WO2014178536A1 - Method for providing recommendation item, and recording medium for recording program and apparatus for same - Google Patents

Abstract

Disclosed are a method for providing a recommendation item, a recording medium for recording a program for the same, and an apparatus for the same. In detail, user log data relating to a web site is collected from a terminal which has accessed a web site through a script inserted into the web site, a recommendation result for one or more recommendation items relating to each source item provided from the web site is calculated using a recommendation algorithm on the basis of the user log data, and a recommendation item list relating to a source item requested from the recommendation result is extracted in response to a recommendation item list request received from a web server of the web site, and is transmitted to the web server.

Description

Method for providing recommended item, recording medium and apparatus for recording program therefor

The present invention relates to a method for providing a recommendation item, and more particularly, to a method for providing a recommendation item for recommending another item related to an item served by a web site, a recording medium, and an apparatus for recording a program therefor.

As the use of the Internet becomes more common, users can acquire contents related to various fields such as news, knowledge documents, games, and videos through the Internet. In addition, many service companies have opened shopping malls and provide users with a huge variety of products and services.

However, since information, products, and services that can be acquired through the Internet are increasing exponentially, a lot of time is required for consumers to find information or products that they really want. Accordingly, various Internet sites provide services that provide appropriate recommendation information such as related video recommendation and product recommendation to visited users, and there are companies that provide a recommendation system as a solution. However, in order to introduce such a recommendation system, a large number of servers capable of processing a large amount of data are basically required, and further development is required to link the recommendation result to an actual service.

Disclosure of Invention An object of the present invention is to propose a method for providing a recommendation item that can be directly applied to its own service without additional development on a web site that provides various contents, products or services, and a recording medium and a device recording a program therefor.

In addition, an object of the present invention is to propose a method for providing a recommended item that can recommend a more suitable item by reflecting the feedback generated from the user in the recommendation result, a recording medium and a device recording a program therefor.

The technical problems to be achieved in the present invention are not limited to the technical problems mentioned above, and other technical problems not mentioned above will be clearly understood by those skilled in the art from the following description. Could be.

In one aspect of the present invention, a service apparatus, comprising: a communication unit for data transmission and reception, a user log data, a storage unit storing a recommendation result, and a website from a terminal connected to a website through a script inserted in the website; Collecting user log data related to the related data, using the recommendation algorithm based on the user log data, calculating recommendation results for one or more recommended items related to each source item serviced by the website, and receiving the result from the web server of the website. And a control unit for extracting a recommendation item list related to the requested source item from the recommendation result in response to the recommendation item list request and transmitting the same to the web server.

Preferably, the controller may calculate a click rate for the recommendation item associated with each source item through the click log data and the view log data for the recommendation item.

Preferably, the controller may exclude a recommendation item whose clickthrough rate is lower than a threshold value for a predetermined period of time from the recommendation result.

Preferably, the controller may classify the user log data into a predetermined condition or arbitrarily into a plurality of groups, and calculate a recommendation result by using a different recommendation algorithm for each group.

Preferably, the controller may calculate a click rate for each group through the click log data and the view log data of the recommended item, and calculate a recommendation result using a recommendation algorithm used for the group having the highest click rate.

Preferably, when the terminal is connected through a link of the recommended item connected to the service device, the controller may obtain click log data for the recommended item through the access URL of the terminal.

Preferably, the controller may extract user data including a predetermined parameter value from a visited URL (Uniform Resource Locator) included in the user log data and obtain the click log data for the recommended item.

Preferably, if there is a recommendation item list request from the web server, the controller may obtain view log data for the recommendation item through the recommendation item list.

Preferably, the controller may extract operation information related to the operation of the terminal from the item information displayed on the visit URL and the visit URL from the visit URL (Uniform Resource Locator) included in the user log data.

Preferably, the controller divides and stores the recommendation result for each source item, and hashes the string including a service identifier for identifying a website, an item identifier for the source item, and a recommendation item identifier for the recommendation item related to the source item. By applying the hash function, you can store the calculated value with the recommended result.

Preferably, the controller may transmit a list of recommended items related to the requested source item in a JavaScript Object Notation (JSON) format or an iframe format.

Preferably, the user log data may include a domain of a web site, a visit URL (Uniform Resource Locator) of the terminal, a user identifier, and visit time information.

Preferably, the user identifier is generated by a web browser cookie of the terminal, and may be generated by combining a random number and a generation time.

According to another aspect of the present invention, there is provided a method for providing a recommendation item, wherein the service device collects user log data related to a web site from a terminal connected to the web site through a script inserted into the web site. Calculating a recommendation result for at least one recommendation item associated with each source item served by the website using a recommendation algorithm based on the user log data; and requesting the recommendation item list request received from the web server of the website by the service apparatus. And in response to extracting a recommendation item list related to the requested source item from the recommendation result and transmitting it to the web server.

Preferably, the service device may further include calculating a click rate for the recommendation item associated with each source item through the click log data and the view log data for the recommendation item.

Preferably, the service device may further include the step of excluding a recommendation item whose recommendation rate is lower than a threshold value for a predetermined period from the recommendation result.

Preferably, in the calculating of the recommendation result, the service device may classify the user log data into a predetermined condition or arbitrarily into a plurality of groups, and calculate the recommendation result by using a different recommendation algorithm for each group.

Preferably, the step of calculating the recommendation result, the service device calculates the clickthrough rate for each group through the click log data and view log data for the recommended item, and calculates the recommendation result by the recommendation algorithm used for the group having the highest click rate. can do.

Preferably, when the terminal is connected through a link of the recommended item to which the service device is connected to the service device, the method may further include obtaining click log data for the recommended item through the access URL of the terminal.

Preferably, the method may further include extracting, by the service apparatus, user data including a predetermined parameter value from a visit URL (Uniform Resource Locator) included in the user log data, as the click log data for the recommended item. .

Preferably, the service device may further include acquiring view log data for the recommended item through the recommended item list when there is a recommendation item list request from the web server.

Preferably, the user device may further include extracting operation information related to the operation of the terminal from the item information displayed on the visit URL and the visit URL from the visit URL (Uniform Resource Locator) included in the user log data.

Preferably, the service device stores and stores the recommended results for each source item, but hashes a string including a service identifier for identifying a website, an item identifier for the source item, and a recommended item identifier for the recommended item related to the source item. The method may further include storing a value calculated by applying the (hash) function together with the recommended result.

According to an embodiment of the present invention, it is possible to directly apply to its own service without additional development on the web site providing various contents, products or services.

In addition, according to an embodiment of the present invention, it is possible to recommend a more suitable item by providing feedback reflected in the recommendation result based on the click / view log data generated from the user.

The effects obtainable in the present invention are not limited to the above-mentioned effects, and other effects not mentioned will be clearly understood by those skilled in the art from the following description. .

BRIEF DESCRIPTION OF THE DRAWINGS The accompanying drawings, included as part of the detailed description in order to provide a thorough understanding of the present invention, provide embodiments of the present invention and together with the description, describe the technical features of the present invention.

1 is a diagram schematically illustrating a system to which a recommended item providing service apparatus according to an embodiment of the present invention may be applied.

2 is a diagram illustrating a configuration of a service device according to an embodiment of the present invention.

3 is a diagram illustrating a method for providing a recommended item according to an embodiment of the present invention.

4 is a diagram illustrating a user log data collection method according to an embodiment of the present invention.

5 is a diagram illustrating a user log data collection method according to an embodiment of the present invention.

6 is a diagram illustrating a method of providing a recommendation item list according to an embodiment of the present invention.

7 is a diagram illustrating a list of recommended items provided in an iframe format according to an embodiment of the present invention.

8 illustrates an example of displaying a recommendation item list provided in an iframe format on a web page according to an embodiment of the present invention.

9 is a diagram illustrating a method of collecting click log data according to an embodiment of the present invention.

10 is a diagram illustrating a click rate for each recommended item according to an embodiment of the present invention.

11 is a diagram illustrating a recommendation rate for each algorithm according to an embodiment of the present invention.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings. The detailed description, which will be given below with reference to the accompanying drawings, is intended to explain exemplary embodiments of the present invention and is not intended to represent the only embodiments in which the present invention may be practiced. The following detailed description includes specific details in order to provide a thorough understanding of the present invention. However, one of ordinary skill in the art appreciates that the present invention may be practiced without these specific details.

In some instances, well-known structures and devices may be omitted or shown in block diagram form centering on the core functions of the structures and devices in order to avoid obscuring the concepts of the present invention.

