KR102164814B1 - Method for providing product recommendation service based on deep-learning neural network - Google Patents

Abstract

A method of providing a product recommendation service based on a deep learning neural network according to an embodiment of the present invention is a method in which an online shopping mall server provides a customized recommendation product based on a deep learning neural network, wherein the product of the product posted in the online shopping mall Obtaining information; Acquiring purchase preference information of the online shopping mall users based on the product information, user's body shape information, and purchase history information to establish a purchase preference information database; Detecting purchase preference information of the user when the user accesses the online shopping mall; Comparing the detected purchase taste information of the user with the purchase taste information of another user to calculate a purchase taste similarity; Selecting the other user whose calculated purchase taste similarity is greater than or equal to a predetermined standard as a similar user of the user; And extracting a list of previously purchased products of the selected similar user and providing them as the recommended products to the user.

Description

Deep learning neural network based product recommendation service provision method {METHOD FOR PROVIDING PRODUCT RECOMMENDATION SERVICE BASED ON DEEP-LEARNING NEURAL NETWORK}

The present invention relates to a method of providing a service for recommending a product based on a deep learning neural network.

In more detail, it relates to a method of providing a user-customized recommended product based on a deep learning neural network in an online shopping mall.

In modern times, with the development of information and communication technologies (ICT, Information&Communications Technologies), the online/internet shopping mall industry is growing day by day.

Here, an online shopping mall or an Internet shopping mall means a place where products can be purchased and sold online through a network such as the Internet.

These online/internet shopping malls (hereinafter, online shopping malls) require differentiated services that can increase the competitiveness of shopping malls as the number of individuals rapidly increases in recent years, and various solutions to provide them are required.

Meanwhile, a neural network that mimics the human brain has been devised, and as deep learning technology that combines and analyzes and learns external data by computer itself, AI (Artificial Intelligence) is a breakthrough. As a result, computers can replace humans to analyze vast amounts of data and cluster or classify objects or data.

The present invention aims to provide a service for recommending customized products based on the body shape and purchase history of online shopping mall users by using a deep learning neural network in order to provide differentiated services that can increase the competitiveness of online shopping malls. To do.

However, the technical problems to be achieved by the present invention and the embodiments of the present invention are not limited to the technical problems as described above, and other technical problems may exist.

A method of providing a product recommendation service based on a deep learning neural network according to an embodiment of the present invention is a method in which an online shopping mall server provides a customized recommendation product based on a deep learning neural network, wherein the product of the product posted in the online shopping mall Obtaining information; Acquiring purchase preference information of the online shopping mall users based on the product information, user's body shape information, and purchase history information to establish a purchase preference information database; Detecting purchase preference information of the user when the user accesses the online shopping mall; Comparing the detected purchase taste information of the user with the purchase taste information of another user to calculate a purchase taste similarity; Selecting the other user whose calculated purchase taste similarity is greater than or equal to a predetermined standard as a similar user of the user; And extracting a list of previously purchased products of the selected similar user and providing them as the recommended products to the user.

In this case, the step of generating product information of the product posted on the online shopping mall includes obtaining fit numerical information of the product through the deep learning neural network, and the fit numerical information is obtained from a tight-size fit. Refers to information expressed by converting the degree of fit of clothes to over-sized fit into numerical values.

In addition, generating product information of a product posted on the online shopping mall includes obtaining information on a clothing style category of the product through the deep learning neural network, wherein the clothing style category information is previously learned. It is a standard for classifying clothes that are preset by data.

Here, the product information includes numerical information, image information, fit numerical information, and clothing style category information of the product.

In addition, the step of establishing a purchase preference database by acquiring purchase preference information of the online shopping mall users based on the product information, the user's body shape information, and purchase history information may include information about the user's preferred fit value or preferred clothing style category information. And acquiring, wherein the preferred fit value information is an arbitrary value obtained based on an average value of the fit value information of at least one product previously purchased by the user in the online shopping mall.

In addition, the preferred clothing style category information includes counting the clothing style categories to which at least one or more products previously purchased by the user in the online shopping mall are classified and belonged to, and at least one or more clothing with a high counting value. This is the information on the style category.

In addition, calculating a purchase taste similarity by comparing the detected purchase taste information of the user with the purchase taste information of another user includes the detected purchase taste information of the user and the purchase taste information of another user. And outputting the similarity of the purchase taste with the user for each of at least one other user by inputting into a deep learning neural network.

In addition, calculating a purchase taste similarity by comparing the detected purchase taste information of the user with the purchase taste information of another user may include a body shape information category, a preferred fit value category, or a preferred clothing style included in the purchase taste information. And first extracting at least one or more other users by measuring a degree of similarity between the user and the other user in the category.

On the other hand, the method for providing a product recommendation service based on a deep learning neural network according to an embodiment of the present invention includes the steps of generating interest level information indicating a degree of interest of the user with respect to the provided recommended product, and the generated interest level information. It further includes the step of utilizing when the user selects the similar user.

