KR101745604B1 - Item recommend system and method for recommending item - Google Patents

Abstract

In one embodiment of the present invention, a user information storage unit for storing a plurality of information of a user accessing a web site, a user utilization information preprocessing unit for preprocessing the information stored in the user information storage unit and converting the information into graph data, A graph database for storing the graph data converted by the preprocessing unit in the user-used information preprocessing unit, a graph database for computing specific data among the items stored in the graph database, And an recommendation unit that recommends the item selected by the operation unit to the user in one or more ways.

Description

Item recommendation system and item recommendation method

Embodiments of the present invention relate to an item recommendation system and an item recommendation method.

Due to the rapid growth of e - commerce, consumers often judge online stores at the same level as well as in - store stores, and in some cases they prefer online shops.

In addition, due to the development of such electronic commerce technology, online auction sites, open markets, social commerce, and various other types of purchase sites are being implemented on-line in addition to simple online shopping malls.

The online purchase method using the e-commerce can easily access the goods, the tickets and other items, and can easily visit and purchase without any need of the user to visit the in-store, so that the time is getting longer And the method is also expanding.

However, online purchases using e-commerce can not be viewed by the user directly by eyes, and since the user visits the site and contacts only the information provided on the screen, it is easy for the user to easily receive various information not. This is especially true for users who are not frequent visitors to a particular site or who are not familiar with e-commerce purchases.

In particular, in recent years, the number of stores that can be purchased online, for example, the number of users who use the websites, increases exponentially, and the number of items handled by the websites increases, making it difficult to provide specific information to users .

Therefore, there is a growing need for a system and method for recommending appropriate items to users.

However, it is not easy to handle and process the information about the users who visit the website and the items provided by the website, and in particular, by increasing the number of users and items, There is a limit to progress.

Embodiments of the present invention provide an item recommendation system and an item recommendation method using the item recommendation system, which can facilitate recommendation work on items to users visiting a web site.

In one embodiment of the present invention, a user information storage unit for storing a plurality of information of a user accessing a web site, a user utilization information preprocessing unit for preprocessing the information stored in the user information storage unit and converting the information into graph data, A graph database for storing the graph data converted by the preprocessing unit in the user-used information preprocessing unit, a graph database for computing specific data among the items stored in the graph database, And an recommendation unit that recommends the item selected by the operation unit to the user in one or more ways.

The similarity determination unit may further include a similarity determination unit that determines a similarity between the items in the embodiment, and the operation unit can select the particular item using the similarity between the items from the similarity determination unit.

In the present embodiment, the similarity determination unit may calculate the similarity between items using an offline arrangement processing method.

In the present embodiment, the operation unit may include an associated user selecting unit for selecting an associated user related to the user, and an associated item selecting unit for selecting an associated item related to the related user.

In the present embodiment, the related user selected by the associated user selecting unit is another user having a relationship with one of the users or items related to similar items, The associated item may be an item having a relationship with the other user or items similar to the one having no relation with the one user.

In the present embodiment, the user information storage unit may include a user information first storage unit and a user information second storage unit for storing different information.

In the present embodiment, the user use information first storage unit may store basic information of a user and one or more information related to a user's activity.

In this embodiment, one or more pieces of information related to the activity of the user may include purchase information or shopping cart information.

In the present embodiment, the user-use-information second storage unit may store user's visit information.

In this embodiment, a real-time processing unit for updating the user visit information in real time may be further included.

In the present embodiment, the real-time processing unit may process the visit to determine whether the visit of the user is the first visit.

In the present embodiment, the user-use-information storage unit may further include a temporary storage unit, and the temporary storage unit may receive the information stored in the first user-information storage unit and the second user- To the user use information preprocessing unit.

In the present exemplary embodiment, the temporary storage unit may temporarily store information in the form of a message queue.

In the present embodiment, the user-utilization-information preprocessing unit forms a relation between the user and a plurality of items provided in the website, and can convert the relation into graph data on the basis of the relation.

According to another embodiment of the present invention, there is provided a method of browsing a web site, comprising: acquiring a plurality of usage information of a user accessing a web site; converting a plurality of usage information of the user into preprocessed graph data; Selecting one or more specific items from items provided in the website by calculating specific data from the data stored in the graph database and recommending the selected item to the user in one or more ways; And starts an item recommendation method including the item.

The method may further include calculating a degree of similarity between the items in the embodiment, and the step of selecting the specific item may select the specific item using the degree of similarity between the items.

In the present embodiment, the step of selecting the specific item may include a step of identifying an item having a relationship with a user of the user, and a step of selecting an item having a similarity to the identified item.

In this embodiment, the step of calculating the degree of similarity between the items may calculate the degree of similarity between the items by the offline placement processing method.

In the present exemplary embodiment, the step of calculating the similarity between the items may receive the user's implicit preference of the user-use information, and may calculate the similarity between the items using the implicit preference.

In the present embodiment, the implicit preference may include purchase information or visit information of the user.

In the present exemplary embodiment, the step of selecting one or more specific items among the items provided in the website by searching specific data among the data stored in the graph database may include selecting an associated user related to the user And associating items associated with the associated user.

In the present exemplary embodiment, the step of selecting an associated user related to the user may include selecting an item having a relation with one of the users or another user having a relationship with similar items, And the step of selecting an item may include selecting items having no relationship with the one user among the items having a relationship with the other user or similar items.

In the present embodiment, the step of acquiring the user use information may include storing and using one or more pieces of information related to the user's basic information and the user's activities.

In this embodiment, one or more pieces of information related to the activity of the user may include purchase information or shopping cart information.

In the present embodiment, the step of acquiring the user use information may include a step of storing and using the visit information of the user.

In the present embodiment, the step of acquiring the user use information may further include updating the user visit information in real time.

In the present embodiment, the step of updating the user visit information in real time may further include a step of determining whether the visit of the user is a first visit.

The method may further include storing the usage information in a temporary storage unit after acquiring a plurality of usage information of a user connected to the web site in the present embodiment, And the forming step uses information stored in the temporary storage unit.

The step of storing the usage information in the temporary storage unit after acquiring a plurality of usage information of the user connected to the website may include acquiring a plurality of usage information of the user and temporarily storing the information in the form of a message queue As shown in FIG.

In the present embodiment, the step of pre-processing a plurality of utilization information of the user and forming the user information as graph data comprises forming a relation between the users and a plurality of items provided by the website, And converting the data into data.

Other aspects, features, and advantages will become apparent from the following drawings, claims, and detailed description of the invention.

The item recommendation system and the item recommendation method using the item recommendation system according to the present embodiment can facilitate the recommendation work on the items to the users visiting the website.

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a schematic diagram for illustrating an item recommendation system and its operation according to an embodiment of the present invention. FIG.
FIG. 2 is a view showing an alternative embodiment of the user-used information storage unit of the item recommendation system of FIG. 1. FIG.
Fig. 3 is a diagram showing a modification of the item recommendation system of Fig. 1. Fig.
4 is a schematic flowchart sequentially illustrating a method of recommending an item using the item recommendation system of FIG.
5 is a flowchart illustrating an alternative embodiment of step S5 of FIG.
6 is a flowchart illustrating the modification of Fig.
FIG. 7 is a graph converted through the preprocessing step of FIG.
FIG. 8 is a diagram for explaining item similarity calculation of FIG. 4; FIG.
FIG. 9 is a view for schematically illustrating an item recommendation system and its operation according to another embodiment of the present invention.
FIG. 10 is a schematic flowchart sequentially illustrating a method of recommending an item using the item recommendation system of FIG. 9. FIG.
11 is a flowchart for explaining an alternative embodiment of step C5 of Fig.
12 is a graph converted through the preprocessing step of FIG.
13 is a diagram for schematically illustrating an item recommendation system and its operation according to another embodiment of the present invention.
14 is a schematic flowchart sequentially illustrating a method of recommending an item using the item recommendation system of FIG.
Figs. 15A and 15B are flowcharts illustrating respective alternative embodiments of step P5 of Fig. 14, respectively.
16 is a graph converted through the preprocessing step of FIG.
Fig. 17 is a diagram for explaining the item similarity degree calculating step of Fig. 14; Fig.

BRIEF DESCRIPTION OF THE DRAWINGS The present invention is capable of various modifications and various embodiments, and specific embodiments are illustrated in the drawings and described in detail in the detailed description. The effects and features of the present invention and methods of achieving them will be apparent with reference to the embodiments described in detail below with reference to the drawings. The present invention may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings, wherein like reference numerals refer to like or corresponding components throughout the drawings, and a duplicate description thereof will be omitted .

In the following embodiments, the terms first, second, and the like are used for the purpose of distinguishing one element from another element, not the limitative meaning.

In the following examples, the singular forms "a", "an" and "the" include plural referents unless the context clearly dictates otherwise.