Throughout the specification, when a part is said to "comprising" (or including) a component, this means that it may further include other components, except to exclude other components unless specifically stated otherwise. do. In addition, the terms “… unit”, “… unit”, “module”, etc. described in the specification mean a unit that processes at least one function or operation, which may be implemented by hardware or software or a combination of hardware and software. have. Also, "a or an", "one", "the", and the like are used differently in the context of describing the present invention (particularly in the context of the following claims). Unless otherwise indicated or clearly contradicted by context, it may be used in the sense including both the singular and the plural.

Specific terms used in the following description are provided to help the understanding of the present invention, and the use of such specific terms may be changed to other forms without departing from the technical spirit of the present invention.

As a result recommended by the present invention to the user, the present invention includes all kinds of goods and services that can be distributed through wired and wireless networks, content such as digitally generated texts, sounds, images, videos, software such as computer programs or applications, and the like. For the convenience of the following description, the term "item" will be described collectively.

1 is a diagram schematically illustrating a system to which a recommended item providing service apparatus according to an embodiment of the present invention may be applied.

Referring to FIG. 1, a system to which a recommended item providing service apparatus according to an embodiment of the present invention may be applied includes a plurality of terminals 10, a web server 20, a service apparatus 30, and a network 40. Can be configured.

The terminal 10 transmits and receives information to and from the web server 20 and / or the service device 30 through the network 40, and refers to a device that can be used by a user who displays such information or receives information from the user. In particular, in the present invention, the terminal 30 may access the web server 20 and receive various information about the item and a list of recommended items provided by the service device 30. Here, the recommended item refers to a list of one or more recommended items related to a specific item, and may be generated based on user log data collected by the service device 30 from the web server 20 and / or the terminal 10. .

In addition, the terminal 10 transmits the selection information on the item selected by the user to the web server 20, thereby using the selected item (for example, purchasing, storing a shopping cart, downloading, transferring to another terminal, etc.). Various functions can be performed. In addition, the terminal 10 may perform a function of transmitting user log data to the service device 30, which will be described later. Preferably, the terminal 10 may transmit / receive information with the web server 20 and / or the service device 30 through a web browser stored therein, but is not limited thereto.

As illustrated in FIG. 1, the terminal described herein may be implemented in various forms. For example, mobile phones, smart phones, laptop computers, digital broadcasting terminals, tablet PCs, PDAs (Personal Digital Assistants), PMPs (Portable Multimedia Players), navigation, MP3 players, etc. In addition to the mobile terminal of, a fixed terminal such as a smart TV (Smart TV), a set-top box, a desktop computer, etc. can be used, and any terminal that can send and receive messages with other users over a network, The device is also applicable to the terminal of the present invention.

The web server 20 collectively refers to servers of a service provider that provides various services on the web, and is not limited to a specific server or a server that provides a specific service. In particular, in the present invention, the web server 20 provides the above-described item and various information about the item to the terminal 10 connected to the web server 20, the terminal 10 (that is, the user) The recommended item list for the selected item may be received from the service device 30 and provided to the terminal 10. In addition, the web server 20 may perform a function of transmitting user log data to the service device 30, which will be described later.

The service device 30 refers to a device that performs a function for providing the web server 20 with a list of recommended items for various items provided by the web server 20. In particular, in the present invention, the service device 30 collects and processes user log data from the terminal 10 and / or the web server 20 and calculates a list of recommended items for each item to the web server 20. Can provide. Here, the user log data refers to a set of information indicating a history such as which terminal 10 (ie, user) performed what operation on which item on which web server and when.

In addition, the service device 30 collects view log data and / or click log data from the terminal 10 and / or the web server 20 or from the collected user log data. The reprocessing may provide a list of recommended items in which feedback is reflected according to a user's visit or operation. Here, the view log data refers to a set of information indicating a history of providing a web page displaying a specific item to the terminal 10 by connecting the web server 20 of the terminal 10 (that is, the user). do. The click log data is included in the recommendation item list by the terminal 10 (that is, the user) by clicking on the recommendation item list provided to the web server 20 by the service device 30. A web page displaying a recommendation item clicked through a link (for example, a hyperlink or a hot link, etc.) means a set of information representing a history provided to the terminal 10.

The service device 30 may be implemented as a set of various devices to support the above-described operation. For example, a storage server capable of storing and managing log data (eg, user logs, view logs, click log data), a collection server that collects log data, a clustering server that converts or processes log data, It may be implemented including an API (Application Program Interface) server for calculating a list of recommended items and providing a list of recommended items to a web server, a feedback processing server for modifying a list of recommended items based on a feedback result, or selecting a recommendation algorithm. . In addition, each module performing a function corresponding to the above-described operation may be implemented in one service device 30. Hereinafter, it will be described on the assumption that it is implemented by one service device 30 for convenience of description.

The terminal 10, the web server 20, and the service device 30 transmit and receive various related information through the network 40. In this case, the network 40 may use various types of networks. For example, 3G mobile communication, 4G mobile communication, WLAN (Wireless LAN), Wi-Fi, Wibro, Wimax, High Speed Downlink Packet Access (HSDPA), or Ethernet Wired communication methods such as Ethernet, xDSL (ADSL, VDSL), Hybrid Fiber Coaxial Cable (HFC), Fiber to The Curb (FTTC), and Fiber To The Home (FTTH) can be used. It may be. Further, in addition to the above-described communication method, it may include all other types of communication methods that are well known or developed in the future.

2 is a diagram illustrating a configuration of a service device according to an embodiment of the present invention.

The configuration of the service device 30 illustrated in FIG. 2 illustrates functionally divided functional elements, and any one or more components may be implemented by being physically integrated with each other.

2, the service device 30 according to an embodiment of the present invention may include a communication unit 210, a control unit 220, and a storage unit 230. The communication unit 210, the control unit 220, and the storage unit 230 constituting the service device 30 may be functionally connected to each other to perform a function according to the present invention.

In more detail with respect to each component, the communication unit 210 may include one or more modules that enable wired / wireless communication with one or more terminals 10 and / or the web server 20. In particular, in the present invention, the communication unit 210 may receive the user log data from the one or more terminals 10 and / or the web server 20. In addition, the communication unit 210 may receive a recommendation item list request message from the web server 20, and transmit the recommendation item list to the web server 20 in response to the recommendation item list request message. In addition, when a connection request is received from the terminal 10 by clicking on a user's recommendation item on a web page provided by the web server 20, a redirect URL (Reform) Uniform Resource Locator (URL) is provided to the corresponding terminal 10 requesting the connection. Can transmit

The controller 220 is configured to perform overall control of the service device 30, and controls the flow of signals for performing the functions of the communication unit 210 and the storage unit 250. In this case, the controller 220 may include an operating system (OS), an application program, and a process device for driving each component, for example, a central processing unit (CPU).

In particular, in the present invention, the controller 220 may include a data collector 221, a data processor 222, a recommendation calculator 223, a recommendation result provider 224, and a feedback processor 225. Can be.

The data collector 221 may perform a function of collecting the above-described user log data from the terminal 10 and / or the web server 20 and storing the user log data in the storage 230.

The data processor 222 extracts data necessary for calculation in order to calculate the recommended item from the user log data collected and stored by the data collector 221, and converts or processes the extracted data into a data format suitable for calculation. The stored user log data may be stored in the storage 230. In addition, a function of converting or processing the recommendation calculation result calculated and stored by the recommendation calculation unit 223 into a format for providing the web server 20 through the Internet to the database 230 may be stored in the database 230. Can be.

The recommendation calculator 223 calculates a recommendation item using various recommendation algorithms from the converted user log data processed by the data processor 222, and stores the recommendation calculation result of calculating the recommendation item in the storage 230. To perform the function.

When the recommendation result providing unit 224 receives the recommendation item list request message for the specific item from the web server 20, the recommendation item list extraction unit 224 extracts the recommendation item list for the corresponding item from the recommendation result of the storage unit 230 and the web server 20. ) Can be sent. Also, remove items from the recommendation result (or recommendation item list), add new recommendation items to the recommendation result (or recommendation item list), add additional information to the recommendation result (or recommendation item list), etc. Postprocessing the recommendation result (or recommendation item list) may be performed.

The feedback processor 225 collects click log data and / or view log data, reprocesses the recommendation result (or recommendation item list) based on the collected click log data and / or view log data, or recommends the calculator 223. ) May determine a recommendation algorithm used for recommendation calculation.

First, in order to collect click log data, the feedback processing unit 225 receives a connection request from the terminal 10 by clicking on a user's recommendation item on a web page provided by the web server 20, and uses the click log data as the click log data. It performs a function of storing in the storage unit 230, it may perform a function of transmitting a redirect (redirect) URL to the corresponding terminal 10, the connection request. In addition, the feedback processing unit 225 may perform a function of extracting data having a specific parameter or format from the user log data and storing the data in the storage unit 230 as click log data.