In addition, the method for providing a product recommendation service based on a deep learning neural network according to an embodiment of the present invention further includes providing a user interface for performing a follow request function for the selected similar user to the user.

In this case, the interest level information is generated based on at least one or more of the number of times the user selects the recommended product and a purchase amount.

In addition, generating interest level information indicating a degree of interest of the user with respect to the provided recommended product includes selecting a preferred purchase preference category of the user based on the interest level information.

In addition, the step of utilizing the generated interest level information when selecting the similar user by the user may include matching and storing information on the preferred purchase preference category with the user, and then with the user when the user selects the similar user. And selecting the similar user of the user by giving a weight to the matched and stored preferred purchase preference category.

On the other hand, the providing of a user interface for performing a follow application function for the selected similar user to the user may include, when a following user who is a followed user purchases the product in the online shopping mall, the following user Providing following user purchase information including at least one or more of an alarm notifying that a product has been purchased, information on the product, a product review of the following user for the product, and a wearing shot to a follower user who is a follower And providing a discount means that can be used when purchasing the product in the online shopping mall to the following user having a predetermined number or more of the follower users.

The method for providing a product recommendation service based on a deep learning neural network according to an embodiment of the present invention provides a service for recommending a customized product using a deep learning neural network based on the body shape and purchase history data of an online shopping mall user. It is possible to increase competitiveness through the provision of differentiated services of the company, enable users to use online shopping malls more efficiently, and create an effect of increasing sales through user-customized recommended products.

In addition, the method of providing a product recommendation service based on a deep learning neural network according to an embodiment of the present invention builds its own database based on the body shape and purchase history data of a plurality of users, and based on the data of the self-established database. By performing deep learning through a deep learning neural network, a faster and more efficient product recommendation service can be provided.

Moreover, the method for providing a product recommendation service based on a deep learning neural network according to an embodiment of the present invention uses a deep learning neural network to generate data necessary for a product recommendation service of an online shopping mall, thereby generating a vast amount of data by artificial intelligence. It can be effectively analyzed and classified or clustered.

In addition, the method for providing a product recommendation service based on a deep learning neural network according to an embodiment of the present invention generates and utilizes user's feedback information on the recommended product, so that not only artificial intelligence judgment through the deep learning neural network but also the user's perception It is possible to implement a service that provides customized recommended products by considering appropriate judgment.

In addition, the method for providing a product recommendation service based on a deep learning neural network according to an embodiment of the present invention provides a reward for users who spread the influence on product purchase to a plurality of users, so that users are more actively online. There is an effect of encouraging to perform activities through shopping malls, improving shopping mall operation competitiveness, and improving sales.

However, the effects obtainable in the present invention are not limited to the above-mentioned effects, and other effects not mentioned may be clearly understood from the following description.

1 is a conceptual diagram of a system for providing a product recommendation service based on a deep learning neural network according to an embodiment of the present invention.
2 is an internal block diagram of an online shopping mall server according to an embodiment of the present invention.
3 is an internal block diagram of a deep learning neural network server according to an embodiment of the present invention.
4 is a flowchart illustrating a method of providing a product recommendation service based on a deep learning neural network according to an embodiment of the present invention.
5 is a flowchart illustrating a process of selecting a similar user of a first user according to an embodiment of the present invention in more detail.

Since the present invention can apply various transformations and have various embodiments, specific embodiments are illustrated in the drawings and will be described in detail in the detailed description. Effects and features of the present invention, and a method of achieving them will be apparent with reference to the embodiments described later in detail together with the drawings. However, the present invention is not limited to the embodiments disclosed below and may be implemented in various forms. In the following embodiments, terms such as first and second are not used in a limiting meaning, but for the purpose of distinguishing one component from another component. In addition, expressions in the singular include plural expressions unless the context clearly indicates otherwise. In addition, terms such as include or have means that the features or elements described in the specification are present, and do not preclude the possibility of adding one or more other features or elements in advance. In addition, in the drawings, the size of components may be exaggerated or reduced for convenience of description. For example, the size and thickness of each component shown in the drawings are arbitrarily shown for convenience of description, and the present invention is not necessarily limited to what is shown.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings, and when describing with reference to the drawings, the same or corresponding constituent elements are assigned the same reference numerals, and redundant descriptions thereof will be omitted. .

1 is a conceptual diagram of a system for providing a product recommendation service based on a deep learning neural network according to an embodiment of the present invention.

Referring to FIG. 1, a product recommendation service system based on a deep learning neural network may include an online shopping mall server 100, a deep learning neural network server 200, and a terminal 300.