In the following embodiments, terms such as inclusive or possessive are intended to mean that a feature, or element, described in the specification is present, and does not preclude the possibility that one or more other features or elements may be added.

In the drawings, components may be exaggerated or reduced in size for convenience of explanation. For example, the size and thickness of each component shown in the drawings are arbitrarily shown for convenience of explanation, and thus the present invention is not necessarily limited to those shown in the drawings.

In the following embodiments, the x-axis, the y-axis, and the z-axis are not limited to three axes on the orthogonal coordinate system, and can be interpreted in a broad sense including the three axes. For example, the x-axis, y-axis, and z-axis may be orthogonal to each other, but may refer to different directions that are not orthogonal to each other.

If certain embodiments are otherwise feasible, the particular process sequence may be performed differently from the sequence described. For example, two processes that are described in succession may be performed substantially concurrently, and may be performed in the reverse order of the order described.

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a schematic diagram for illustrating an item recommendation system and its operation according to an embodiment of the present invention. FIG.

Referring to FIG. 1, a plurality of users U1, U2, U3,... Un accesses a web site WS via a network NW. The item recommendation system 100 may be connected to the web site WS to receive a plurality of information related to the user. For example, a web site (WS) can be a site including a social commerce site, an online shopping mall, an open market, and various other e-commerce transactions. The network NW may be of various forms capable of connecting a plurality of users U1, U2, U3 ... Un and a web site WS.

In FIG. 1, the item recommendation system 100 is shown separately from the web site WS, but this is an embodiment. That is, as an alternative embodiment, the item recommendation system 100 may be integrated with the web site WS, and the item recommendation system 100 may be provided in the web site WS. As another alternative embodiment, item recommendation system 100 and web site WS may be formed to serve some functions.

Referring to FIG. 1, the item recommendation system 100 mainly includes a user use information storage unit 110, a user utilization information preprocessing unit 120, a graph database 130, an operation unit 150, a similarity determination unit 160 And a recommendation unit 170. [

The user use information storage unit 110 includes a user use information first storage unit 101 and a user use information second storage unit 102. The user use information first storage unit 101 and the user use information second storage unit 102 store different user usage related information. Details of this will be described later.

The user use information preprocessing execution unit 120 proceeds to a preprocessing step to transform user usage information stored in the user use information storage unit 110 into a form suitable for storing in the graph database 130. [ More details will be described later.

The graph database 130 stores information modified appropriately through the user-use information preprocessing unit 120, that is, information transformed into a graph form.

The item similarity degree determiner 160 calculates and determines the similarity degree between the items.

The operation unit 150 performs the operation using the user use information stored in the graph database 130 and the item similarity determined by the item similarity degree determiner 160 when necessary.

The item recommendation system 100 of the present embodiment will be described in more detail.

The user use information first storage unit 101 of the user use information storage unit 110 stores basic information of the user and one or more information related to the activity of the user. The basic information of the user stored in the user information first storage unit 101 may be information for identifying the user. For example, the user's basic information may be a user's name, a user's ID, a nickname, a password, or an e-mail account. One or more pieces of information related to the activity of the user stored in the user-used information first storage unit 101 may be real-time activity information of the user. For example, one or more pieces of information associated with a user's activity may be purchase information or shopping cart information. That is, one or more pieces of information related to a user's activity may be an item purchased by a user through a web site (WS) or an item stored by a user in a shopping cart through a web site (WS). Of course, the above example is one example for explaining the present embodiment, and it is possible to store various real-time activity information of the user.

The user use information second storage unit 102 of the user use information storage unit 110 may store the visit information of the user. That is, the visiting information may be stored in the user-used information second storage unit 102 when the user visits the web site WS via the network NW.

The user use information storage unit 110 may have various forms.

FIG. 2 is a diagram illustrating an alternative embodiment of the user-used information storage unit 110 of the item recommendation system 100 of FIG. For the sake of convenience of explanation, different things from those described in the above embodiments will be mainly described.

Referring to FIG. 2, the user use information storage unit 110 'of the alternative embodiment of the present invention includes a user usage information first storage unit 101', a user usage information second storage unit 102 ', a real time processing unit (SR) And a temporary storage unit (MQ).

The user use information first storage unit 101 'stores basic information of a user and one or more information related to a user's activity. The user use information first storage unit 101 'may include a Relational Data Base Management System (RDBMS) to easily store such information. In addition, as an alternative embodiment, the user use information first storage unit 101 'can easily process a large amount of data including the mongo DB.

The user use information second storage unit 102 'may store the visit information of the user. For example, the user use information second storage unit 102 'stores visit information of a user in the form of a weblog.

The real-time processing unit SR updates the visit information of the user stored in the user-used information second storage unit 102 'in real time. In addition, the real-time processing unit SR judges whether the user U1, U2, U3 ... Un has no existing visit record as shown in FIG. 2, In the case of the first visitor, the visiting information of the user is processed so as to know that it is the first visitor, and then stored in a temporary storage unit (MQ) to be described later.

The temporary storage unit MQ receives information stored in the user-used information first storage unit 101 ', that is, one or more pieces of information related to the user's basic information and the user's activities. Also, the temporary storage unit (MQ) receives the information processed through the real-time processing unit (SR) of the visit information of the user stored in the second user storage unit 102 '.

The temporary storage unit (MQ) makes it easy to perform preprocessing in the user-used information preprocessing unit 120 by using information of the user. For example, the temporary storage unit MQ may temporarily store user information in the form of a message queue.

The user-used information storage unit 110 'as an optional embodiment of the present invention described above with reference to FIG. 2 is one of various examples that the item recommendation system 100 of the present invention can include. That is, it may include structures that share some configurations with those shown in FIG. 2, or embodiments that further include additional additional configurations in addition to those shown in FIG. 2.

The user-used information preprocessing unit 120 is connected to the user-used information storage unit 110.

The user use information preprocessing execution unit 120 performs preprocessing to fetch user use information from the user use information storage unit 110 and store the user use information in the graph database 130. [

The user use information preprocessing execution unit 120 forms a relation between a plurality of users U1, U2, U3 ... Un and the plurality of items shown in FIG. 1, (U1, U2, U3 ... Un) and a plurality of items.

The graph database 130 stores graph data transformed by the user-use-information preprocessing unit 120, that is, data representing graphs of relationships between users and a plurality of items. The graph data stored in the graph database 130 may display the relationship between users and items as nodes, relations, and properties.

The details of this will be further described with reference to FIG. 7 to be described later.

The graph database 130 can use various databases, for example, Neo4J which is an open type DB based on Java.

In addition, the user-used information preprocessing unit 120 may be configured independently of the graph database 130 as shown in FIG. However, the present embodiment is not limited to this, and the user utilization information preprocessing unit 120 may be formed integrally with the graph database 130.

The similarity determination unit 160 calculates and determines similarities between items. The similarity determination unit 160 may calculate and determine similarities between items using an offline batch processing method.

The similarity determination unit 160 can calculate and determine the similarity between items in various ways. For example, the user's implicit preference can be used. That is, the similarity degree between the items can be calculated and determined using the user's purchase, visit, or other implicit preferences. Also, the similarity determination unit 160 may calculate the similarity between items periodically, for example, at an interval of one hour, by an offline arrangement processing method. At this time, purchasing, visiting information indicating the user's implicit preference is periodically provided, and the similarity between the items can be calculated and determined.

For this, the similarity degree determiner 160 may have various alternative embodiments. As shown in FIG. 3, the similarity degree determiner 160 " may be connected to the user information storage 110 " have. Fig. 3 is a diagram showing a modification of the item recommendation system of Fig. 1. Fig.

Referring to FIG. 3, the similarity determination unit 160 '' of the item recommendation system 100 '' receives user utilization information, that is, user purchase information and user's visit information from the user utilization information storage unit 110 ' have.

The item recommendation system 100 "of FIG. 3 is substantially similar to the item recommendation system 100" of FIG. 1 except that the user use information storage unit 110 ", the user utilization information preprocessing unit 120 ", the graph database 130 & , A similarity determination unit 160 "and a recommendation unit 170 ".

Except for the relationship between the similarity determination unit 160 "and the user-used information storage unit 110 ", the description of the remaining members is omitted because it is the same as the item recommendation system 100 of FIG.

The degree of similarity between a plurality of items is determined through calculation and determination of similarities between items of the similarity degree determiner 160. Items having similarity to each other can be defined as similar item groups as shown in FIG. 8 will be described more specifically in the following embodiments.

* The calculation unit 150 retrieves specific desired data among the graph data stored in the graph database 130 at a desired time. In addition, the operation unit 150 retrieves the item similarity information obtained by the calculation and determination by the similarity degree determiner 160.