In addition, in order to collect the view log data, the feedback processor 225 may receive a recommendation item list request message from the web server 20, and store the same in the database 230 as view log data. .

Meanwhile, in the example of FIG. 2, the data collector 221, the data processor 222, the recommendation calculator 223, the recommendation result provider 224, and the feedback processor 225 are illustrated as respective blocks. It may be integrated into one device and implemented, or as described above, some or each of them may be implemented in physically different devices. In addition, a plurality of devices that perform the same function may be implemented in parallel.

The storage unit 230 is a means for storing data and programs necessary for the operation of the service device 30, and basically stores an operating program and an application program to be executed by the service device 30. In particular, in the present invention, the storage unit 230 is the user log data collected by the data collection unit 241, the converted user log data processed by the data processing unit 222 and the recommendation results, recommendation calculation unit 223 The recommendation calculation result calculated by) and the view log data and the click log data collected by the feedback collector 245 may be stored.

Such a storage unit 250 may be implemented by various storage means. For example, the storage unit 250 may include a flash memory type, a hard disk type, a multimedia card micro type, a card type memory (for example, SD or XD). Memory (RAM), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EPEROM), programmable read-only memory (PROM) At least one type of storage medium may include a magnetic memory, a magnetic disk, and an optical disk. The storage unit 250 may be inside or outside the controller 220 and may be connected to the controller 220 by various well-known means.

Hereinafter, for clarity of explanation, the user log data processed by the data processing unit 222 is referred to as 'converted user log data' separately from the 'user log data' collected by the data collection unit 221. In addition, the result calculated by the recommendation calculation unit 223 is referred to as a 'recommended calculation result', and the recommendation calculation result processed by the data processing unit 222 is referred to as a 'recommended result'. Also, the recommendation result provided by the recommendation result providing unit 224 about the specific item requested by the web server 20 in the recommendation result is referred to as a 'recommended item list'. That is, the recommendation result encompasses a list of recommended items recommended for all items.

Hereinafter, an operation for providing a recommendation item in the service device 30 will be described with reference to FIG. 3.

3 is a diagram illustrating a method for providing a recommended item according to an embodiment of the present invention.

Referring to FIG. 3, the service device 30 may collect the above-described user log data from the terminal 10 and / or the web server 20 (S301). Here, the service device 30 may collect user log data from the terminal 10 and / or the web server 20 synchronously (ie, real time collection) or asynchronously (eg, collecting at a specific period). The user log data may be continuously collected from the terminal 10 and / or the web server 20 regardless of the collection method.

The service device 30 may extract data necessary for the recommendation calculation in order to calculate the recommendation item from the user log data collected in operation S301, and convert the extracted data into a data format suitable for the recommendation calculation (S303). Here, if the format of the user log data collected from the plurality of terminals 10 and / or the plurality of web servers 20 is the same in step S301 and is suitable for calculating the recommended items, the service device 30 performs the data conversion step. It may not be performed, that is, step S303 may be omitted.

The service device 30 may calculate a recommendation item using various recommendation algorithms using the user log data converted in S303 and store the recommendation calculation result in operation S305. Here, the service device 30 may calculate a recommendation item for each recommendation algorithm for all items. In addition, the user log data may be arbitrarily divided into a plurality of groups according to a specific criterion, and a recommendation item may be calculated using a different recommendation algorithm for each group. In addition, the time point at which the service device 30 calculates the recommended item may or may not be the same as the time point at which the user log data is collected in step S301, and may be calculated with a certain period.

The service device 30 may convert the recommendation calculation result calculated in step S305 into a suitable data format in order to provide the web server 20 through the Internet (S307). As described above, if the service device 30 does not perform the data conversion step in step S303, step S307 may also be omitted.

The service device 30 may perform post-processing on the calculated recommendation result (or recommendation item list), and then provide the recommendation item list to the corresponding web server 20 collecting user log data (S309). In this case, the service device 30 may provide the recommendation item list to the web server 20 at regular intervals, or the service device 30 may request a recommendation item list for a specific item from the web server 20 to the corresponding item. The recommended item list may be transmitted to the web server 20.

The service device 30 may collect the user's feedback and reflect it in the recommendation (S311). That is, the service device 30 reprocesses the recommendation result (or recommendation item list) based on the collected click log data and / or view log data (branch to step S309), or the recommendation calculation unit 223 calculates the recommendation. A recommendation algorithm to be used can be determined (branch to step S305) (S311).

Hereinafter, each step described in the example of FIG. 3 will be described in more detail.

The service device 30 (particularly, the data collection unit 221) may collect the above-described user log data from the terminal 10 and / or the web server 20 (S301).

As described above, the user log data refers to a set of information indicating which user performed what action on which item on which web server and when. In addition, the user log data includes information for identifying a browser mounted on the terminal 10 used by the user, information for limiting exposure to a specific user for other items, and a member ID of the user managed by the web server 20. (ID: Identifier) information and the like.

Table 1 illustrates information that may be included in user log data according to the present invention.

Table 1

User Identification Information	Product identification	Motion identification	Motion occurrence time
User A	Goods 2	shopping basket	20XX-YY-ZZ aa: bb: cc
User B	Goods	3	purchase	20XX-YY-ZZ aa: bb: cc
User B	Item	5	visit	20XX-YY-ZZ aa: bb: cc

The service device 30 may use the following method for collecting such user log data.

First, the service device 30 may access the web server 20 to extract and obtain user log data, or may receive and obtain user log data transmitted from the web server 20. It demonstrates with reference.

4 is a diagram illustrating a user log data collection method according to an embodiment of the present invention.

Referring to FIG. 4, when the terminal 10 is connected and performs a specific operation, the web server 20 access path information of the web page including the user log data to inform the service device 30 that the specific operation has been performed. A connection request message including a may be transmitted to the service device 30 (S401). Here, the web server 20 may transmit a connection request message to the service device 30 using the POST HTTP protocol.

When the service device 30 receives the access request message from the web server 20, the service device 30 accesses the web server 20 based on the access path information of the web page included in the access request message and logs the user included in the web page. The data is extracted and obtained (S403). Here, the service device 30 may extract user log data from the web server 20 using the GET HTTP protocol.

Subsequently, the service device 30 stores the obtained user log data (S405).

In addition, unlike the example of FIG. 4, the user log data on the operation of the user terminal 10 connected to the web server 20 from the web server 20 is transferred to the service device 30 through the TCP / IP protocol. By transmitting, the user log data may be obtained from the service device 30.

Next, the service device 30 may insert a script into a web site serviced by the web server 20 and obtain user log data from the user terminal 10 through the inserted script. A description with reference to FIG. 5 below.

5 is a diagram illustrating a user log data collection method according to an embodiment of the present invention.

Referring to FIG. 5, the service device 30 may insert a script into a web site provided by the web server 20 (S501). For example, the service device 30 inserts a script such as "<script src = http: //collector.recopick.com/plugin.js> </ script>" into a web page provided by the web server 20. can do.

Thereafter, when the terminal 10 accesses the web server 20 (S503) and visits a web page provided by the web server 20, the terminal 10 executes a script inserted in the web page (S505). The log data of the user may be automatically transmitted to the service device 30 by the script (S507). For example, if the above-described script is executed in a web browser mounted on the terminal 10, the terminal 10 provides information on URLs, visit times, and values of parameters included in the script. To the device 30.

Information collected by the service device 30 from the terminal 10 by using the above-described script is as follows.

-User IDentifier (UID) generated based on web browser cookie

-Visit URL information of the user

-The user's referrer information

Connection time information

-User Agent Information of Web Browser

-Other Meta Information

The user identifier (UID) means a value for distinguishing a user. When the terminal 10 executes the script, the terminal 10 first finds whether a user identifier value is stored in a cookie of a web browser, and if the user identifier is stored, the value is a service device. Transfer to 30. On the other hand, if the user identifier is not stored in the cookie of the web browser, a random numeric value and a creation time are combined (for example, 33619442.1366005959274) to generate information that can identify the user, and transmit the generated information to the service device 30. .

Visit URL information means URL information on the web page of the web server 20 to which the access. And, the reference address (referrer) information means the URL information about the previously visited web page when visiting the web page of the web server 20 through a link. The access time information means time information that the terminal 10 visits a web page of the web server 20. The user agent information of the web browser refers to information for distinguishing the web browser used by the terminal 10 when accessing the web server 20. For example, the information for identifying whether the web browser used in the terminal 10 is a web browser for a mobile device or a web browser for a desktop may correspond thereto. Such information may be collected by a web browser mounted on the terminal 10 to transmit the entire information synchronously or asynchronously to the service device 30.

