr. B of the same generation) is recommended. On the other hand, when the method of the above embodiment is used, when Mr. A purchases clothes, a person similar to Mr. A is identified based on the fact that the characteristic item “color” is often selected. To do. Then, when Mr. A selects “blue” as the characteristic item “color”, he recommends Mr. A blue clothing (for example, a blue shirt) purchased by a person similar to Mr. A (for example, Mr. C). . As a result, information on clothes purchased in the past by users with similar search tendencies can be presented to the user as recommended information, so that information suitable for the user's request can be presented. Of course, it is also possible to present Mr. A with the blue shirt purchased by Mr. C and the T-shirt purchased by Mr. B based on the emphasis filtering method. In addition, it is possible to present information as described above also in products other than clothes, such as shoes and hats.

  Also, for example, it is assumed that the purchase history of each user's rice cooker at the EC site is as shown in FIG. In this case, assuming that the emphasis filtering method is used, for Mr. A who purchases a new rice cooker, he has purchased a large rice cooker in the past. Will recommend A rice cooker. On the other hand, when the method of the above embodiment is used, when Mr. A purchases the rice cooker, the person who is similar to Mr. A based on the fact that the characteristic item “size” is often selected ( For example, Mr. B and Mr. C) are specified. When Mr. A selects “Large” as the characteristic item “Size”, Mr. A recommends to Mr. A a rice cooker of size “Large” purchased by a person similar to Mr. A. In addition, when Mr. A selects “small” as the characteristic item “size”, a rice cooker of size “small” purchased by a person similar to Mr. A is recommended to Mr. A. As a result, information on products purchased in the past by users with similar search tendencies can be presented to the user as recommended information, so that information suitable for the user's request can be presented. In addition, it is possible to present information as described above also in home appliances and other products other than the rice cooker.

  Further, for example, a travel product on a travel plan search site may be an item to be searched. For example, it is assumed that the reservation history of each user in the reservation site of the accommodation destination is as shown in FIG. The item “purpose” in FIG. 14 stores the purpose of stay “business trip” and “leisure” selected by the user. In this case, if the emphasis filtering method is used, Mr. A who is trying to make a new reservation for a hotel (there is a high frequency of making reservations for a hotel near Tokyo Station), a hotel near Tokyo Station that another person has reserved. I will recommend it. On the other hand, when the method of the above embodiment is used, when Mr. A reserves a hotel, he often selects the characteristic item “purpose”, and thus a person similar to Mr. A (for example, , Mr. B and Mr. C). When Mr. A selects “leisure” as the characteristic item “purpose”, the hotel recommends the hotel of the purpose “leisure” reserved by a person similar to Mr. A. Thereby, it is possible to present information suitable for the user's request.

  Furthermore, for example, information on rental properties on a rental property search site may be used as an item to be searched. In this case, the search item (characteristic item) includes location, rent, floor plan, distance from the station, and the like.

  The above processing functions can be realized by a computer. In that case, a program describing the processing contents of the functions that the processing apparatus should have is provided. By executing the program on a computer, the above processing functions are realized on the computer. The program describing the processing contents can be recorded on a computer-readable recording medium (except for a carrier wave).

  When the program is distributed, for example, it is sold in the form of a portable recording medium such as a DVD (Digital Versatile Disc) or a CD-ROM (Compact Disc Read Only Memory) on which the program is recorded. It is also possible to store the program in a storage device of a server computer and transfer the program from the server computer to another computer via a network.

  The computer that executes the program stores, for example, the program recorded on the portable recording medium or the program transferred from the server computer in its own storage device. Then, the computer reads the program from its own storage device and executes processing according to the program. The computer can also read the program directly from the portable recording medium and execute processing according to the program. Further, each time the program is transferred from the server computer, the computer can sequentially execute processing according to the received program.

  The above-described embodiment is an example of a preferred embodiment of the present invention. However, the present invention is not limited to this, and various modifications can be made without departing from the scope of the present invention.