Then, this is calculated to perform the machining step. In particular, the operation unit 150 can perform various operations in real time by calculating data stored in the graph database 130, extract items related to a user based on a specific user, And can easily perform an operation of extracting items having a high degree of similarity.

Also, although not shown, the similarity determination unit 160 may be connected to the user-used information preprocessing unit 120 as an alternative embodiment. Information on the item similarity degree determined by the similarity degree determiner 160 may be transformed into graph data.

The recommendation unit 170 may be connected to the operation unit 150 to recommend items extracted from the operation unit 150 to a specific user. That is, it is possible to perform a recommendation for items having a similarity with a specific user and having a relation of purchase, visit, etc., and items having no relationship with the specific user.

The recommendation unit 170 can perform recommendation work in various ways. For example, the recommendation unit 170 may form a pop-up window to display an item on a web site when the specific user visits the web site, temporarily or permanently display an item on a part of the web site, You may. However, the present embodiment is not limited to this, and it is needless to say that the present invention may include the recommendation unit 170 that performs the recommendation process in various ways.

4 is a schematic flowchart sequentially illustrating a method of recommending an item using the item recommendation system of FIG.

Referring to FIG. 4, the item recommendation method of the present embodiment includes a step of acquiring user information (S1), a preprocessing step S2, a storing step S3, an item similarity calculating step S4, an operation step S5, S6).

The user utilization information acquisition step S1 proceeds in the user utilization information storage unit 110 of the item recommendation system 100 of FIG. 1 as described above. That is, the information of the user, for example, the purchase information of the user, the shopping cart information, or the information of the user is stored by using the user information first storage unit 101 and the user information second storage unit 102 of the user information storage unit 110 And acquires the visit information.

For example, in the form shown in Table 1 below, the user-use-information storage unit 110 may store user's usage information.

I ( ₁ ) I ( ₂ ) I ( ₃ ) I ( ₄ ) U1 purchase visit U2 purchase U3 visit U5 purchase purchase

U1, U2, U3, and U5 in Table 1 refer to different users who access the Web site, respectively. I ( ₁ ), I ( ₂ ), I ( ₃ ), and I ( ₄ ) in Table 1 refer to the different items provided on the website, respectively.

That is, Table 1 will be described in detail as follows.

The user U1 purchases the item I ( ₁ ) and visits the item I ( ₂ ). User U2 has purchased item I ( ₂ ). The user U3 visited the item I ( ₄ ). The user U5 has purchased items I ( ₃ ) and I ( ₄ ).

In addition, when the user-use-information storage unit 110 'shown in FIG. 2 is applied as an alternative embodiment, the information can be stored in the temporary storage unit MQ.

Table 1 above illustrates four users and four items by way of example. In addition, two kinds of user utilization information, namely purchase and visit information, are disclosed. It is needless to say that information for a plurality of users and a plurality of items can be acquired and stored in the user use information acquisition step S1 for convenience of explanation. It is also possible to use other information such as shopping cart information, storage box information, wish list information, and the like in addition to purchase and visit as user use information.

The content related to storing other user-used information is as described in the above embodiments.

In the preprocessing step S2, the user-information-information preprocessing unit 120 acquires and stores the acquired and stored information in the user-used information obtaining step S1 to form a relationship between users and items, and transforms the graph into a graph. I.e., a storage step S3 to be described later.

7 is a graph that is changed through the preprocessing step S2.

For convenience of explanation, FIG. 7 shows graph data obtained by pre-processing using the information of Table 1 as it is.

Referring to FIG. 7, the graph data includes a plurality of nodes, and a relation is formed between the plurality of nodes. Each node has properties U1, U2, U3, U5 and items I ( ₁ ), I ( ₂ ), I ( ₃ ), I ( ₄₎ )) As an attribute. Also, relations have attributes (p) and (v) as attributes.

In Fig. 7, the relationship is shown only in one direction. However, it goes without saying that a bi-directional relation may be formed between each node as an example.

Referring to FIG. 7, all of the data shown in Table 1 are shown. That is, a user (U1) is purchased with respect to item (I ₍₁₎₎ purchased, and item to (I ₍₂₎₎ the item (I ₍₂₎₎ by the visit information, the user (U2) with respect to information, user (U3) the item (I ₍₄₎₎ visit information and the user (U5) with respect to item (I ₍₃₎₎ and the item (I ₍₄₎₎ the information is purchased is displayed as a graph the data with respect to the .

That is, in the preprocessing step S2, the user utilization information preprocessing unit 120 processes the information shown in Table 1 to generate graph data corresponding to Table 1. [

The information between each user U1, U2, U3, U5 and each item I ( ₁ ), I ( ₂ ), I ( ₃ ), I ( ₄ ) can be easily known through the graph data. Information on the relation between each user U1, U2, U3 and U5 and each item I ( ₁ ), I ( ₂ ), I ( ₃ ), I ( ₄ ) The relationship between the users U1, U2, U3 and U5 can be known.

In the storage step S3, the graph data formed using the user-used information preprocessing unit 120 is stored in the graph database 130. [ As described in the above embodiments, the user-utilization-information preprocessing unit 120 and the graph database 130 may be separate independent members, or may be in one integrated form as another alternative embodiment.

In the item similarity calculation step S4, the similarity between items is calculated and determined. In the item similarity calculation step S4, similarity between items is calculated by various methods. That is, as described above, the degree of similarity between items can be calculated and determined by an offline batch processing method.

In addition, the implicit preference of the user can be used. That is, the similarity degree between the items can be calculated and determined using the user's purchase, visit, or other implicit preferences. Also, the similarity determination unit 160 may calculate the similarity between items periodically, for example, at an interval of one hour, by an offline arrangement processing method. At this time, purchasing, visiting information indicating the user's implicit preference is periodically provided, and the similarity between the items can be calculated and determined.

3, the similarity determination unit 160 " may be connected to the user-used information storage unit 110 " to receive information.

The item similarity calculation step S4 may be performed to select a group for each item having a high degree of similarity as shown in FIG.

Referring to Figure 8 of the items contained in the group A _{(I (1), I (} 3), I (4), I (7), I (10), I (13) ...) the degree of similarity to eunseo Items I ( ₂ ), I ( ₅ ), I ( ₆ ), I ( ₁₅ ), I ( ₁₉ ), I ( ₂₀ ) Items.

The group selected by the item similarity calculated in the similarity calculation step S4 and the similarity obtained through the similarity can be stored in a predetermined storage. For example, in the similarity determination unit 160 shown in FIG. 1, and may be stored in a separate storage, notably a temporary storage, not shown.

As another example, it can be stored in the temporary storage unit MQ shown in FIG. In this case, as an alternative embodiment, the item similarity information stored in the temporary storage unit MQ may be included in the above-described graph data by performing the preprocessing step S2 and stored in the graph data storage step S3 .

In the calculation step S5, the calculation is performed using the information obtained through the storage step S3 and the item similarity calculation step S4. In particular, in the calculation step S5, the stored graph data is calculated in the storage step S3 to perform various operations in real time, and items related to the user are extracted based on a specific user, It is possible to easily perform an operation of extracting items having a high degree of similarity.

The calculation step S5 may take various forms, which will be described with reference to Fig. 5 is a flowchart illustrating an alternative embodiment of step S5 of FIG.

Referring to FIG. 5, the operation step S5 includes a step S51 of recognizing an item having a relation with a user and a step S52 of identifying items having high similarity to the item obtained in the step S51.

Step S51 of identifying an item having a relation with a user is to identify an item having a relation with a specific user, and referring to FIG. 7, it identifies items having a relation with a specific user, for example, the user U1. The item I ( ₁ ) having the relationship between the user U1 and the purchase p and the item I ( ₂ ) having the relationship between the user U1 and the visit v.

In the step S52 of recognizing the items having high similarity to the items identified in the step S51, the items having the similarity with the items I ( ₁ ) and the items having the similarity with the items I ( ₂ ) . Referring to FIG. 8, items having similarity to items I ( ₁ ) are items I ( ₃ ), I ( ₄ ), I ( ₇ ), I ( ₁₀ ), I ( ₁₃ ) Items I ( ₅ ), I ( ₆ ), I ( ₁₅ ), I ( ₁₉ ), I ( ₂₀ ) ...) are items having similarity to item I ( ₂ ).

In the recommendation step S6, an operation for recommending the selected items to the user through the operation step S5 is performed. That is, the user (U1) Item _{(I (3), I (} 4), I (7), I (10), I (13) ...) at least one of a like or the item (I ( ₅ ), I ( ₆ ), I ( ₁₅ ), I ( ₁₉ ), I ( ₂₀ )

The recommendation step S6 can be recommended in various ways, and it can be performed in various ways as described above. For example, a specific user U1 may form a pop-up window so that items are displayed on the website at the time of visiting, The user can temporarily or permanently display the item on a part of the screen of the mobile phone, or announce items by e-mail. However, the present embodiment is not limited to this, and it goes without saying that the recommendation process can be performed in various ways.