The meta information means other additional information to be considered when the service device 30 calculates a recommendation item. For example, the meta information may identify a user more accurately, such as member information, on a web site operated by a web server. Information or information on whether an item requires adult certification may be applicable. Such meta information is transmitted to the service device 30 by the terminal 10 when the script embedded in the web page of the web server 20 has a format for requesting meta information. . For example, when the inserted script has a form of "plugin.js? Key = value & key2 = value2", the terminal 10 transmits values (value1, value2) for each parameter (key, key2) to the service device 30. To be sent).

The service device 30 that receives the user log data from the terminal 10 in the manner described above stores the obtained user log data (S509).

If the user log data collected through the same method as the example of Figs. 4 to 5 are illustrated in Table 2 below.

Table 2 illustrates user log data collected by the service device 30.

TABLE 2

KEY	VALUE
ref	http://vanillashu.com/front/php/product.php?product_no=13806&main_cate_no=1
site	vanillashu.co.kr
uid	44209103.136609
url	http://vanillashu.co.kr/front/php/category.php?cate_no=30
time	20XX-YY-ZZ aa: bb: cc

The service device 30 (particularly, the data processing unit 222) extracts data necessary for the recommendation calculation in order to calculate the recommendation item from the user log data collected in step S301, and recommends the extracted data for the calculation. The data format may be transformed and stored (S303).

First, the service device 30 extracts the service identifier SERVICE_ID from the collected user log data by using the domain of the visited URL information. For example, when the visit URL information is "http://www.foodfly.co.kr/restaurants/show/11378", the service device 30 may check the domain "www.foodfly.co. As a service identifier corresponding to "kr", 1 may be assigned (SERVICE_ID = 1). At this time, the domain and service identifier of the web site operated by each web server 20 may be mapped one-to-one. In addition, the table information in which the domain of the web site and the service identifier are mapped may be stored in the service device 30. In this case, the service device 30 may extract the service identifier value by referring to the previously stored table. have.

In addition, the service device 30 extracts an item identifier (ITEM_ID) and an action identifier (ACTION_ID) from the visited URL information by using a regular expression predefined for each service identifier (SERVICE_ID). Here, the regular expression means a formula for extracting a value of a required parameter from URL information, and a different regular expression may be used for each service identifier and / or operation identifier. For example, "/restaurants\/show\/(.*)$/" is used as a regular expression for extracting a 'VISIT' operation from a URL having the service identifier 1 (SERVICE_ID = 1). In this case, the path '/ restaurants \ / show \ /' indicates a 'VISIT' operation (ACTION_ID = VISIT), and the parameter value 11378 located after the path indicates an item identifier (ITEM_ID = 11378). ). Here, \ is a variable indicating that '/' exists to distinguish each path when a plurality of paths are included in the regular expression, and $ is a variable indicating that the regular expression is finished. In addition, "/order\/(.*) as a" PURCHASE "regular expression at a URL such as" http://www.foodfly.com/order/11378 "with the same service identifier (SERVICE_ID = 1). $ / "May be used, in which case the '/ order \ /' path indicates a 'PURCHASE' action (action_id = purchase), and the parameter value 11378, located after the path, indicates the item identifier ( item_id = 11378).

Meanwhile, if the user member ID value of the web server 20 is transmitted and stored as meta information in the user log data collection step (S301), the service device 30 uses the corresponding member ID as the user identifier (USER_ID), If not, the terminal 10 uses the user identifier USER_IDENTIFIER generated based on the web browser cookie as the user identifier USER_ID.

In addition, the service device 30 converts the user identifier USER_ID and the item identifier ITEM_ID extracted or acquired through the above-described method into a predefined numeric identifier, and the numeric identifier predefined in the extracted operation identifier ACTION_ID. To give. In this case, the service device 30 may convert each user identifier, an operation identifier, and an item identifier into a predetermined numeric identifier, such as a one-to-one mapping relationship between a domain and a service identifier, or optionally convert to a numeric identifier having a continuous integer value ( Or grant).

Table 3 illustrates the converted user identifier table.

TABLE 3

Service identifier (SERVICE_ID)	Original user identifier (ORG_USER_ID)	Converted User Identifier (CONVERTED_USER_ID)
One	chaehyun	One
One	44209103.136609	2
One	68709103.015529	3

In Table 3, the original identifier 'chaehyun' indicates a case where the user member information value of the web server 20 is delivered and used as the user identifier, and '44209103.136609' or '68709103.015529' is based on the web browser cookie of the terminal 10. This is the case that the generated value is used as a user identifier.

Table 4 illustrates the converted item identifier table.

Table 4

Service identifier (SERVICE_ID)	Original item identifier (ORG_ITEM_ID)	Converted Item Identifier (CONVERTED_ITEM_ID)
One	11378	One
One	1245	2
One	21456	3

Table 5 illustrates the translated action identifier table.

Table 5

Action Name (ACTION_NAME)	Action Identifier (ACTION_ID)
VISIT	One
PURCHASE	101
Recommended algorithm 2 (RECOMMEND_VISIT_ALGORITHM_2)	2
Recommended algorithm 3 (RECOMMEND_VISIT_ALGORITHM_3)	3
BASKET	201

In Table 5, the recommendation algorithm x (RECOMMEND_VISIT_ALGORITHM_x) represents an operation in which the user terminal 10 accesses the web server 20 having the corresponding service identifier through a recommendation item list calculation result calculated using the recommendation algorithm x. .

The user log data finally converted as described above is shown in Table 6 below.

Table 6 illustrates the converted user log data.

Table 6

Version (VERSION)	Converted User Identifier (CONVERTED_USER_ID)	Converted Item Identifier (CONVERTED_ITEM_ID)	Service identifier (SERVICE_ID)	Action Identifier (ACTION_ID)	TIME
0	36368996	1404200	4	One	20XX-YY-ZZ aa: bb: cc
0	36346053	39861	4	One	20XX-YY-ZZ aa: bb: cc
0	4093797	46666	8	One	20XX-YY-ZZ aa: bb: cc
0	36352376	29640	4	One	20XX-YY-ZZ aa: bb: cc
0	4093800	97415	8	One	20XX-YY-ZZ aa: bb: cc
0	36344208	39160	4	One	20XX-YY-ZZ aa: bb: cc
0	36343108	3812418	4	One	20XX-YY-ZZ aa: bb: cc
0	36356140	3841659	4	One	20XX-YY-ZZ aa: bb: cc
0	36395318	3856063	4	One	20XX-YY-ZZ aa: bb: cc

Referring to Table 6, the service device 30 may assign different version information to distinguish the converted user log data according to the items of the information included in the converted user log data. For example, in the case of the converted user log data having version 0, as shown in the example of Table 6, the converted user identifier, the converted item identifier, the service identifier, the operation identifier, the time item, but the conversion having the version 1 In the case of the user log data, information items other than those included or not included in version 0 may be added.

The service device 30 (particularly, the recommendation calculator 223) may calculate the recommendation item using various recommendation algorithms using the user log data converted in S303, and store the recommendation calculation result (S305).

The service device 30 may calculate a recommendation item by applying various recommendation algorithms so as to apply a recommendation algorithm suitable for the nature of the service provided by the web server 20 providing the recommendation result. In other words, based on the collected total user log data collected by applying a variety of recommendation algorithms to calculate the recommended items according to each recommendation algorithm, or group the collected total converted user log data arbitrarily or according to a specific condition, Each group may calculate recommendation items according to different recommendation algorithms. For example, the converted user log data having an odd user identifier (USER_ID) and the converted user log data having an even number may be distinguished, and a recommended item may be calculated using an A recommendation algorithm and a B recommendation algorithm, respectively.

The service device 30 calculates similarity with all items except for the corresponding items for all items belonging to the converted user log data included in the predetermined range (ie, all or groups). For example, if a total of 100 items belong to the converted user log data within a predetermined range, the similarity between the first item and 99 items other than the corresponding item is calculated, and the 99 items other than the second item and the corresponding item are calculated. The similarity with other items is calculated for all items by calculating the similarity between them.

The service device 30 may use various recommendation algorithms to calculate the similarity between two items. For example, jacquard similarity, log likelihood similarity, Pearson correlation, or the like may be used. These similarity calculation methods are already known methods and will not be described in detail below. Hereinafter, description will be made on the assumption that jacquard similarity is used as a recommendation algorithm for convenience of description, and jacquard similarity is defined as in Equation 1 below.