Here, each component of FIG. 1 may be connected through a network. The network refers to a connection structure capable of exchanging information between nodes such as the online shopping mall server 100, the deep learning neural network server 200, and the terminal 300, and an example of such a network is 3GPP (3rd Generation Partnership Project) network, LTE (Long Term Evolution) network, WIMAX (World Interoperability for Microwave Access) network, Internet, LAN (Local Area Network), Wireless LAN (Wireless Local Area Network), WAN (Wide Area Network) , Personal Area Network (PAN), Bluetooth (Bluetooth) network, satellite broadcasting network, analog broadcasting network, Digital Multimedia Broadcasting (DMB) network, and the like, but are not limited thereto.

However, the components shown in FIG. 1 are not essential for implementing a product recommendation service system based on a deep learning neural network, so the product recommendation service system based on a deep learning neural network described in this specification includes the configurations listed above. It may have more or fewer elements than elements.

-Online shopping mall server

First, in an embodiment of the present invention, the online shopping mall server 100 may implement an online shopping mall that provides a product recommendation service based on a deep learning neural network.

In detail, the online shopping mall server 100 may provide a place where products can be purchased and sold online through a network such as the Internet, and the online shopping mall server 100 and the user's terminal 300 and/or other Data transmission/reception may be performed between servers, and data processing related to a product recommendation service based on a deep learning neural network may be operated based on various data transmitted and received.

In detail, referring to FIG. 2, the online shopping mall server 100 may include a data transmission/reception unit 110, a data processing unit 120, a payment processing unit 130, and a database 140.

First, the data transmission/reception unit 110 forms a communication channel between the online shopping mall server 100 and other servers (eg, deep learning neural network server 200) and/or the terminal 300, and is based on a deep learning neural network. It is possible to perform a series of wired/wireless communication for transmitting and receiving data related to the product recommendation service.

The data transmission/reception unit 110 includes technical standards or communication methods for mobile communication (eg, Global System for Mobile communication (GSM), Code Division Multi Access (CDMA)), High Speed Downlink Packet Access (HSDPA), Radio with at least one of a base station, an external terminal 300, and an arbitrary server on a mobile communication network built according to HSUPA (High Speed Uplink Packet Access), LTE (Long Term Evolution), LTE-A (Long Term Evolution-Advanced), etc. Can send and receive signals.

In addition, the data transmission/reception unit 110 includes WLAN (Wireless LAN), Wi-Fi (Wireless-Fidelity), Wi-Fi (Wireless Fidelity) Direct, DLNA (Digital Living Network Alliance), WiBro (Wireless Broadband), WiMAX ( World Interoperability for Microwave Access) can also transmit and receive wireless signals.

In addition, the data transmission/reception unit 110 includes Bluetooth™, Radio Frequency Identification (RFID), Infrared Data Association (IrDA), Ultra Wideband (UWB), ZigBee, Near Field Communication (NFC), Wi- Short-range communication may be supported using at least one of Wireless-Fidelity (Fi), Wi-Fi Direct, and Wireless USB (Wireless Universal Serial Bus) technologies.

Next, the data processing unit 120 may perform a series of data processing related to a product recommendation service based on a deep learning neural network based on transmission/reception and various data generated.

In an embodiment, the data processing unit 120 includes at least one of a body shape similarity, a fit similarity, and a clothes style similarity between the first user and each other user, based on purchase preference information of the first user and a plurality of other users. Can be calculated.

Next, the payment processing unit 130 may perform a payment process by performing a user's purchase procedure for a product posted on the online shopping mall.

Next, the database 140 may store any one or more of application programs, data, and commands required for functional operation according to an embodiment of the present invention, and a product information database 141 and a purchase preference information database 142 for each user It may include.

Here, the product information database 141 may store and manage product information for each product on various products posted on the online shopping mall.

For example, the product information database may store size and/or numerical information for each product (ie, clothes), image information, fit numerical information, clothing style category information, and the like.

In addition, the user-specific purchase preference information database 142 may store and manage user-specific purchase preference information generated based on user-specific body shape information, preferred fit numerical information, and preferred clothing style information collected from a plurality of users. .

In addition, a more detailed description will be given in the detailed description of a method for providing a product recommendation service based on a deep learning neural network to be described below.

The database 140 may be a variety of storage devices such as ROM, RAM, EPROM, flash drive, hard drive, etc., and is a web storage that performs a storage function of the database 140 on the Internet. May be.

-Deep Learning Neural Network Server

On the other hand, in the embodiment of the present invention, the deep learning neural network server 200 performs deep-learning suitable for the attribute of the data to be derived based on various data received from the online shopping mall server 100. This can be done so that a product recommendation service based on a deep learning neural network can operate smoothly.

In detail, referring to FIG. 3, the deep learning neural network server 200 may include a fit numerical calculation neural network 210, a clothes style classification neural network 220, and a purchase preference similarity output neural network 230. .

First, the fit numerical calculation neural network 210 may perform deep learning based on the clothing numerical information of the product information for a specific product received from the online shopping mall server 100, and the specific product as a result of the deep learning. You can print information about the fit value.