It is needless to say that the description of the user U1 as a specific user is for convenience of explanation and can perform recommendation work for any one of the other users. It is needless to say that the number of users and the number of items shown in the present embodiment and drawings are for convenience of description and are not limited thereto.

The item recommendation system 100 of the embodiments of the present invention and the recommendation method using the same can easily store user-use information through the user-used information storage unit 110, and as an alternative embodiment, You can mark it as a visitor and save it.

Then, the information stored in the user-use-information storage unit 110 is processed in the user-use-information preprocessing unit 120 to transform the relation between the user and the items into simplified graph data. The graph data is stored in the graph database 130.

The information between the user and the items is not a big problem when the user and the items are small. However, when the number of users and the number of items are increased and the relation type (for example, purchase, visit) increases, It is not easy to use such stored information.

However, in the present embodiment, information between the user and the items is stored in the graph database 130 in the form of graph data. Through this, the relationship between users and items can be easily grasped and analyzed. In particular, only the graph data is stored in the graph database 130, and no separate operation is performed on the relationship between users and items. It is possible to improve the utilization ability of the user-utilized information through the graph database 130 because unnecessary calculation operations are not performed.

Further, the graph database 130 can be used to easily store additional information using nodes and relations.

In addition, the operation unit 150 performs various operations by performing an operation operation using the relation between the users and the items, which are grasped by the easy method in the processing step. In other words, rather than computing all the data stored in the form of graph data, only the information including a relation related to a necessary user or a required item is performed, and only the operation is performed by performing the operation on the information.

Accordingly, it is possible to perform a quick calculation in real time, so that it is possible to easily carry out calculations based on the latest information at a desired time, for example, a relation information sorting operation for each user and items.

In addition, since the recommendation work based on the item is judged by judging the similarity between the items, the speed of recommendation work can be easily increased.

In this case, in the present embodiment, the degree of similarity between items can be calculated. In an alternative embodiment, the degree of similarity can be calculated by using an off-line arrangement method, and the degree of similarity between items can be easily calculated using the customer's implicit preference. This can increase the efficiency of customer referral work.

FIG. 9 is a view for schematically illustrating an item recommendation system and its operation according to another embodiment of the present invention.

For convenience of explanation, the description will be focused on the differences from the above-described embodiment.

Referring to FIG. 9, a plurality of users U1, U2, U3 ... Un accesses a web site WS via a network NW. The item recommendation system 200 may be connected to the web site WS to receive a plurality of information related to the user. The contents related to the web site WS are the same as those of the above-described embodiment, and thus a detailed description thereof will be omitted.

9, the item recommendation system 200 mainly includes a user-use information storage unit 210, a user-used information preprocessing unit 220, a graph database 230, an operation unit 250, and a recommendation unit 270. [ . The operation unit 250 includes an associated user selection unit 251 and an associated item selection unit 252.

The user use information storage unit 210 includes a user usage information first storage unit 201 and a user usage information second storage unit 202. The user use information first storage unit 201 and the user use information second storage unit 202 store different user usage related information. The details of this are the same as those of the above-described embodiment and thus will not be described.

The user use information preprocessing execution unit 220 proceeds to a preprocessing step in order to transform the user use information stored in the user use information storage unit 210 into a form suitable for storing in the graph database 230. [ The detailed contents are the same as those of the above-mentioned embodiment and will be omitted.

The graph database 230 stores information appropriately modified through the user-used information preprocessing unit 220. [

The arithmetic operation unit 250 performs arithmetic operation using user utilization information stored in the graph database 230 when necessary.

The operation unit 250 includes an associated user selection unit 251 and an associated item selection unit 252. The associated user selection unit 251 selects another user having a relationship with items having a relationship with a specific user. The associated item selection unit 252 selects an item having no relationship with the specific user among the items related to the user selected by the related user selection unit 251. [ Details of this will be described later.

Although not shown, the user-used information storage unit 210 of the item recommendation system 200 of the present embodiment may have various optional embodiments, and may have the embodiment of FIG. 2 described above, for example. That is, the user use information storage unit 210 of the item recommendation system 200 of the present embodiment may include a user use information first storage unit, a user use information second storage unit, a real time processing unit, and a temporary storage unit. The detailed contents including the real-time processing unit and the temporary storage unit are the same as those of the embodiment of FIG. 2 described above, and therefore, detailed description thereof will be omitted.

The user use information preprocessing execution unit 220 is connected to the user use information storage unit 210 to fetch the user use information from the user use information storage unit 210 and store the user use information in the graph database 230 .

The user use information preprocessing execution unit 220 forms a relation between a plurality of users U1, U2, U3 ... Un and the plurality of items shown in FIG. 9, (U1, U2, U3 ... Un) and a plurality of items.

The graph database 230 stores graph data modified in the user-use-information preprocessing unit 220, that is, data representing graphs of relationships between users and a plurality of items. The graph data stored in the graph database 230 may display the relationship between users and items as nodes, relations, and properties.

The details of this will be further described with reference to FIG.

An example of the graph database 230 is the same as the above-described embodiment. Also, as in the above-described embodiment, the user-utilization-information preprocessing unit 220 may be configured independently of the graph database 230, A user-used information preprocessing unit 220 may be formed.

The operation unit 250 loads specific desired data among the graph data stored in the graph database 230 at a desired time.

Then, this is calculated to perform the machining step. In particular, the operation unit 250 can perform various operations in real time by calculating data stored in the graph database 230. That is, the associated user selection unit 251 identifies items having a relationship with a specific user, and selects other users having a relationship with these items. The association item selection unit 252 may perform an operation of selecting items having no relationship with the specific user among the items having a relationship with the other users.

The recommendation unit 270 may be connected to the operation unit 250 to recommend items selected by the operation unit 250 to a specific user.

The recommendation unit 270 can perform recommendation work in various ways. For example, the recommendation unit 270 may form a pop-up window to display an item on a web site when the specific user visits the web site, temporarily or permanently display an item on a part of the web site, You may. However, the present embodiment is not limited to this, and it is needless to say that the present invention may include the recommendation unit 270 that performs the recommendation process in various ways.

FIG. 10 is a schematic flowchart sequentially illustrating a method of recommending an item using the item recommendation system of FIG. 9. FIG.

Referring to FIG. 10, the item recommendation method of the present embodiment includes a user use information obtaining step C1, a preprocessing step C2, a storing step C3, an calculating step C5, and a recommending step C6.

The user access information acquisition step C1 proceeds in the user access information storage unit 210 of the item recommendation system 200 of FIG. 9 as described above. That is, the information of the user, for example, the purchase information of the user, the shopping cart information, or the information of the user is stored by using the user use information first storage unit 201 and the user use information second storage unit 202 of the user use information storage unit 210 And acquires the visit information.

For example, in the form shown in Table 2 below, the user use information storage unit 210 may store usage information of a user.

I ( ₁ ) I ( ₂ ) I ( ₃ ) I ( ₄ ) U1 purchase visit purchase U2 visit purchase purchase U3 purchase visit

In Table 2, U1, U2, and U3 indicate different users who access the Web site, respectively. I ( ₁ ), I ( ₂ ), I ( ₃ ), and I ( ₄ ) in Table 2 refer to the different items provided on the website, respectively.

That is, Table 2 will be described in detail as follows.

The user U1 purchases the item I ( ₁ ) and the item I ( ₃ ), and visits the item I ( ₂ ). User U2 purchases item I ( ₂ ) and item I ( ₄ ) and visits item I ( ₁ ). The user U3 purchases the item I ( ₃ ) and visits the item I ( ₄ ).

Also, as an alternative embodiment, when the user use information storage unit 210 applies the structure of the alternative embodiment of FIG. 2, the information may be stored in the temporary storage unit.

Table 2 above illustratively shows three users and four items. In addition, two kinds of user utilization information, namely purchase and visit information, are disclosed. It is needless to say that information for a plurality of users and a plurality of items can be acquired and stored in the user-use-information obtaining step C1 for convenience of explanation. It is also possible to use other information such as shopping cart information, storage box information, wish list information, and the like in addition to purchase and visit as user use information.

The content related to storing other user-used information is as described in the above embodiments.

In the preprocessing step (C2), the user-information pre-processing unit (220) acquires and stores the acquired and stored information in the user-utilized information obtaining step (C1) to form a relationship between users and items and transforms the same into a graph. That is, the storage step C3, which will be described later.

12 is a graph that is changed through the preprocessing step (C2).

For convenience of explanation, FIG. 12 shows graph data obtained by pre-processing using the information of Table 2 as it is.

Referring to FIG. 12, the graph data includes a plurality of nodes, and a relation is formed between the plurality of nodes. I2, U3, and I3, I ( ₁ ), I ( ₂ ), I ( ₃ ), I ( ₄ ) As an attribute. Also, relations have attributes (p) and (v) as attributes.