Equation 1

Referring to Equation 1, the jacquard similarity between two items is determined by the number of users who visited both items (the number of users who accessed both web pages displaying both items) and the number of unions of users who visited both items (the two items The number of users who have access to any one or more of the displayed web pages is calculated by dividing the number of users who have accessed both web pages displaying two items), and as a result, the more users who visit both items, the greater the similarity between the two items. . For example, the number of users who access the web page displaying item A is 20, the number of users who access the web page displaying item B is 10, and the number of users who visited both the web pages displaying item A and B If the number is 5, the result of calculating the jacquard similarity is equal to 5 / (20 + 10-5) = 1/5.

The service device 30 calculates all similarities for all items belonging to the user log data converted within a predetermined range (all or in a group) using the recommendation algorithm, and then recommends n items in the order of similarity for each item. Calculate it as an item. In this case, the service device 30 calculates all similarities for all items, extracts an item satisfying a specific condition (for example, the number of users who visited both items> = 50 and the similarity> 0.1) and then extracts the items. N items may be calculated as recommended items in order of similarity for each item among the items.

The method of calculating the recommendation item using the jacquard similarity described above is just one example, and as described above, the service device 30 may calculate the recommendation item for each item using various recommendation algorithms. In addition, even when the recommendation scores of the recommended items for each item are calculated using different recommendation algorithms, the recommendation scores may all be set at the same level (eg, between 0 and 1). As described above, the result of recommendation calculation according to each recommendation algorithm is shown in Table 7 below.

Table 7 illustrates the recommendation calculation results calculated according to each recommendation algorithm.

TABLE 7

Converted Item Identifier (CONVERTED_ITEM_ID)	Translated suggestion item identifier (RECOMMENDED_ITEM_ID)	Referral Score (SCORE)	Method (METHOD)
One	11	0.05511811023622047	2
One	164	0.0894854586129754	2
One	218	0.04025423728813559	2
One	272	0.11278195488721804	2

Referring to Table 7, the SCORE value indicates a value for calculating the similarity between the item and the recommended item according to the recommendation algorithm, and the method METHOD indicates an identification number for the recommendation algorithm used for the similarity calculation.

The service device 30 (particularly, the data processor 222) may convert the recommendation calculation result calculated in step S305 into a suitable data format in order to provide the web server 20 (S307).

First, the service device 30 changes the identifier of the item converted from the recommendation calculation result to the original identifier, and adds and stores the service identifier and the calculated date.

Table 8 illustrates the converted format of the recommendation calculation result.

Table 8

Original item identifier (ORG_ITEM_ID)	Original suggested item identifier (ORG_RECOMMENDED_ITEM_ID)	Referral Score (SCORE)	Method (METHOD)	Service identifier (SERVICE_ID)	DATE
123	11149	0.055118	2	One	20XX-YY-ZZ
123	11113	0.089485	2	One	20XX-YY-ZZ
123	740	0.040254	2	One	20XX-YY-ZZ
123	179	0.112782	2	One	20XX-YY-ZZ

Subsequently, the service device 30 may convert and store the recommendation calculation result into a data format suitable for finally providing the web server 20 through the Internet. At this time, the service device 30 may be stored in the JSON (JavaScript Object Notation) format as shown in Table 9 below. However, the JSON format is just one example, and the service device 30 may store the data in various data formats such as Extensible Markup Language (XML) format or YAML Ain't Markup Language (YAML) format. have. The following description assumes the JSON format.

Table 9 illustrates the recommended result format that is finally converted and stored.

Table 9

HASH	Service identifier (SERVICE_ID)	Original item identifier (ORG_ITEM_ID)	COUNT	JSON	TIME
4cfc30c5e1c1bbcf	One	11057	3	[{"id": "11649", "score": 0.059867, "method": "2"}, {"id": "11480", "score": 0.052219, "method": "2"}, { "id": "12116", "score": 0.052083, "method": "2"}]	20XX-YY-ZZ aa: bb: cc
c000365f48786906	One	11078	2	[{"id": "11822", "score": 0.055385, "method": "2"}, {"id": "11021", "score": 0.033816, "method": "2"}]	20XX-YY-ZZ aa: bb: cc
96028a9562ba33de	One	11099	10	[{"id": "325", "score": 0.100656, "method": "2"}, {"id": "553", "score": 0.061425, "method": "2"}, { "id": "849", "score": 0.055133, "method": "2"}, {"id": "622", "score": 0.051887, "method": "2"}, {"id ":" 464 "," score ": 0.051515," method ":" 2 "}, {" id ":" 435 "," score ": 0.050532," method ":" 2 "}, {" id ": "11614", "score": 0.046632, "method": "2"}, {"id": "786", "score": 0.044872, "method": "2"}, {"id": "11456 "," score ": 0.041322," method ":" 2 "}, {" id ":" 11771 "," score ": 0.040925," method ":" 2 "}]	20XX-YY-ZZ aa: bb: cc

Referring to Table 9, the hash value means a value calculated so as not to store the recommended item list for the corresponding item when the recommended item list recommended for the specific item is the same, and the service identifier (SERVICE_ID) and the item. The identifier (ITEM_ID) and the recommended item identifier (RECOMMENDED_ITEM_ID) are combined into one string, and then calculated by applying a hash function to the combined string. By using such a hash value, even if the recommendation item of the recommendation item list for a specific item is changed or the recommendation item of the recommendation item list is not changed unless the recommendation item of the recommendation item list is recommended, the recommendation is changed. The item list may not be updated. That is, since the recommendation score for the recommendation item is not included as a variable for calculating the hash value, when the recommendation item list is calculated the next day, even if the recommendation score of the recommendation item is changed, the number of recommended items is changed, If the order does not change, the hash value does not change.

The count value indicates the number of recommended items and how many items are recommended for a particular item. The JSON value includes information on the number of recommended items corresponding to the count value. For example, the JSON value may include a recommendation item identifier, a recommendation score, and recommendation algorithm information.

The service device 30 (particularly, the recommendation result providing unit 224) performs post-processing on the calculated recommendation result (or recommendation item list), and then recommends to the corresponding web server 20 that has collected user log data. An item list may be provided (S309).

This will be described in more detail with reference to FIG. 6 below.

6 is a diagram illustrating a method of providing a recommendation item list according to an embodiment of the present invention.

Referring to FIG. 6, the service device 30 may perform post-processing on the converted stored recommendation result (or recommendation item list) (S601).

Post-processing of the recommendation result (or recommendation item list) refers to an operation of adding, deleting, or changing a recommendation item from the recommendation result (or recommendation item list) or adding, deleting, or changing information items included in the recommendation item list. . This post-processing process may be automatically performed in the service device 30, or may be performed in the service device 30 by a request from the web server 20.

In detail, the service device 30 may remove the prohibited item from the recommendation result (or the recommended item list). For example, when the administrator of the web server 20 prohibits the recommendation of item B for item A, item B may be removed from the recommended item list for item A. FIG. In addition, in the case of an item that should not be recommended for any item, such as when a certain product is out of stock or when a particular store is no longer in operation, the item may be removed from the list of all recommended items (ie, the recommendation result). When removing the recommended prohibited item from the recommendation result (or recommendation item list), the service device 30 arbitrarily deletes an item to be deleted from the previously generated recommendation result (or recommendation item list), or The recommendation score can be changed to a lower value or a variable value can be adjusted so that a lower value is calculated to remove the prohibited item from the recommendation result (or recommendation item list).

In addition, the service device 30 may add any item that does not belong to the recommendation result (or the recommendation item list) as the recommendation item. For example, if the administrator of the web server 20 does not like the result of the calculated recommendation item list, or if the recommendation item list for the specific item is not generated, the service device 30 selects the specific item. It may be included in the recommendation result (or recommendation item list) as a recommendation item. Even in this case, the service device 30 arbitrarily adds an item to be added from a previously generated recommendation result (or a recommendation item list), or creates a recommendation item list composed of specific items or recommends an item to be added. You can adjust the value of the variable to change the score to a higher value or to calculate a higher value.

In addition, the information item constituting the recommendation result (or recommendation item list) may include only the recommendation item identifier, recommendation score, and recommendation algorithm information as shown in the example of Table 9 above, but the service device 30 may also recommend the recommendation result (or It is also possible to add a new information item constituting a list of recommended items. For example, the service device 30 may include a recommended item name, a URL of a thumbnail of the recommended item, price information of the recommended item, and the like in the recommended item list.

The web server 20 transmits a message to the service device 30 for requesting a list of recommended items regarding items displayed on the web page to which the terminal 10 accesses (S603). The recommended item list request message includes a service identifier for the requesting web server 20, an item identifier for the requested item, and may further include version information, meta information, and the like.