In addition, the following additional remarks are disclosed regarding description of the above embodiment.
(Supplementary note 1) In the item search screen, information on the search item selected when the user searches for the item is acquired.
Storing information on the search items acquired in the acquisition process in a storage unit that stores information on search items selected by a plurality of users in the past;
Referring to the storage unit, identify a user similar to the user,
Outputting information on items referred to in the past by the similar user as information on items recommended for the user,
A recommended item output method characterized in that a computer executes processing.
(Additional remark 2) In the process to output, the item specified from the search item which the said user selected on the search screen among the information of the item which the said similar user referred in the past, and the value of the said input search item The recommended item output method according to appendix 1, wherein the information is output as recommended item information.
(Supplementary note 3) The supplementary note 1 or 2, wherein the computer further executes a process of deleting the information of the search item stored in the storage unit based on a time corresponding to the price of the searched item. 2. The recommended item output method according to 2.
(Additional remark 4) In the process to output, the information of the item referred to by the similar user within the period determined based on the item searched by the user is output as the item information recommended for the user. The recommended item output method according to any one of appendices 1 to 3, wherein:
(Supplementary Note 5) In the item search screen, information on the search item selected when the user searches for the item is acquired.
Storing information on the search items acquired in the acquisition process in a storage unit that stores information on search items selected by a plurality of users in the past;
Referring to the storage unit, identify a user similar to the user,
Outputting information on items referred to in the past by the similar user as information on items recommended for the user,
A recommended item output program which causes a computer to execute processing.
(Additional remark 6) In the process to output, the item specified from the search item which the said user selected on the search screen among the information of the item which the said similar user referred in the past, and the value of the said input search item The recommended item output program according to appendix 5, wherein the information is output as recommended item information.
(Supplementary note 7) The supplementary note 5 or 5, wherein the computer further executes a process of deleting the information of the search item stored in the storage unit based on a time corresponding to the price of the searched item. 6. The recommended item output program according to 6.
(Additional remark 8) In the process to output, the information of the item referred to by the similar user within the period determined based on the item searched by the user is output as the information of the item recommended for the user. The recommended item output program according to any one of appendices 5 to 7, characterized by:
(Supplementary Note 9) In the item search screen, an acquisition unit that acquires information on a search item selected when the user searches for an item;
A storage processing unit that stores information on the search items acquired in the acquisition process in a storage unit that stores information on search items selected by a plurality of users in the past;
A specifying unit for specifying a user similar to the user with reference to the storage unit;
An output unit that outputs information on items referred to by the similar user in the past as item information recommended for the user;
A recommended item output device comprising:
(Additional remark 10) The said output part of the item specified from the search item which the said user selected on the search screen among the information of the item which the said similar user referred in the past, and the value of the said input search item The recommended item output device according to appendix 9, wherein the information is output as recommended item information.
(Additional remark 11) The recommended item output device of Additional remark 9 or 10 further provided with the deletion part which deletes the information of the said search item memorize | stored in the said memory | storage part based on the time according to the price of the searched item.
(Additional remark 12) The said output part outputs the information of the item referred to by the said similar user within the period defined based on the item which the said user searched as the information of the item recommended with respect to the said user. The recommended item output device according to any one of appendices 9 to 11, which is characterized.

10 servers (recommended item output device)
12 User evaluation acquisition unit (acquisition unit, storage processing unit)
14 User classification part (specification part, deletion part)
16 Information presentation part (output part)
22 Evaluation value table (storage unit)

Claims (6)

In the item search screen, get the information of the search item selected when the user searched for the item,
Storing information on the search items acquired in the acquisition process in a storage unit that stores information on search items selected by a plurality of users in the past;
Referring to the storage unit, identify a user similar to the user,
Outputting information on items referred to in the past by the similar user as information on items recommended for the user,
A recommended item output method characterized in that a computer executes processing.   In the process of outputting, information on items specified from the search item selected by the user on the search screen and the value of the input search item among items of information referred to by the similar user in the past, The recommended item output method according to claim 1, wherein information is output as recommended item information.   3. The computer further executes a process of deleting information on the search item stored in the storage unit based on a time corresponding to a price of the searched item. Recommended item output method.   In the output process, information on an item referred to by the similar user within a period determined based on an item searched by the user is output as item information recommended for the user. The recommended item output method as described in any one of Claims 1-3. In the item search screen, get the information of the search item selected when the user searched for the item,
Storing information on the search items acquired in the acquisition process in a storage unit that stores information on search items selected by a plurality of users in the past;
Referring to the storage unit, identify a user similar to the user,
Outputting information on items referred to in the past by the similar user as information on items recommended for the user,
A recommended item output program which causes a computer to execute processing. In the item search screen, an acquisition unit that acquires information on a search item selected when the user searches for an item;
A storage processing unit that stores information on the search items acquired in the acquisition process in a storage unit that stores information on search items selected by a plurality of users in the past;
A specifying unit for specifying a user similar to the user with reference to the storage unit;
An output unit that outputs information on items referred to by the similar user in the past as item information recommended for the user;
A recommended item output device comprising:

Images