In Fig. 12, the relation is shown only in one direction. However, it goes without saying that a bi-directional relation may be formed between each node as an example.

Referring to FIG. 12, all the data described in Table 2 are shown. That is, the user (U1) the item (I ₍₁₎₎ and the item (I ₍₃₎₎ purchase, the item (I ₍₂₎₎ is a the visit information, the user (U2) with respect to items of (I ₍₂ )) and the item (I ₍₄₎₎ purchase, purchase an item (I (visit information, the user (U3) with respect to item _{1)) (I (3)} ) of, and the item (I ₍₄₎₎ to The visited information is displayed as one graph data.

That is, in the preprocessing step (C2), the user information preprocessing unit 220 processes the information described in Table 2 to generate graph data corresponding to Table 2.

Information between each user U1, U2, U3 and each item I ( ₁ ), I ( ₂ ), I ( ₃ ), I ( ₄ ) can be easily known through the graph data. In addition, through the information for each user (U1, U2, U3) and for each item _{(I (1), I (} 2), I (3), I (4)) relationship (relation) between each user ( U1, U2, U3, U5.

In the storage step (C3), the graph data formed using the user-utilization-information preprocessing unit 220 is stored in the graph database 230. FIG. As described in the above embodiments, the user-utilization-information preprocessing unit 220 and the graph database 230 may be separate independent members, or may be in one integrated form as another alternative embodiment.

In the operation step (C5), the operation is performed using the information obtained through the storage step (C3). In particular, in the operation step (C5), the stored graph data is calculated in the storage step (C3) to perform various operations in real time, and items related to the user are extracted based on a specific user, Lt; RTI ID = 0.0 > and / or < / RTI >

The computation step C5 can take various forms, which will be described with reference to Fig. 11 is a flowchart for explaining an alternative embodiment of step C5 of Fig.

11, the operation step (C5) includes a step (C51) of identifying an item having a relation with a specific user, a step (C52) of identifying users having a relationship with the item identified at the step C51, (C53) of selecting users except for the specific user among the users, a step (C54) of identifying items having a relationship with the users selected in the step C53, and a step (C54) of correlating the items (Step C55) of selecting an item having no item.

Step C51 of identifying an item having a relation with a user is to identify an item having a relation with a specific user, and referring to FIG. 12, it identifies items having a relation with a specific user, for example, the user U1. That is, items I ( ₁ ) and I ( ₃ ) having a relationship of the user U1 and purchase p and an item I ( ₃ ) having the relationship of the user U1 and visit v, ₂ ).

In the step C52 of recognizing the users having relations with the items identified in the step C51, the items identified in the step C51, i.e., items I ( ₁ ), I ( ₂ ) ( ₃ )). That is, the user U1 and the user U2 who are users who have relations with the item I ( ₁ ) and those who have the relationship with the item I ( ₂ ) And grasps the user U1 and the user U3 who are users who have a relationship with the item I ( ₃ ).

In step C53 of selecting the users other than the specific user among the users identified in step C52, the user identified in step C52, namely, the user U1, the user U2, and the user U3, The user U2 and the user U3 are selected as users excluding the user U1. These users are referred to in this embodiment as associated users. That is, the associated users of user U1 are user U2 and user U3. The task of selecting these related users can be performed by the related user selection unit 251 of the item recommendation system 200 of FIG. 9 described above.

In step C54, items related to the users selected in the step C53 are identified by the users U2 and U3, which are related users, respectively.

First, the items related to the user U2 will be grasped. Referring to FIG. 12, an item I ( ₂ ) and an item I ( ₄ ) having a relationship of a user U2 and a purchase p are grasped and a relationship between a user U2 and a visit v And grasps the item I ( ₁ ). Then, identify your item (I ₍₄₎₎ having the relation of the item (I ₍₃₎₎ and landing (v) having a relationship of (U3) and purchasing (p).

In the step C55 of selecting items not related to the specific user among the items identified in the step C54, the items identified as I (I ( ₁ )), (item the I ₍₂₎₎ and the item (I _(4)), and the user (according to item (I ₍₃ having a relationship with the U3))) and the item (I ₍₄₎₎ items from having no relation with the user (U1) . For convenience, these items are referred to in this embodiment as associated items.

Specifically, the item (I _(1)), the item (I _(2)), the item (I ₍₃₎₎ and the item (I ₍₄₎₎ of the item having no relation with the user (U1) Item (I ₍₄ ). In this case, the associated item of the user U1 is the item I ( ₄ ). The selection of these related items may be performed by the associated item selection unit 252 of the item recommendation system 200 of FIG. 9 described above.

In the recommendation step (C6), an operation for recommending the selected items to the user is performed through the operation step (C5). That is, the item U ( ₄ ) can be recommended to the user U1, which is an item associated with the user U1.

Recommendation step C6 may be recommended in a variety of ways, and may proceed in a variety of ways as described above, such as forming a pop-up window for a particular user U1 to display an item on the website at visit, The user can temporarily or permanently display the item on a part of the screen of the mobile phone, or announce items by e-mail. However, the present embodiment is not limited to this, and it goes without saying that the recommendation process can be performed in various ways.

It is needless to say that the description of the user U1 as a specific user is for convenience of explanation and can perform recommendation work for any one of the other users.

The item recommendation system 200 and the recommendation method using the item recommendation system according to the embodiments of the present invention can easily store user-use information through the user-use information storage unit 210 and, as an alternative embodiment, You can mark it as a visitor and save it.

The user information pre-processing unit 220 preprocesses the information stored in the user-used information storage unit 210 to transform the relation between the user and the items into simplified graph data. The graph data is stored in the graph database 230.

Information between users and items is not a big problem when users and items are small, but it is not easy to store them when the number of users and the number of items increases and the relationship type (for example, purchase, visit) increases, It is not easy to use such stored information.

However, in the present embodiment, information between the user and the items is stored in the graph database 230 in the form of graph data. Through this, the relationship between users and items can be easily grasped and analyzed. In particular, only the graph data is stored in the graph database 230, and no separate operation is performed on the relationship between users and items. It is possible to improve the ability to utilize the user-utilized information through the graph database 230 since unnecessary calculation operations are not performed.

Further, the graph database 230 can be used to easily store additional information using nodes and relations.

In addition, the operation unit 250 carries out various operations by performing the calculation operation using the relation between the users and the items grasped by the easy method in the processing step. In other words, rather than computing all the data stored in the form of graph data, only the information including a relation related to a necessary user or a required item is performed, and the operation is performed by performing only the operation related thereto.

Accordingly, it is possible to perform a quick calculation in real time, so that it is possible to easily carry out calculations based on the latest information at a desired time, for example, a relation information sorting operation for each user and items.

In particular, it is possible to select other users having an association with items having a relation with a specific user, i.e., an associated user, and select items having no relationship with a specific user, i.e., related items, Since recommendation work is carried out, recommendation work can be effectively performed in real time. That is, only the information about the users who will perform the recommendation work and the information about the users related thereto can be retrieved from the graph database and can be operated in a short time, so that the recommendation work can be easily performed in real time at any time.

13 is a diagram for schematically illustrating an item recommendation system and its operation according to another embodiment of the present invention.

For convenience of explanation, the description will be focused on the differences from the above-described embodiment.

Referring to FIG. 13, a plurality of users U1, U2, U3,... Un accesses a web site WS via a network NW. The item recommendation system 300 may be connected to the web site WS to receive a plurality of information related to the user. The contents related to the web site WS are the same as those of the above-described embodiment, and thus a detailed description thereof will be omitted.

Referring to FIG. 13, the item recommendation system 300 includes a user use information storage unit 310, a user utilization information preprocessing unit 320, a graph database 330, an operation unit 350, a similarity determination unit 360 And a recommendation unit 370. The operation unit 350 includes an associated user selection unit 351 and an associated item selection unit 352. [

The user use information storing unit 310 includes a user using information first storing unit 301 and a user using information second storing unit 302. [ The user use information first storage unit 301 and the user use information second storage unit 302 store different user usage related information. The details of this are the same as those of the above-described embodiment and thus will not be described.

The user use information preprocessing unit 320 proceeds to a preprocessing step to transform the user use information stored primarily in the user use information storage unit 310 into a form suitable for storing in the graph database 330. [ The detailed contents are the same as those of the above-mentioned embodiment and will be omitted.

The graph database 330 stores the modified information appropriately through the user-used information preprocessing unit 320.

The item similarity degree determiner 360 calculates and determines the degree of similarity between the items.

The operation unit 350 performs the operation using the user use information stored in the graph database 330 and the item similarity information determined by the item similarity degree determiner 360 when necessary.