Next, the clothes style classification neural network 220 may perform image deep learning based on the clothes image information of the product information for a specific product received from the online shopping mall server 100, and as a result of the deep learning, You can output the most suitable clothing style category for the product. In this case, the clothing style category may be preset based on previously learned data.

Finally, in the embodiment, the neural network 230 outputting the similarity of purchase preferences may receive purchase preference information for a first user and purchase preference information for a plurality of other users excluding the first user from the online shopping mall server 100. I can.

In addition, the purchase taste similarity output neural network 230 may compare the received first user purchase taste information with the other user purchase taste information and output a purchase taste similarity with the first user for each of at least one other user.

In addition, a more detailed description will be given in the detailed description of a method for providing a product recommendation service based on a deep learning neural network to be described below.

-Terminal

In addition, in an embodiment of the present invention, the terminal 300 is a terminal 300 capable of accessing various applications and/or programs related to a product recommendation service based on a deep learning neural network, and includes a product recommendation service based on a deep learning neural network. It may include a portable terminal 300 in which an application providing a related function is installed or a fixed terminal 300 in which a program is installed.

For example, the terminal 300 is a smart phone that is a portable terminal 300, a digital broadcasting terminal 300, a mobile phone, a personal digital assistant (PDA), a portable multimedia player (PMP), a navigation system, and a tablet PC. ), a wearable device, and smart glass.

In addition, the terminal 300 is a product based on a deep learning neural network based on wired/wireless communication, such as a desktop PC, which is a fixed terminal 300, a laptop computer, and a personal computer such as an ultrabook. A computing device capable of accessing a program related to the recommendation service may further be included.

In more detail, the terminal 300 may include a communication unit, a display unit, an input unit, a memory, and a control unit.

First, the communication unit establishes a communication channel between the terminal 300 and another server and/or terminal 300, and performs a series of wired/wireless communications for transmitting and receiving data related to a product recommendation service based on a deep learning neural network. can do.

Next, the display unit may output and provide a graphic image related to a product recommendation service based on a deep learning neural network.

For example, the display unit may output and display various product images of an online shopping mall as a graphic.

Such display units include liquid crystal displays (LCD), thin film transistor-liquid crystal displays (TFT LCDs), organic light-emitting diodes (OLEDs), and flexible displays. , It may include at least one of a 3D display and an e-ink display.

Next, the input unit may detect a user's input related to a product recommendation service based on a deep learning neural network.

For example, the input unit may detect an input of the user's body shape information, and may detect a user's selection input, purchase input, and the like for recommended products.

In addition, the input unit and the display unit may be implemented by being coupled to a touch screen.

Next, the memory may store any one or more of an application program, data, and command required for a functional operation according to an embodiment of the present invention.

That is, the memory may store applications, programs, various data, commands, etc. related to a product recommendation service based on a deep learning neural network.

Such a memory may be a variety of storage devices such as ROM, RAM, EPROM, flash drive, hard drive, and the like, or may be a web storage that performs a storage function of the memory over the Internet.

Finally, the control unit may control and drive the overall operation of each unit described above.

These controls include application specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers ( It may be implemented using at least one of micro-controllers), microprocessors, and electrical units for performing other functions.

-Deep learning neural network based product recommendation service provision method

Hereinafter, a method of providing a product recommendation service based on a deep learning neural network in an online shopping mall will be described in detail with reference to the drawings.

4 is a flowchart illustrating a method of providing a product recommendation service based on a deep learning neural network according to an embodiment of the present invention.

Referring to FIG. 4, the online shopping mall server 100 (hereinafter, shopping mall server) may generate product information for each product of the online shopping mall. (S101)

Here, the product information refers to at least one of classification (ie, product category) information, option information, numerical information, image information, fit numerical information, clothes style category information, and/or product purchase satisfaction information. It means information that includes information.

At this time, the shopping mall server 100 may receive and obtain product category information, option information, numerical information, and/or image information of a product from another server or user terminal 300, etc. A deep learning neural network can be used to obtain style category information.

In detail, the shopping mall server 100 may first receive and obtain at least one of product category information, option information, and numerical information from another server or the user terminal 300.

In this case, the product category information is information on a category in which the product is classified, and may include, for example, a top, a bottom, an outer, and an accessory.

Also, the product option information may be information including size information and/or color information of the product.

Further, the numerical information of the product means dimension information measured for a specific area of the product, and may include, for example, information about the shoulder width of the product, information about the chest measurement, and the like.

Meanwhile, the shopping mall server 100 may acquire product fit values and clothing style category information based on artificial intelligence through a deep learning neural network.

In detail, the shopping mall server 100 first performs an operation of calculating a fit value based on the product size and/or numerical information to the neural network 210 for calculating the fit value for each product. Deep learning can be performed by putting the product size and/or numerical information.