Upon receiving the recommendation item list request message from the web server 20, the service device 30 extracts a recommendation item list for the requested item (S605), and transmits the extracted recommendation item list to the web server 20 (S607). .

Here, the service device 30 may transmit the recommendation item list in the format of the recommendation item list generated in the web server 20 (ie, JSON format), or may transmit the recommendation item list in the iframe format. In the case of providing a list of recommended items in JSON format, after adding a separate design in the web server 20, a list of recommended items should be provided on the web page, whereas in the case of an iframe, the web server 20 does not need additional processing. There is an advantage that can be applied directly to a web page.

First, for example, in the case of transmitting a list of recommended items in the JSON format, the web server 20 is a service device in a format such as "http://api.recopick.com/VERSION/recommend/SERVICE_ID/ITEM_ID?OPTIONS". The list of recommended items may be requested at 30. As such, when requesting a recommendation item list in JSON format, the recommendation item list request message may include a service identifier, an item identifier, version information, and other meta information.

For example, when the web server 20 requests a list of recommended items from the service device 30 such as "http://api.recopick.com/2/recommend/1/11057?field=meta", the request is made. The version of the recommended item list is 2, the service identifier is 1, the identifier of the requesting item is 11057, and it indicates a case of additional request for meta information. An example of the recommended item list provided by the service device 30 to the web server 20 in response to the request is shown in Table 10 below.

Table 10 illustrates a list of recommended items provided in JSON format.

Table 10

[{"id": "11649", "score": 0.061947, "method": "2", "image": {"src": "http://www.foodfly.com/upload/restaurant/ 8320977315090c487b5f3d.jpg "}," title ":" XXX "}, {" id ":" 11480 "," score ": 0.05137," method ":" 2 "," image ": {" src ":" http: //www.foodfly.co/kr/upload/restaurant/1192962858506d36dc8981f.jpg"},"title":"YYY},{"id":"12116","score":0.042345,"method":"2 " , "image": {"src": "http://www.foodfly.com/upload/restaurant/1138839503513839a99ea0e.jpg"}, "title": "ZZZ"}]

Referring to Table 10, the recommendation item list may include a recommendation item identifier (“id”), a recommendation score (“score”), and recommendation algorithm information (“method”). In addition, Table 10 illustrates a case in which the service device 30 includes additional meta information in the recommended item list when additional meta information is additionally requested, and image information (“image”) of the recommended item is included as additional meta information. ), And recommended item name information ("title").

Next, for example, in the case of transmitting a list of recommended items in iframe format, the web server 20 reads "http://api.recopick.com/VERSION/banner/SERVICE_ID/show?item_id=ITEM_ID&direction= [horizontal | The recommended item list may be requested from the service device 30 in a format such as vertical & count = recommend_size ". As such, when requesting a list of recommended items in an iframe format, a service identifier, an item identifier, a frame direction (horizontal or vertical direction), a frame size, version information along with the number of recommended items, and the like may be included.

For example, the web server 20 requests a list of recommended items from the service device 30 such as "http://api.recopick.com/2/banner/6/show?item_id=59&direction=horizontal&count=4". In this case, the version of the requested recommendation item list is 2, the service identifier is 6, the identifier of the requesting item is 59, the horizontal direction in the direction of the frame, and the number of recommended items is 4.

In addition, for example, the web server 20 may recommend a list of recommended items to the service device 30 such as "http://api.recopick.com/2/banner/6/show?item_id=59&direction=vertical&count=2". In the case of requesting, the version of the requested recommendation item list is 2, the service identifier is 6, the identifier of the requesting item is 59, the direction perpendicular to the frame direction, and the number of recommended items is 2.

8 illustrates a list of recommended items provided by the service device 30 to the web server 20 in response to the request.

7 is a diagram illustrating a list of recommended items provided in an iframe format according to an embodiment of the present invention.

Referring to FIG. 7A, when the web server 20 requests four recommended items in the horizontal direction, in response to the request, the recommended items displayed in the horizontal direction are four recommended items provided by the service device 30. In FIG. 7B, when the web server 20 requests two recommended items in the vertical direction, the number of recommended items provided by the service device 30 is two and vertical in response thereto. An example of the recommended item list displayed in the direction is illustrated.

8 illustrates an example of displaying a recommendation item list provided in an iframe format on a web page according to an embodiment of the present invention.

Referring to FIG. 8, the recommendation item list in the iframe format provided by the service device 30 may be displayed on the web page of the web server 20 in the form provided by the service device 30 as shown in FIG. 7. Can be. Thus, since the service device 30 provides the list of recommended items to the web server 20 in the form of a banner, the web server 20 may not perform an additional data processing procedure.

Meanwhile, although FIG. 6 illustrates a method of transmitting a recommendation item list at the request of the web server 30 after the service device 30 performs post-processing, the service device 30 from the web server 20. When the recommended item list request message is received, the post-processed recommendation item list may be transmitted to the web server 20 after the post-processing is performed on the requested recommendation item list. That is, the step of performing post-processing may be performed between steps S605 and S607.

The service device 30 (particularly, the feedback processor 225) may collect the user's feedback and reflect it in the recommendation (S311).

The feedback of the user may be calculated according to the click rate of the user, and the recommended view log data and the user's recommended click log data may be collected to calculate the click rate.

First, how to collect click log data.

As a first method, when a recommendation item included in the recommendation item list of the user is clicked, it is redirected to the service device 30 to obtain click log data of the user, which will be described with reference to FIG. 9.

9 is a diagram illustrating a method of collecting click log data according to an embodiment of the present invention.

Referring to FIG. 9, when the terminal 10 selects and inputs a specific recommendation item included in the recommendation item list displayed on the web page of the web server 20 (for example, click or touch input) (S901). In operation S903, the terminal 10 requests access to the service device 30 through the link URL of the recommended item. That is, the recommendation item may have URL information for moving to the service device 30 instead of URL information for moving to a web page displaying the corresponding recommendation item.

For example, each recommendation item (s) belonging to the list of recommended items may have a link URL in the format "http://api.recopick.com/VERSION/banner/SERVICE_ID/pick?source=SOURCE_ITEM_ID&pick=TARGET_ITEM_ID&method=ALGORITHM". Can have The link URL of each recommendation item includes the version information of the recommendation item list to which the recommendation item belongs, the service identifier, the source item (the item to which the recommendation item is displayed), the target item (the recommendation item selected and clicked by the user). ) Identifier, algorithm information for calculating the selected recommendation item, and the like.

For example, if items 132, 178, and 200 are recommended in item 59 of the web server 20 having the service identifier SERVICE_ID 6, and the user clicks item recommendation 132 recommended by the algorithm 2, the user The terminal 10 moves to the service device 30 through a URL such as "http://api.recopick.com/1/banner/6/pick?source=59&pick=132&method=2".

The service device 30 stores click log data for the terminal 10 attempting to access the service device 30 (S905). As described above, the recommendation item link URL may include version information of the recommendation item list to which the corresponding recommendation item belongs, a service identifier, a source item identifier, a target item identifier, algorithm information for calculating the selected recommendation item, and the like. ) Can determine which recommendation item is clicked by the user from which item on which web server 20 through the URL that the terminal 10 attempts to access, and can confirm that the recommendation item is calculated by which recommendation algorithm. have.

The service device 30 transmits the redirect URL to the terminal 10 so that the terminal 10 can reconnect to the web server 20 (S907). For example, the service device 30 may terminal the URL such as "http://hellonature.net/goods/132?recopick=2" so that the terminal 10 may access a web page displaying item 132 selected by the terminal 10. 10 can be transmitted.

Upon receiving this, the terminal 10 accesses the web server 20 through the received redirect URL (S909).

Since the user's click information is transmitted to the service device 30 whenever the user clicks on the recommendation item by using the method as shown in the example of FIG. 9, the service device 30 stores the user's click log data on the recommendation item in real time. Can be stored.

In a second method, a predetermined specific parameter is added to a variable constituting a link of the recommendation item, and if the predetermined specific parameter described above exists in the visit URL information of the user in the user log data collected in step S301, it is added to the recommendation. Can be judged by click. That is, the data including the predetermined parameter in the visited URL may be extracted from the user log data and stored as the click log data. For example, recopick = METHOD may be added to the link variable of the recommendation item, and if recopick exists in the user's visit URL in the user log data, this may be determined as a click by recommendation. After checking the item identifier (ITEM_ID) in the reference address (Referrer) information, the source item identifier may be checked using the item identifier (ITEM_ID). However, when the reference address information has the URL of the service device 30, it may not be determined as a recommended click because it is already collected as click log data in the first method.

The service device 30 described above may collect click log data using the first method and / or the second method, and store the collected click log data.