The operation unit 350 includes an associated user selection unit 351 and an associated item selection unit 352. [ The associated user selection unit 351 selects another user having a relationship with items related to a specific user or a user having a relationship with items similar to those related to a specific user. The associated item selection unit 352 selects an item having a relationship with the user selected by the associated user selection unit 351 or an item having no relation with the specific user among similar items. Details of this will be described later.

Although not shown, the user use information storage unit 310 of the item recommendation system 300 of the present embodiment may have various optional embodiments, and may have the embodiment of FIG. 2 described above, for example. That is, the user use information storage unit 310 of the item recommendation system 300 of the present embodiment may include a user use information first storage unit, a user use information second storage unit, a real time processing unit, and a temporary storage unit. The detailed contents including the real-time processing unit and the temporary storage unit are the same as those of the embodiment of FIG. 2 described above, and therefore, detailed description thereof will be omitted.

The user-used information preprocessing unit 320 is connected to the user-used information storage unit 310.

The user use information preprocessing unit 320 performs preprocessing to fetch user use information from the user use information storage unit 310 and store the user use information in the graph database 330. [

The user use information preprocessing unit 320 forms a relation between a plurality of users U1, U2, U3 ... Un and a plurality of items shown in FIG. 13, (U1, U2, U3 ... Un) and a plurality of items.

The graph database 330 stores graph data modified by the user-use-information preprocessing unit 320, that is, data representing graphically the relationship between users and a plurality of items. The graph data stored in the graph database 330 may display the relationship between users and items as nodes, relations, and properties.

Details related to this will be described later.

An example of the graph database 330 is the same as the above-described embodiment. Also, as in the above-described embodiment, the user-utilization-information preprocessing unit 320 may be configured independently of the graph database 330, A user-used information preprocessing unit 320 may be formed.

The similarity determination unit 360 calculates and determines the similarity between the items. The similarity determination unit 360 may calculate and determine similarities between items in an offline batch processing manner.

The similarity determination unit 360 can calculate and determine the similarity between items in various ways, for example, the user's implicit preference can be used. That is, the similarity degree between the items can be calculated and determined using the user's purchase, visit, or other implicit preferences. In addition, the similarity determination unit 360 may calculate the similarities between items periodically, for example, at intervals of one hour, in an offline arrangement processing manner. At this time, purchasing, visiting information indicating the user's implicit preference is periodically provided, and the similarity between the items can be calculated and determined.

3, the similarity degree determiner 360 may also be an alternative embodiment. The similarity degree determiner 360 may be used by the similarity degree determiner 360 as a user And may be connected to the information storage unit 310 to receive information. The detailed contents are the same as those of the above-mentioned embodiment and will be omitted.

The degree of similarity between a plurality of items is determined through calculation and determination of similarity between items of the similarity degree determiner 360, and items having similarity to each other can be defined as a similar item group as shown in FIG. 17 will be described more specifically in the following embodiments.

The calculation unit 350 loads specific desired data among the graph data stored in the graph database 330 at a desired time. In addition, the operation unit 350 retrieves the item similarity information obtained by the calculation and determination by the similarity degree determiner 360.

Then, this is calculated to perform the machining step. In particular, the operation unit 350 can perform various operations in real time by calculating data stored in the graph database 330, extract items related to a user based on a specific user, And can easily perform an operation of extracting items having a high degree of similarity.

The recommendation unit 370 may be connected to the operation unit 350 to recommend items selected by the operation unit 350 to a specific user.

The recommendation unit 370 can perform recommendation work in various ways. An example of the recommendation method of the recommendation unit 370 may be the same as or similar to the one described in the above embodiment, and thus a detailed description thereof will be omitted.

14 is a schematic flowchart sequentially illustrating a method of recommending an item using the item recommendation system of FIG.

Referring to FIG. 14, the item recommendation method of the present embodiment includes a user utilization information acquisition step P1, a preprocessing step P2, a storage step P3, an operation step P5, and a recommendation step P6.

The user utilization information acquisition step P1 proceeds in the user utilization information storage unit 310 of the item recommendation system 300 of FIG. 13 as described above. That is, the information of the user, for example, the purchase information of the user, the shopping cart information, or the information of the user is stored by using the user use information first storage unit 301 and the user use information second storage unit 302 of the user use information storage unit 310 And acquires the visit information.

For example, in the form shown in Table 3 below, the user-use-information storage unit 310 may store user's usage information.

I ( ₁ ) I ( ₂ ) I ( ₃ ) I ( ₄ ) I ( ₅ ) U1 purchase visit purchase U2 visit purchase purchase U3 purchase visit purchase

In Table 3, U1, U2, and U3 indicate different users who access the Web site, respectively. I ( ₁ ), I ( ₂ ), I ( ₃ ), I ( ₄ ), and I ( ₅ ) in Table 1 refer to the different items provided on the website, respectively.

That is, Table 3 will be described in detail as follows.

The user U1 purchases the item I ( ₁ ) and the item I ( ₃ ), and visits the item I ( ₂ ). User U2 purchases item I ( ₂ ) and item I ( ₄ ) and visits item I ( ₁ ). The user U3 purchases item I ( ₃ ) and item I ( ₅ ) and visits item I ( ₄ ).

Also, as an alternative embodiment, when the user use information storage unit 310 applies the structure of the alternative embodiment of FIG. 2, the information may be stored in the temporary storage unit.

Table 3 above illustratively shows three users and five items. In addition, two kinds of user utilization information, namely purchase and visit information, are disclosed. It is needless to say that information for a plurality of users and a plurality of items can be acquired and stored in the user-use-information obtaining step P1 for convenience of explanation. It is also possible to use other information such as shopping cart information, storage box information, wish list information, and the like in addition to purchase and visit as user use information.

The content related to storing other user-used information is as described in the above embodiments.

In the preprocessing step P2, the user-information-preprocessing unit 320 acquires and stores the acquired and stored information in the user-used information obtaining step P1 to form a relationship between users and items, and transforms the generated graph into a graph. That is, the storage step P3, which will be described later.

12 is a graph that is changed through the preprocessing step P2.

For convenience of explanation, FIG. 16 shows graph data obtained by pre-processing using the information of Table 3 as it is.

Referring to FIG. 16, the graph data includes a plurality of nodes, and a relation is formed between the plurality of nodes. Each node has properties U1, U2, U3 and items I ( ₁ ), I ( ₂ ), I ( ₃ ), I ( ₄ ) I ( ₅ )). Also, relations have attributes (p) and (v) as attributes.

In Fig. 16, the relation is shown only in one direction. However, it goes without saying that a bi-directional relation may be formed between each node as an example.

Referring to FIG. 16, all the data described in Table 3 are shown. That is, the user (U1) the item (I ₍₁₎₎ and the item (I ₍₃₎₎ purchase, the item (I ₍₂₎₎ is a the visit information, the user (U2) with respect to items of (I ₍₂ ) And item I ( ₄ ), information visited for item I ( ₁ ), user U3 purchases item I ( ₃ ) and item I ( ₅ ) The information visited for the item I ( ₄ ) is displayed as one graph data.

That is, in the preprocessing step P2, the user information pre-processing unit 320 processes the information described in Table 3 to generate graph data corresponding to Table 3. [

Through the graph data for each user (U1, U2, U3) and for each item _{(I (1), I (} 2), I (3), I (4), I (5)) easily know information between . Further, the information for each user (U1, U2, U3) and for each item _{(I (1), I (} 2), I (3), I (4), I (5)) relationship (relation) between the U1, U2, and U3 can be known through the user interface (U1, U2, U3).

In the storing step P3, the graph data formed using the user-used information preprocessing unit 320 is stored in the graph database 330. [ As described in the above embodiments, the user-utilization-information preprocessing unit 320 and the graph database 330 may be separate independent members, or may be in one integrated form as another alternative embodiment.

In the item similarity calculation step P4, the degree of similarity between the items is calculated and determined. In the item similarity calculation step (P4), similarity between items is calculated by various methods. That is, as described above, the degree of similarity between items can be calculated and determined by an offline batch processing method.

In addition, the implicit preference of the user can be used. That is, the similarity degree between the items can be calculated and determined using the user's purchase, visit, or other implicit preferences. In addition, the similarity determination unit 360 may calculate the similarities between items periodically, for example, at intervals of one hour, in an offline arrangement processing manner. At this time, purchasing, visiting information indicating the user's implicit preference is periodically provided, and the similarity between the items can be calculated and determined.

For this, the similarity determination unit 360 may be connected to the user-used information storage unit 310 to receive information as an alternative embodiment as described above.

The item similarity calculation step P4 can select the group for each item having a high degree of similarity as shown in FIG.

17, the items I ( ₁ ), I ( ₃ ), I ( ₄ ), I ( ₇ ), I ( ₁₀ ), I ( ₁₃ ) Items I ( ₂ ), I ( ₅ ), I ( ₆ ), I ( ₁₅ ), I ( ₁₉ ), I ( ₂₀ ) Items.