Here, the fit numerical information means information expressed by converting the degree of fit of clothes from a tight-size fit to an over-size fit into a numerical value. That is, in the embodiment, the fit numerical information may be an arbitrary value between 0 and 100.

That is, the shopping mall server 100 may determine whether the product is a tight-size fit or an over-size fit according to a predetermined criterion based on the fit numerical information of the product.

For example, when the fit value information of a specific product is 80, the shopping mall server 100 may determine the product as an over-size fit according to a preset criterion.

In addition, in order to determine the clothing style category of the product, the shopping mall server 100 determines the clothing style category for each product on the basis of the image information of the product. Image information can be inserted to perform image deep learning.

Here, the clothing style category may be a standard for classifying clothes preset based on previously learned data, for example, casual, street, vintage, modern, feminine, dandy, minimalism, undisclosed simalism, layered, classic, sporty, etc. Can contain various categories of.

Thus, the shopping mall server 100 may acquire the clothing style category information of the product determined through image deep learning, and automatically classify and manage the product according to the clothing style category of the acquired product.

In addition, the shopping mall server 100 may utilize fit numerical information and/or clothing style category information obtained for a product when posting a corresponding product on an online shopping mall.

For example, when displaying a specific product through an online shopping mall, the shopping mall server 100 inserts and provides fit value information and/or clothing style category information of a specific product in the name (ie, product name) of a specific product. I can.

In this way, the shopping mall server 100 may perform effective data clustering using artificial intelligence by acquiring fit numerical information and clothing style category information required for product recommendation services of an online shopping mall through a deep learning neural network. .

In addition, the shopping mall server 100 may provide a user satisfaction evaluation interface through an online shopping mall in order to obtain product purchase satisfaction information.

In detail, the shopping mall server 100 provides a user satisfaction evaluation interface to a user who has purchased a product, and the overall satisfaction with the product and/or the similarity between the expected fit of the product and the actual fit after purchasing the product Information, etc. can be obtained.

In addition, the shopping mall server 100 may include the acquired product purchase satisfaction information in product information, and may include and post product purchase satisfaction information obtained when the product is posted on the online shopping mall.

That is, the shopping mall server 100 includes classification (ie, product category) information, option information (ie, size information and/or color information), numerical information, image information, fit numerical information and clothing style category information, A product recommendation service based on a deep learning neural network may be provided by utilizing at least one or more of the product purchase satisfaction information to provide a product recommendation service based on objective data to a user.

Next, the shopping mall server 100 may acquire purchase preference information of an online shopping mall user to establish a purchase preference information database 142 for each user. (S103)

In detail, the shopping mall server 100 may generate purchase preference information of a user based on at least one of information about a user's body type, a preferred fit value, a preferred clothing style category, and a preferred size.

At this time, the user's body shape information may be obtained from an input through the user's terminal 300 when the user subscribes to the online shopping mall, for example, an inverted triangle body shape, a triangle body shape, an X shape body shape, a circular body shape and a rectangular body shape. Any one of the body types may be information selected by the user.

In addition, the preferred fit numerical information may be an arbitrary numerical value obtained based on the body shape information of the user and the average fit numerical value of at least one product (clothing) previously purchased by the user in the online shopping mall.

In another embodiment, the preferred fit numerical information may be information on a preferred fit selected by the user through a preferred fit selection interface provided to the user through an online shopping mall.

For example, the preferred fit numerical information may be information that the user selects one of a loose fit, a regular fit, and a tight fit through a preferred fit selection interface.

In addition, the preferred clothing style category information includes counting clothing style categories to which at least one product (clothing) previously purchased by the user in the online shopping mall is classified and belonged to, and determines at least two or more clothing style categories with a high counting value. It can be obtained by discriminating and selecting.

In another embodiment, the preferred clothing style category information may be information on a preferred clothing style selected by the user through a preferred clothing style selection interface provided to the user on the online shopping mall.

For example, the preferred clothing style category information can be obtained by the user at least one of Simple Basic, SPA/Chipchic, Contemporary, Formal/Career, Traditional, Unique, Trendy, Street, Luxury, Sports, Girly, Morentic, and Modern. The clothes style may be information selected through a preferred clothes style selection interface.

At this time, the determined at least two or more clothing style categories may be classified into a main clothing style category and a sub clothing style category.

For example, when the top two clothing style categories with high counting values are 10 casual categories and 2 maximalism categories, the shopping mall server 100 serves as a main clothing style category and a maximalism category. They can be classified into clothing style categories.

Also, the preferred size information may be information on a preferred clothes size input and/or selected by the user through a preferred clothes size selection interface provided to the user through an online shopping mall.

For example, the preferred size information may be information that the user selects any one of sizes 44 or less, 55 sizes, 66 sizes, and 77 sizes or more through the preferred clothes size selection interface.

The shopping mall server 100, which has obtained at least one of the user's body type information, preferred fit numerical information, preferred clothing style information, and preferred size information, is Purchase preference information can be generated, and the generated purchase preference information can be converted into a database and managed.