Table 11 illustrates the format of the click log data.

Table 11

Log Type	Version (VERSION)	Service identifier (SERVICE_ID)	Click item (CLICK ITEM)	SOURCE ITEM	Algorithm (METHOD)	TIMESTAMP
click	2	One	597	117	2	20XX-YY-ZZ aa: bb: cc
click	2	One	11556	11816	2	20XX-YY-ZZ aa: bb: cc

Referring to Table 11, the click log data may include a log type (click or view), version information on the recommendation item list, service identifier, click item information, source item information, recommendation algorithm, click time information, and the like.

Next, referring to the method of collecting the view log data, the service device 30 may obtain and store the view log data through the recommendation item list provided whenever the web server 20 requests the recommendation item list.

Table 12 illustrates the format of view log data.

Table 12

Log Type	Version (VERSION)	Service identifier (SERVICE_ID)	Click item (CLICK ITEM)	SOURCE ITEM	RECOMMENDED LIST (ITEM_ID: METHOD, ITEM_ID: METHOD, ITEM_ID: METHOD)	Algorithm (METHOD)	TIMESTAMP
viewing	2	One	-	182	158: 2,73: 2,566: 2,85: 2,89: 2	2	20XX-YY-ZZ aa: bb: cc
viewing	2	14	-	13830	13844: 2,13841: 2,13828: 2,13843: 2,13845: 2	2	20XX-YY-ZZ aa: bb: cc
viewing	2	14	-	13670	13732: 2,13534: 2,13780: 2,13698: 2,13822: 2	2	20XX-YY-ZZ aa: bb: cc
viewing	2	14	-	13670	13732: 2,13534: 2,13780: 2,13698: 2,13822: 2	2	20XX-YY-ZZ aa: bb: cc
viewing	2	6	-	43	15: 2,299: 2,316: 2,329: 2,16: 2,426: 2,92: 2,210: 2,276: 2,83: 2	2	20XX-YY-ZZ aa: bb: cc
viewing	2	One	-	176	512: 2,153: 2,809: 2,637: 2,177: 2	2	20XX-YY-ZZ aa: bb: cc
viewing	2	6	-	178	41: 2,398: 2,115: 2,421: 2,230: 2,12: 2,372: 2,177: 2,45: 2,293: 2	2	20XX-YY-ZZ aa: bb: cc
viewing	2	14	-	13795	13796: 2,13785: 2,13787: 2	2	20XX-YY-ZZ aa: bb: cc

Referring to Table 12, the click log data may include a log type (click or view), version information for the recommendation item list, service identifier, click item information, source item information, recommendation list, recommendation algorithm, click time information, and the like. Can be.

As such, the service device 30 that collects the click log data and the view log data may calculate the click rate using the click log data and the view log data. The service device 30 may add up the total number of views exposed to a specific item by date, and also add up the number of times a user clicks, and then calculate a click rate between each source item and each recommendation item.

Table 13 illustrates the clickthrough rate.

Table 13

DATE	Algorithm (ALGORITHM)	Source item identifier (SOURCE_ITEM_ID)	Featured Item Identifier (RECOMMENDED_ITEM_ID)	Click count (CLICK COUNT)	VIEW COUNT	Clickthrough rate (CLICK RATIO)
20XX-YY-ZZ	2	123	234	4	100	4%
20XX-YY-ZZ	2	123	345	11	100	11%
20XX-YY-ZZ	2	123	456	0	100	0%
20XX-YY-ZZ	2	123	789	7	100	7%
20XX-YY-ZZ	3	123	111	2	100	2%
20XX-YY-ZZ	3	123	222	3	100	3%
20XX-YY-ZZ	3	123	333	0	100	0%
20XX-YY-ZZ	3	123	444	5	100	5%

Referring to Table 13, Algorithm 2's YYYY ZZ Day 20 20XX rate for item 123 is (4 + 11 + 0 + 7) / 100 = 22%, and Algorithm 3's YYYYYY ZZ Day 20 Clickthrough Rate ( 2 + 3 + 0 + 5) / 100 = 10%.

The service device 30 may calculate a click rate between each source item and each recommendation item, and then designate the item as a prohibited item if the click rate is below a certain threshold (or below) for a predetermined period of time. That is, if the recommended item B for item A is less than 1% for 7 days, item B may be removed from the recommended item list for item A. In addition, the service device 30 may designate a prohibition item when the number of click counts between each source item and each recommendation item is less than (or less than) a certain threshold for a predetermined period. When removing the recommended prohibited item from the recommendation result (or recommendation item list), the service device 30 arbitrarily deletes an item to be deleted from the previously generated recommendation result (or recommendation item list), or The recommendation score can be changed to a lower value or a variable value can be adjusted so that a lower value is calculated to remove the prohibited item from the recommendation result (or recommendation item list).

This feedback process may be automatically performed in the service device 30. In addition, the service device 30 provides the web server 20 with the click rate information of the corresponding item and the click rate information or the number of clicks between the recommended item and the recommended item for each item serviced by the web server 20, and the web server 20. The feedback processing may be performed in the service device 30 according to a request from the service device 30.

10 is a diagram illustrating a click rate for each recommended item according to an embodiment of the present invention.

As illustrated in FIG. 10, the service device 30 may provide the web server 20 with information about the source item 101 and schedule information about the source item 101 such as the name of the source item 101 and average click-through rate information for a predetermined period of time. The overall referral clickthrough rate 102 may be provided for a period of time. In addition, the recommendation item list 103 for the source item 101, a name, an algorithm, a recommendation score, a click rate between the source items, an operation input button, and the like may be provided for each recommendation item. Through such information, the administrator of the web server 20 may check the clickthrough rate of the source item 101 along with the clickthrough rate of the source item 101 for each recommendation item by date. In the case of FIG. 10, the click rate of the recommendation item 104 displayed at the bottom stays at about 0% for a certain period. As such, when the web server 20 administrator selects the prohibition of recommendation button for the recommendation item having a low clickthrough rate, the web server 20 transmits it to the service device 30 so that the service device 30 transmits the recommendation item to the source item. Recommendation 101 may be prohibited. In addition, when the web server 20 administrator selects the first exposure button separately, the web server 20 transmits it to the service device 30, so that the service device 30 recommends a score (score) of the selected recommendation item. Regardless of this, the recommendation item list may be generated to display the corresponding recommendation item at the highest priority (for example, at the top or the leftmost side).

In addition, the service device 30 may automatically select the most suitable algorithm for each web server 20 in consideration of click rate information or click count information.

First, in order to apply the algorithm having the highest efficiency (that is, the highest click rate or the highest number of clicks) for each service provided by the web server 20, the service device 30 generates a plurality of user log data in a certain condition or arbitrarily. It is divided into groups and provides a list of recommended items calculated by different algorithms for each group. For example, the service device 30 arbitrarily divides the user log data into groups A and B (for example, group A when the odd user identifier is odd, group B when the even number is odd), and recommends algorithm 1 to the group A. , The group B may calculate a recommendation using algorithm 2, and provide the web server 20 with a list of recommended items. In addition, the service device 20 is a user connected to the web server 20 through the mobile terminal and the group connected to the web server 20 through the fixed terminal using the user agent information of the web browser in the user log data. The web server 20 may provide a list of recommended items by dividing log data and calculating recommendations through different algorithms.

In addition, the service device 30 may calculate the recommendation by selecting the algorithm having the highest click rate or the algorithm having the highest number of clicks as the algorithm of the corresponding service by separately calculating the click rate for each group. The feedback process may also be automatically performed by the service device 30, and the service device 30 may provide the web server 20 with click rate information or click count information for each user group (that is, for each algorithm). 20, and a feedback process may be performed in the service device 30 by a request from the web server 20.

11 is a diagram illustrating a recommendation rate for each algorithm according to an embodiment of the present invention.

As shown in FIG. 11, since the recommendation rate for Algorithm 1 applied to user group A is higher than the recommendation rate for Algorithm 2 for a predetermined period of time, the service device 30 recommends an algorithm for the web server 20 that provides the service. Can be selected.

Embodiments according to the present invention may be implemented by various means, for example, hardware, firmware, software, or a combination thereof. In the case of a hardware implementation, an embodiment of the present invention may include one or more application specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), FPGAs ( field programmable gate arrays), processors, controllers, microcontrollers, microprocessors, and the like.