The group selected by the item similarity calculated in the similarity calculation step P4 and the similarity obtained through the similarity can be stored in a predetermined storage. For example, within the similarity determination unit 360 shown in FIG. 1, and may be stored in a separate repository, not shown, in particular, a temporary repository.

As another example, the item similarity information stored in the temporary storage unit MQ can be stored in the temporary storage unit MQ as described in the embodiment shown in FIG. 2, and the preprocessing step P2 is performed And then stored in the graph data and stored in the graph data storage step P3.

In the calculation step P5, the calculation is performed using the information obtained through the storage step P3 and the item similarity calculation step P4. In particular, in the calculation step P5, the stored graph data may be calculated in the storage step P3 to perform various operations in real time, and items related to the user may be extracted based on the specific user, Lt; RTI ID = 0.0 > and / or < / RTI >

Operation step P5 may have various alternative embodiments, which are flowcharts that respectively illustrate alternative embodiments of step P5 of Fig. 14, respectively.

Referring to FIG. 15A, the operation step (P5) includes a step (P51) of identifying an item having a relation with a specific user, a step (P52) of selecting items having high similarity to the item identified in the step P51, (P53) of selecting users who are related to the selected item from among the users having a relationship with the selected item in the step P53, determining (P54) items related to the users selected in the step P53, (P55) of selecting an item having no relationship with the specific user.

The step P51 of selecting an item having a relation with the user is to grasp an item having a relation with a specific user, and referring to FIG. 16, it grasps items having a relation with a specific user, for example, the user U1. That is, items I ( ₁ ) and I ( ₃ ) having a relationship of the user U1 and purchase p, ₂ ).

In step P52, items having similarity to the item identified in step P51 are compared with the items I ( ₁ ), I ( ₂ ), and I ( ₃ ) To identify high-ranked items. Referring to FIG. 17, items having similarity to items I ( ₁ ) and I ( ₃ ) are items I ( ₄ ), I ( ₇ ), I ( ₁₀ ), I ( ₁₃ ) .), and the item (I ₍₂₎₎ and the degree of similarity is high items are the items _{(I (5), I (} 6), I (15), I (19), I (20) ...).

That is, finally the item _{(I (4), I (} 7), I (10), I (13) ...) or the item _{(I (5), I (} 6), I (15), I ( ₁₉ ), I ( ₂₀ ) ...) are selected.

In the step P53 of selecting users except for the specific user among the users having relations with the item selected in the above step P52, the user selects the items selected in the above step P52, that is, the item I ( ₁ ) ₃ )), for example, items other than the user U1 among the users having a relationship with the item I ( ₄ ) or the item I ( ₅ ). First, the users having a relationship with item I ( ₄ ) are user U2 and user U3. And, the user having a relationship with the item I ( ₅ ) is the user U3. Among these users, the user U1 is excluded. Therefore, the user U2 and the user U3 are finally selected.

The associated user selecting unit 351 of the calculating unit 350 may perform the selecting operation of the user U2 and the user U3. That is, an operation can be performed to select another user (referred to as an associated user in this embodiment) having a relationship with items similar to items having relations with items having a relation with the specific user U1.

In the step P54 of grasping items related to the users selected in the step P53, that is, the related users, items having relations with the selected users, i.e., the user U2 and the user U3 are identified . 16, the user U2 has a relation with the item I ( ₁ ), the item I ( ₂ ) and the item I ( ₄ ), the user U3 has the relationship with the item I ( ₃ ) Item I ( ₄ ), and item I ( ₅ ). That is, the items identified in step (P54) are items (I _(1)), the item (I _(2)), the item (I _(3)), the item (I ₍₄₎₎ and the item (I ₍₅₎₎ .

Of the items identified in the P54 step, step (P55) for selecting an item having no relationship with the particular user of the items identified in step _{(P54) (I (1)} , I (2), I (3), I ( ₄ ), I ( ₅ )) that are not related to the user (U1). That is the item of the _{(I (1), I (} 2), I (3), I (4), I (5)) the item having a relationship with the user (U1) of the _{(I (1), I (} 2) I ( ₄ ), I ( ₅ )) excluding the items I ( ₃ ) and I ( ₃ ) The association item selection unit 352 of the operation unit 350 performs the item selection operation. That is, an item having a relation with the related user and having no relation with the specific user (referred to as an associated item in this embodiment) is selected.

In the recommendation step P6, an operation for recommending the selected items to the user is performed through the operation step P5. That is, the user U1 can recommend items I ( ₄ ) and I ( ₅ ) to the user U1.

The recommendation step P6 can be recommended in various ways, which is the same as the above-described embodiment, and a detailed description thereof will be omitted.

It is needless to say that the description of the user U1 as a specific user is for convenience of explanation and can perform recommendation work for any one of the other users.

Another embodiment among the various types of selective embodiments of the calculation step P5 will be described.

Referring to FIG. 15B, the operation step P5 'includes a step P51' of identifying an item having a relation with a specific user, a step P51 'of excluding users having a relationship with the item identified at the step P51' A step of selecting a user (P52 '), a step (P53') of identifying an item having a relationship with the users selected in the P52 ', a step (P54' And a step P55 'of selecting items not related to the specific user among the items identified in the step P54'.

The step P51 'of selecting an item having a relation with the user is to identify an item having a relation with a specific user, and referring to FIG. 16, the item having a relationship with a specific user, for example, the user U1 is identified . That is, items I ( ₁ ) and I ( ₃ ) having a relationship of the user U1 and purchase p, ₂ ).

In step P52 'of selecting a user other than the specific user in step P51' among the users having a relationship with the item identified in step P51 ', the items identified in step P51', that is, items I ( ₁ ) ), Item (I ( ₂ )), and item (I ( ₃ )). That is, the user U1 and the user U2 who are users who have relations with the item I ( ₁ ) and those who have the relationship with the item I ( ₂ ) And grasps the user U1 and the user U3 who are users who have a relationship with the item I ( ₃ ). The user (U2) and the user (U3) excluding the user U1, which is a specific user, are finally selected.

The associated user selecting unit 351 of the calculating unit 350 may perform the selecting operation of the user U2 and the user U3. In other words, it is possible to perform an operation of selecting another user (referred to as an associated user in this embodiment) having a relationship with items having a relation with the specific user U1.

In the step P53 'of grasping the items related to the users selected in the step P52', the user U2 and the user U3 grasp the items having the respective relations.

First, the items related to the user U2 will be grasped. 16, an item I ( ₂ ) and an item I ( ₄ ) having a relation of a user U2 and a purchase p are grasped and a relationship between a user U2 and a visit v item (I ₍₁₎₎ to identify and user items with the relationship (U3) and purchasing (p) item (I _(3)), the item (I ₍₅₎₎ and landing (v) having a relationship ( I ( ₄ ).

The item I ( ₁ ), the item I ( ₂ ), and the item I ( ₃ ) are selected in the step P54 'of selecting an item having a high degree of similarity to the item identified in the P53' , Items (I ( ₄ )) and items (I ( ₅ )).

Referring to Figure 17, the item (I _(1)), the item (I ₍₃₎₎ or the item (I ₍₄₎₎ and the degree of similarity is high items are items of _{(I (7), I (} 10), I (13 ) ...), and the items having similarities to items I ( ₂ ) and I ( ₅ ) are items I ( ₇ ), I ( ₁₀ ), I ( ₁₃ ) .

In the step P55 'of selecting items not related to the specific user among the items identified in the step P54', an item having no relationship with the user U1 among the items is selected. Finally, the items I ( ₇ ), I ( ₁₀ ), I ( ₁₃ ) ... and the items I ( ₇ ), I ( ₁₀ ), I ( ₁₃ ) . The association item selection unit 352 of the operation unit 350 performs the item selection operation. That is, an item (referred to as an associated item in this embodiment) that does not have a relationship with the specific user among the items similar to the item having a relationship with the related user is selected.

In the recommendation step P6, an operation for recommending the selected items to the user is performed through the operation step P5. That is, the user (U1) Item _{(I (7), I (} 10), I (13) ...) and the item _{(I (7), I (} 10), I (13) ...) Can be recommended.

The recommendation step P6 can be recommended in various ways, which is the same as the above-described embodiment, and a detailed description thereof will be omitted.

It is needless to say that the description of the user U1 as a specific user is for convenience of explanation and can perform recommendation work for any one of the other users.

The item recommendation system 300 of the present embodiment and the recommendation method using the same can easily store user-use information through the user-use-information storage 310 as in the above-described embodiments, and as an alternative embodiment, You can mark and store the information and the first visitor.