In this way, the shopping mall server 100 may generate purchase preference information based on the body shape and purchase history data of the online shopping mall user and use it later, thereby more effectively providing a customized product recommendation service based on a clear basis.

Meanwhile, the shopping mall server 100 may detect purchase preference information of the first user from the database 140 when the first user accesses the online shopping mall site. (S105)

In detail, the shopping mall server 100 may extract the purchase preference information of the first user from the purchase preference database 140 for each user constructed based on the purchase preference information for each user.

In addition, the shopping mall server 100 may select another user (ie, similar user) having purchase taste information similar to the detected purchase taste information of the first user. (S107)

In detail, the shopping mall server 100 may extract purchase preference information of the first user and purchase preference information of other users, which are at least one other user except for the first user, from the purchase preference information database 142 for each user. .

In addition, the shopping mall server 100 may perform deep learning by putting the extracted purchase preference information of the first user and other users into the purchase preference similarity output neural network 230.

Thereafter, the shopping mall server 100 may obtain a purchase preference similarity with a first user for each of at least one other user derived as a result of deep learning from the purchase preference similarity output neural network 230.

In this case, the degree of similarity of purchase preference may be provided by being displayed as a percentage (%) value according to the degree of similarity.

In addition, the shopping mall server 100 may designate at least one or more other users as similar users based on the acquired similarity of purchase preferences with the first user for each other user.

In detail, the shopping mall server 100 may determine at least one or more other users having a high similarity in purchase preferences with the first user according to a predetermined criterion (eg, when the purchase preference similarity is greater than or equal to a predetermined percentage value), and Other users can be set as similar users.

However, in the case of selecting a similar user in the above manner, effective data processing may be difficult because all a plurality of other users must be deep-learned to designate a similar user of the first user.

In order to solve this problem, the shopping mall server 100 in an embodiment of the present invention may perform data processing for selecting a similar user as shown in FIG. 5.

5 is a flowchart illustrating a process of selecting a similar user of a first user according to an embodiment of the present invention in more detail.

Referring to FIG. 5, the shopping mall server 100 may first detect purchase preference information of a first user and purchase preference information of another user. (S201)

In detail, the shopping mall server 100 may extract purchase preference information of the first user and other users from the purchase preference database 140 for each user.

In addition, the shopping mall server 100 may compare the detected first user purchase taste information with other user purchase taste information. (S203)

In detail, the shopping mall server 100 may perform data processing for comparing the first user purchase taste information extracted through the data processing unit 120 with at least one other user purchase taste information.

In this case, the shopping mall server 100 may firstly extract at least one or more other users of a predetermined number or less having purchase preference information of the first user and purchase preference information having a high degree of similarity. (S205)

In detail, the shopping mall server 100 includes 1) a body type information category, 2) a preferred fit numerical category, and 3) a preferred clothing style category, which are each category (hereinafter, referred to as a purchase orientation category) included in the first detected purchase orientation information. Similarity between the first user and other users can be measured.

In addition, the shopping mall server 100 may derive at least one or more and other users less than a predetermined number of users having a high degree of similarity for each purchase preference category according to a preset criterion (eg, a case having a similarity of a predetermined percentage value or more) I can.

That is, the shopping mall server 100 includes: 1) another user having a high similarity to the first user in the body type information category, 2) another user having a high similarity to the first user in the preferred fit numerical category, and 3) a preferred clothing style category. The first user and other users having a high degree of similarity may be selected and extracted first.

For example, when the value of the first user's preferred fit value is 50, the shopping mall server 100 has a similarity of 50% or more to the first user's preferred fit value among the values of the preferred fit value of other users. At least one or more other users having a fit value and less than a predetermined number may be extracted.

Subsequently, the shopping mall server 100 may acquire purchase taste similarity by putting the first extracted purchase taste information of another user and the purchase taste information of the first user into the deep learning neural network. (S207)

In detail, the shopping mall server 100 may perform deep learning by putting the first extracted purchase preference information of another user and the purchase preference information of the first user into the purchase preference similarity output neural network 230.

Thereafter, the shopping mall server 100 may obtain a purchase taste similarity with the first user for each other user extracted at least one primary extracted as a result of deep learning from the purchase taste similarity output neural network 230.

The shopping mall server 100, which obtained the first extracted degree of purchase taste similarity with the first user for each other user, through the deep learning neural network, is at least one or more of the obtained purchase taste similarity higher than a predetermined standard. Other users can be secondarily extracted and selected as similar users. (S209)

That is, the shopping mall server 100 may finally extract other users with high similarity in purchase preferences above a predetermined standard (eg, a preset percentage value) and designate them as similar users.

In this way, the shopping mall server 100 performs the first data processing on its own and then proceeds with the data processing of the deep learning neural network to select similar users based on the first processed data, It is possible to reduce the load on data processing in the network and make the deep learning neural network operate more efficiently and quickly.