In addition, in the case of the implementation by firmware or software, an embodiment of the present invention is implemented in the form of a module, a procedure, a function, etc. for performing the above-described functions or operations, so that the recording medium can be read by various computer means. Can be recorded. Here, the recording medium may include a program command, a data file, a data structure, etc. alone or in combination. Program instructions recorded on the recording medium may be those specially designed and constructed for the present invention, or they may be of the kind well-known and available to those having skill in the computer software arts. For example, the recording medium may be magnetic media such as hard disks, floppy disks and magnetic tapes, optical disks such as Compact Disk Read Only Memory (CD-ROM), digital video disks (DVD), Magnetic-Optical Media, such as floppy disks, and hardware devices specially configured to store and execute program instructions, such as ROM, random access memory (RAM), flash memory, and the like. do. Examples of program instructions may include high-level language code that can be executed by a computer using an interpreter as well as machine code such as produced by a compiler. Such hardware devices may be configured to operate as one or more software modules to perform the operations of the present invention, and vice versa.

As described above, although the present specification and drawings disclose preferred embodiments of the present invention, in addition to the embodiments disclosed herein, other modifications based on the technical idea of the present invention may be implemented in the technical field to which the present invention belongs. It is self-evident to those of ordinary knowledge. In addition, although specific terms are used in the specification and the drawings, they are only used in a general sense to easily explain the technical contents of the present invention and to help the understanding of the present invention, and are not intended to limit the scope of the present invention. Accordingly, the above detailed description should not be construed as limiting in all aspects and should be considered as illustrative. The scope of the invention should be determined by reasonable interpretation of the appended claims, and all changes within the equivalent scope of the invention are included in the scope of the invention.

In addition, the device or the terminal according to the present invention may be driven by a command that causes one or more processors to perform the functions and processes described above. For example, such instructions may include interpreted instructions, such as script instructions such as JavaScript or ECMAScript instructions, or executable instructions or other instructions stored on a computer readable medium. Furthermore, the device according to the present invention may be implemented in a distributed manner over a network, such as a server farm, or may be implemented in a single computer device.

Furthermore, a computer program (also known as a program, software, software application, script or code) mounted on an apparatus according to the invention and executing a method according to the invention comprises a compiled or interpreted language or a priori or procedural language. It can be written in any form of programming language, and can be deployed in any form, including stand-alone programs or modules, components, subroutines, or other units suitable for use in a computer environment. Computer programs do not necessarily correspond to files in the file system. A program may be in a single file provided to the requested program, in multiple interactive files (eg, a file that stores one or more modules, subprograms, or parts of code), or part of a file that holds other programs or data. (Eg, one or more scripts stored in a markup language document). The computer program may be deployed to run on a single computer or on multiple computers located at one site or distributed across multiple sites and interconnected by a communication network.

In addition, in describing the embodiments according to the present invention, although the operations are described in the drawings in a specific order, it is necessary to perform such operations in the specific order or the sequential order shown in order to obtain a desirable result, or all the illustrated operations. Should not be understood to be performed. In certain cases, multitasking and parallel processing may be advantageous. Moreover, the separation of the various system components of the above-described embodiments should not be understood as requiring such separation in all embodiments, and the described program components and systems will generally be integrated together into a single software product or packaged into multiple software products. It should be understood that it can.

The present invention recommends other items related to items served by a web site. Accordingly, it can be applied directly to its own service without additional development on the web site that provides various contents, products or services, and feedback is reflected in the recommendation results based on the click / view log data generated from the user. By providing more suitable items can be recommended. That is, the method of providing the recommendation result of the present invention can be applied to various wired / wireless communication systems. This has industrial applicability because it is not only sufficient marketable or business possibility, but also practically evident.

Claims (24)

1. Communication unit for transmitting and receiving data;

   A storage unit for storing user log data and recommendation results; And

   Collecting the user log data related to the web site from the terminal accessing the web site through a script embedded in the web site, and services on the web site using a recommendation algorithm based on the user log data Calculating a recommendation result for one or more recommendation items related to each source item, and extracting a recommendation item list related to the requested source item from the recommendation result in response to a recommendation item list request received from the web server of the website; A control unit for transmitting to the web server;

   Service device comprising a.
2. The method of claim 1, wherein the control unit,

   And a click rate for a recommendation item associated with each source item through click log data and view log data for the recommendation item.
3. The method of claim 2, wherein the control unit,

   And the recommendation item of which the click rate is continuously below a threshold for a predetermined period is excluded from the recommendation result.
4. The method of claim 1, wherein the control unit,

   And dividing the user log data into a predetermined condition or arbitrarily into a plurality of groups, and calculating the recommendation result using a different recommendation algorithm for each group.
5. The method of claim 4, wherein the control unit,

   And a click rate for each group through click log data and view log data for the recommended item, and calculates the recommendation result using a recommendation algorithm used for the group having the highest click rate.
6. The method of claim 1, wherein the control unit,

   And when the terminal is connected through a link of the recommendation item connected to the service apparatus, click log data for the recommendation item through the access URL of the terminal.
7. The method of claim 1, wherein the control unit,

   And extracting user data including a predetermined parameter value from a visit URL (Uniform Resource Locator) included in the user log data, and obtaining the extracted user data as click log data for the recommended item.
8. The method of claim 1, wherein the control unit,

   And when the recommendation item list request is made from the web server, view log data for the recommendation item through the recommendation item list.
9. The method of claim 1, wherein the control unit,

   And extracting item information displayed on the visit URL and operation information related to the operation of the terminal from the visit URL from the visit URL included in the user log data.
10. The method of claim 1, wherein the control unit,

    The recommendation result is classified and stored for each source item, and a hash is added to a string including a service identifier for identifying the web site, an item identifier for the source item, and a recommendation item identifier for the recommendation item associated with the source item. and a value calculated by applying a hash) function together with the recommendation result.
11. The method of claim 1, wherein the control unit,

    And a list of recommended items related to the requested source item in a JavaScript Object Notation (JSON) format or an iframe format.
12. The method of claim 1,

    The user log data includes a domain of the web site, a visit URL (Uniform Resource Locator) of the terminal, a user identifier, and visit time information.
13. The method of claim 12,

    The user identifier is generated by a web browser cookie of the terminal, characterized in that generated by combining a random number and the generation time.
14. Collecting, by a service device, the user log data related to the web site from a terminal accessing the web site through a script embedded in a web site;

    Calculating, by the service device, a recommendation result for at least one recommendation item associated with each source item served by the web site using a recommendation algorithm based on the user log data; And

    Extracting, by the service apparatus, a recommendation item list related to the requested source item from the recommendation result in response to the recommendation item list request received from the web server of the web site, and transmitting it to the web server;

    Recommended item providing method comprising a.
15. The method of claim 14,

    Calculating, by the service device, a click rate for a recommendation item associated with each source item through click log data and view log data for the recommendation item;

    Recommended item providing method characterized in that it further comprises.
16. The method of claim 15,

    The service device excluding a recommendation item from the recommendation result that the click rate continues for a period of time below a threshold value;

    Recommended item providing method characterized in that it further comprises.
17. The method of claim 14, wherein the calculating of the recommendation result comprises:

    And the service device divides the user log data into a predetermined condition or arbitrarily into a plurality of groups, and calculates the recommendation result by using a different recommendation algorithm for each group.
18. The method of claim 17, wherein the calculating of the recommendation result comprises:

    The service device calculates a click rate for each group through the click log data and the view log data of the recommended item, and calculates the recommendation result using a recommendation algorithm used for the group having the highest click rate. How to deliver the item.
19. The method of claim 14,

    Acquiring click log data for the recommended item through the access URL of the terminal when the terminal is connected through the link of the recommended item to which the service device is connected to the service device;

    Recommended item providing method characterized in that it further comprises.
20. The method of claim 14,

    Extracting, by the service device, user data including a predetermined parameter value from a visit URL (Uniform Resource Locator) included in the user log data, and obtaining the user data as click log data for the recommended item;

    Recommended item providing method characterized in that it further comprises.
21. The method of claim 14,

    Obtaining, by the service apparatus, view log data for the recommendation item through the recommendation item list when the recommendation item list request is made by the web server;

    Recommended item providing method characterized in that it further comprises.
22. The method of claim 14,

    Extracting, by the user device, item information displayed on the visit URL from the visit URL (Uniform Resource Locator) included in the user log data and operation information related to the operation of the terminal from the visit URL;

    Recommended item providing method characterized in that it further comprises.
23. The method of claim 14,

    The service apparatus stores the recommended result for each source item, and stores the classified result, wherein a service identifier for identifying the web site, an item identifier for the source item, and a recommended item identifier for the recommended item associated with the source item are added together. Storing a value calculated by applying a hash function to a string together with the recommendation result;

    Recommended item providing method characterized in that it further comprises.
24. A computer-readable recording medium storing a program for executing the method for providing a recommended item according to any one of claims 14 to 23.

Images