That is, at the machining step through the operation unit 350, various operations are performed by performing calculation operations using the relationship between the users and the items, which are grasped by the above-mentioned easy method. In other words, rather than computing all the data stored in the form of graph data, only the information including a relation related to a necessary user or a required item is performed, and the operation is performed by performing only the operation related thereto.

Accordingly, it is possible to perform a quick calculation in real time, so that it is possible to easily carry out calculations based on the latest information at a desired time, for example, a relation information sorting operation for each user and items.

In addition, since the recommendation work based on the item is judged by judging the similarity between the items, the speed of recommendation work can be easily increased. At the same time, other users having a relationship with items related to a specific user, that is, an associated user, are selected, and items having no relationship with a specific user, i.e., related items, are selected Since recommendation work can be carried out in parallel, various types of recommendation work can be performed in real time. That is, only the information about the users who will perform the recommendation work and the information about the users related thereto can be retrieved from the graph database and can be operated in a short time, so that the recommendation work can be easily performed in real time at any time.

While the present invention has been particularly shown and described with reference to exemplary embodiments thereof, it is evident that many alternatives, modifications and variations will be apparent to those skilled in the art . Accordingly, the true scope of the present invention should be determined by the technical idea of the appended claims.

Embodiments may be represented by functional block configurations and various processing steps. These functional blocks may be implemented in a wide variety of hardware and / or software configurations that perform particular functions. For example, embodiments may include integrated circuit components such as memory, processing, logic, look-up tables, etc., that may perform various functions by control of one or more microprocessors or other control devices Can be employed.

Similar to the components of the present invention may be implemented with software programming or software components, embodiments may include various algorithms implemented in a combination of data structures, processes, routines, or other programming constructs, such as C, C ++ , Java (Java), assembler, and the like. Functional aspects may be implemented with algorithms running on one or more processors. The embodiments may also employ conventional techniques for electronic configuration, signal processing, and / or data processing. Terms such as "mechanism", "element", "means", "configuration" may be used broadly and are not limited to mechanical and physical configurations. The term may include the meaning of a series of routines of software in conjunction with a processor or the like.

The specific implementations described in the embodiments are, by way of example, not intended to limit the scope of the embodiments in any way. For brevity of description, descriptions of conventional electronic configurations, control systems, software, and other functional aspects of such systems may be omitted. Also, the connections or connecting members of the lines between the components shown in the figures are illustrative of functional connections and / or physical or circuit connections, which may be replaced or additionally provided by a variety of functional connections, physical Connection, or circuit connections. Also, unless explicitly mentioned, such as "essential "," importantly ", etc., it may not be a necessary component for application of the present invention.

The use of the terms "above" and similar indication words in the description of the embodiments (in particular in the claims) may refer to both singular and plural. In addition, in the embodiment, when a range is described, it includes the invention to which the individual values belonging to the above range are applied (if there is no description to the contrary), the individual values constituting the above range are described in the detailed description . Finally, the steps may be performed in an appropriate order, unless explicitly stated or contrary to the description of the steps constituting the method according to the embodiment. The embodiments are not necessarily limited to the description order of the steps. The use of all examples or exemplary terms (e.g., etc.) in the examples is for the purpose of describing the embodiments in detail and is not intended to be limited by the scope of the claims, It is not. It will also be appreciated by those skilled in the art that various modifications, combinations, and alterations may be made depending on design criteria and factors within the scope of the appended claims or equivalents thereof.

100, 200, 300: Item recommendation system
110, 110 ' 210, 310:
101, 101 '201, and 301:
102, 102 ' 202, 302:
120, 220, and 320: a user use information pre-
130, 230, 330: a graph database
150, 250, and 350:
MQ: Temporary storage unit
SR: real-time processing unit

Claims (11)

A first user storage unit for storing basic information including identification information of a user accessing a web site and activity information in real time, a second user storage unit for storing visit information of the user, And a temporary storage unit for storing the information in the form of a message queue by receiving the information from the first storage unit and the second storage unit;
A pre-processing unit for receiving and pre-processing information from the temporary storage unit of the user-use-information storage unit to form a relation between a plurality of items provided by the user and the website, and converting the user- Performing part;
A graph database for storing pre-processed and transformed graph data in the user utilization information preprocessing unit;
A degree of similarity determination unit connected to the user-use-information storage unit and determining the degree of similarity between the items based on an implicit preference including purchase or visit of the user as an offline placement processing scheme according to a set time period;
An operation unit for retrieving specific data among the data stored in the graph database and the item similarity information determined by the similarity degree determination unit and selecting one or more specific items among the items provided in the website; And
And a recommendation unit for recommending the item selected by the operation unit to the user in one or more ways,
The user-utilization-information preprocessing unit may associate the mutual association information between the user and the items with a plurality of nodes, a relation between the plurality of nodes, and at least one property having the node and the relation And converts it into graph data,
Each of the user and the items corresponding to an attribute of each of the plurality of nodes,
Wherein a relationship between a node having the user as an attribute and a node having the attribute as an attribute is represented by a line connecting the nodes,
Wherein the relationship between the nodes indicated by the lines has one or more attributes different from each other,
Wherein the attribute of the relationship between the nodes is one of a purchase (P) and a visit (V), the value corresponding to activity information associated with one or more activities of the user on the item through connection to the website,
Wherein a node having the user as an attribute and a node having the attribute as an attribute are connected in a relationship having attributes of purchase (P) and visit (V) ) As attributes are displayed separately. The method according to claim 1,
Wherein the operation unit includes an associated user selecting unit for selecting an associated user related to the user and an associated item selecting unit for selecting an associated item associated with the associated user. 3. The method of claim 2,
And the related user selected by the related user selecting unit is another user having a relationship with one of the users or a relationship with similar items,
The association item selected by the association item selection unit is items having no relationship with the one user among the items having the relation with the other user or similar items,
Wherein the similar items include those determined by the operation unit using similarity determined through the similarity determination unit after determining similarities between the items through the similarity determination unit. The method according to claim 1,
Wherein the at least one item of information related to the activity of the user includes purchase information or shopping cart information. The method according to claim 1,
Wherein the real-time processing unit processes the item to determine whether the visit of the user is the first visit. A first user storage unit storing basic information including identification information of a user accessing a website and real-time activity information, a second user storage unit storing user's visit information, And a temporary storage unit for receiving the information from the first storage unit and the second storage unit and storing the received information in a message queue form, Obtaining usage information;
A relation between a user and a plurality of items provided on the web site is formed by receiving information from the temporary storage unit using a user use information preprocessing unit and preprocessing a plurality of usage information of the user, Converting into graph data on a basis;
Storing the graph data in a graph database;
A degree of similarity determining step of determining the degree of similarity between the items using an implicit preference including a user's purchase or visit using the information of the user-use information storage unit as an offline placement processing method according to a set time period;
Selecting one or more specific items among the items provided in the website by searching for specific data among the data stored in the graph database and item similarity information determined in the similarity determination step; And
Recommending the selected item to one or more methods that inform the user that the selected item is recognized by the user using a recommendation section via a network,
Wherein the step of transforming the graph data into the graph data comprises: correlating information between the user and the items to a plurality of nodes, a relation between the plurality of nodes, and at least one property having the node and the relation. And converts it into graph data,
Each of the user and the items corresponding to an attribute of each of the plurality of nodes,
Wherein a relationship between a node having the user as an attribute and a node having the attribute as an attribute is represented by a line connecting the nodes,
Wherein the relationship between the nodes indicated by the lines has one or more attributes different from each other,
Wherein the attribute of the relationship between the nodes is one of a purchase (P) and a visit (V), the value corresponding to activity information associated with one or more activities of the user on the item through connection to the website,
The node having the attribute as the attribute and the node having the attribute as the attribute are connected in a relationship having attributes of one of purchase (P) and visit (V), and the connection And the relationship of the visit (V) as an attribute are distinguished and displayed. The method according to claim 6,
Wherein the step of selecting the specific item comprises:
Identifying an item having a relationship with a user of the user; And
And selecting an item having a similarity to the identified item. The method according to claim 6,
Wherein the step of selecting one or more specific items among the items provided by the web site by searching specific data among the data stored in the graph database,
Selecting an associated user associated with the user, and selecting an associated item associated with the associated user. 9. The method of claim 8,
Wherein the step of selecting an associated user related to the user selects an item having a relationship with one of the users or another user having a relationship with similar items,
Wherein the step of selecting the associated item is to select items that are not related to the one user among items having a relationship with the other user or similar items,
The similar items include,
Determining the degree of similarity between the items through the similarity determination step, and then selecting the one or more specific items based on the degree of similarity determined through the similarity determination step. The method according to claim 6,
Wherein the one or more pieces of information associated with the activity of the user includes purchase information or shopping cart information. The method according to claim 6,
Wherein the step of updating the user's visit information in real time further includes processing the user's visit to determine whether the visit is the first visit.

Images