Returning to FIG. 4 again, the shopping mall server 100 selecting a similar user may extract a list of previously purchased products of the selected similar user and display it as a recommended product. (S109)

In detail, the shopping mall server 100 may detect a list of previously purchased products of a similar user selected from the database 140 and may provide the detected list of purchased products of a similar user as recommended products to the first user.

In this case, the shopping mall server 100 may list and display recommended products in order according to 1) the order of satisfaction of similar users with respect to recommended products, 2) the order of products recently purchased by similar users.

In addition, the shopping mall server 100 may generate and provide a product list for each user having a high purchase preference similarity including both a first user and a similar user.

Next, the shopping mall server 100 that displays the recommended product and provides it to the first user may generate interest level information of the first user with respect to the displayed recommended product. (S111)

Here, the degree of interest information is information capable of indicating the degree of interest in the recommended product of the first user, which is generated based on at least one of the number of times the first user selects the recommended product and the amount of purchase.

For example, the shopping mall server 100 is, the number of times the first user selects the first recommended product = 10, the purchase amount = 2, the number of times the first user selects the second recommended product = 5, the purchase amount = 1 day In this case, it may be determined that the degree of interest in the first recommended product is higher than the degree of interest in the second recommended product.

The shopping mall server 100 that has generated such interest level information may generate similar user detection base information that can be used when similar users are detected by the first user later, based on the generated interest level information of the first user. (S113)

In detail, the shopping mall server 100 may first detect a second user determined to have the highest interest level of the first user based on the interest level information of the first user.

In addition, the shopping mall server 100 may identify a purchase preference category belonging to the time when the detected second user is first extracted by the data processing unit 120 as another user having a purchase preference similar to the first user.

That is, the shopping mall server 100 may determine which of the plurality of purchase preference categories the category having the highest similarity between the first user and the second user is a body type information category, a fit value information category, and a clothes style category.

In addition, the shopping mall server 100 may select a preferred purchase preference category of the first user from the determined purchase preference category at the time of the first extraction of the second user.

For example, the shopping mall server 100 may set the preferred fit numerical category to the first user's purchase preference category when the detected second user was first extracted by the data processing unit 120 You can select a preferred purchase preference category.

In this way, the shopping mall server 100 that selects the preferred purchase preference category of the first user stores the selected preferred purchase preference category information by matching with the first user, and then detects a similar user for the first user, the first Data processing may be performed by assigning weights to preferred purchase preference categories that are matched with the user and stored.

For example, when the shopping mall server 100 detects a similar user of the first user later, the shopping mall server 100 selects a similar user of the first user to give a weight to the preferred purchase preference category matched with the first user. In the first extraction, the proportion of other users can be increased by extracting more than twice as many other users as compared to the rest of the purchase preference categories from the corresponding preferred purchase preference category.

In this way, the shopping mall server 100 uses the first user's preferred purchase preference category information based on the first user's interest level information as feedback information and uses it as detection base information when selecting a similar user of the first user later. By doing so, it is possible to perform high-level data processing taking into account not only artificial intelligence judgment but also cognitive judgment of the user through the deep learning neural network, and through this, a service providing customized recommended products can be implemented with a higher degree of completion.

Meanwhile, the shopping mall server 100 provides the first user with a list of previously purchased products of the selected similar user as recommended products, and then provides a follow application function for at least one or more of the selected similar users. Can be provided to. (S115)

In detail, the shopping mall server 100 may provide the first user with a user interface through which at least one or more similar users can apply for follow.

And the shopping mall server 100, when a following user, who is a followed user, purchases a product in an online shopping mall in the future, an alarm notifying that the following user has purchased a product, information about the product, and The following user purchase information including product reviews and wearing shots of the corresponding following user may be provided to a first user (ie, a follower user) who is a follower.

That is, the shopping mall server 100 matches and manages similar users that the first user wants to constantly be interested in with the first user, and provides the following user purchase information for the similar user to the first user in real time. , A more differentiated user-customized product recommendation service can be implemented.

In addition, the shopping mall server 100 is a discount means that can be used when purchasing a product in an online shopping mall to a following user who has a predetermined number or more (eg, 10% or more of the total online shopping mall users). (Eg, discount coupons) may be provided.

That is, the shopping mall server 100 can encourage users to more actively perform activities through the online shopping mall by providing compensation according to the compensation for users who spread the influence on product purchase to a plurality of users. Operational competitiveness can be improved and sales can be improved.

As described above, the method for providing a product recommendation service based on a deep learning neural network according to an embodiment of the present invention provides a service for recommending a customized product using a deep learning neural network based on the body shape and purchase history data of an online shopping mall user. , It is possible to increase competitiveness by providing differentiated services of online shopping malls, enable users to use the online shopping mall more efficiently, and create an effect of increasing sales through user-customized recommended products.

In addition, the method for providing a product recommendation service based on a deep learning neural network according to an embodiment of the present invention builds its own database 140 based on the body shape and purchase history data of a plurality of users, and builds its own database 140. ), it is possible to provide a faster and more efficient product recommendation service by performing deep learning through a deep learning neural network.

Moreover, the method for providing a product recommendation service based on a deep learning neural network according to an embodiment of the present invention uses a deep learning neural network to generate data necessary for a product recommendation service of an online shopping mall, thereby generating a vast amount of data by artificial intelligence. It can be effectively analyzed and classified or clustered.

In addition, the method for providing a product recommendation service based on a deep learning neural network according to an embodiment of the present invention generates and utilizes user's feedback information on the recommended product, so that not only artificial intelligence judgment through the deep learning neural network but also the user's perception It is possible to implement a service that provides customized recommended products by considering appropriate judgment.

In addition, the method for providing a product recommendation service based on a deep learning neural network according to an embodiment of the present invention provides a reward for users who spread the influence on product purchase to a plurality of users, so that users are more actively online. There is an effect of encouraging to perform activities through shopping malls, improving shopping mall operation competitiveness, and improving sales.

The embodiments according to the present invention described above may be implemented in the form of program instructions that can be executed through various computer components and recorded in a computer-readable recording medium. The computer-readable recording medium may include program instructions, data files, data structures, etc. alone or in combination. The program instructions recorded in the computer-readable recording medium may be specially designed and configured for the present invention or may be known and usable to those skilled in the computer software field. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tapes, optical recording media such as CD-ROMs and DVDs, and magnetic-optical media such as floptical disks. medium), and a hardware device specially configured to store and execute program instructions, such as ROM, RAM, flash memory, and the like. Examples of the program instructions include not only machine language codes such as those produced by a compiler but also high-level language codes that can be executed by a computer using an interpreter or the like. The hardware device can be changed to one or more software modules to perform the processing according to the present invention, and vice versa.

The specific implementations described in the present invention are examples, and do not limit the scope of the present invention in any way. For brevity of the specification, descriptions of conventional electronic configurations, control systems, software, and other functional aspects of the systems may be omitted. In addition, the connection or connection members of the lines between the components shown in the drawings exemplarily represent functional connections and/or physical or circuit connections, and in an actual device, various functional connections that can be replaced or additionally It may be referred to as a connection, or circuit connections. In addition, if there is no specific mention such as “essential” or “importantly”, it may not be an essential component for the application of the present invention.

In addition, although the detailed description of the present invention has been described with reference to a preferred embodiment of the present invention, the spirit of the present invention described in the claims to be described later if one skilled in the art or those of ordinary skill in the art And it will be appreciated that various modifications and changes can be made to the present invention within a range not departing from the technical field. Therefore, the technical scope of the present invention should not be limited to the content described in the detailed description of the specification, but should be determined by the claims.

Claims (15)

As a method for an online shopping mall server to provide customized recommended products based on a deep learning neural network,
Acquiring product information of products posted on the online shopping mall;
Acquiring purchase preference information of the online shopping mall users based on the product information, user's body shape information, and purchase history information to establish a purchase preference information database;
Detecting purchase preference information of the user when the user accesses the online shopping mall;
Comparing the detected purchase taste information of the user with the purchase taste information of another user to calculate a purchase taste similarity;
Selecting the other user whose calculated purchase taste similarity is greater than or equal to a predetermined standard as a similar user of the user; And
Extracting a list of previously purchased products of the selected similar user and providing the user as the recommended product,
The generating product information of the product posted on the online shopping mall may include obtaining fit numerical information of the product through the deep learning neural network, and obtaining clothing style category information of the product through the deep learning neural network. Including the step of obtaining,
The fit numerical information refers to information expressed by converting the degree of fit of clothes from tight-size fit to over-sized fit into numerical values, and the clothing style category information is clothes preset based on previously learned data. It is a classification criterion, and the product information includes numerical information, image information, fit numerical information, and clothing style category information of the product,
The step of constructing the purchase taste database includes obtaining information on the user's preferred fit value and information on a preferred clothing style category,
Computing the purchase taste similarity by comparing the detected purchase taste information of the user with the purchase taste information of another user may include, in the body shape information category, the preferred fit value category, and the preferred clothing style category included in the purchase taste information. Calculating the degree of similarity for each category between the user and the other user, and first extracting at least one or more other users in the order of the highest average of the calculated similarities for each category, and the first extraction of the other users. Including the step of inputting the purchase taste information and the purchase taste information of the user into the deep learning neural network and outputting the similarity of the purchase taste with the user for each of at least one other user,
Generating interest level information indicating the level of interest of the user with respect to the provided recommended product,
Further comprising the step of utilizing the generated interest level information when the user selects the similar user,
Utilizing the generated interest level information when the user selects the similar user includes: detecting a first user with the highest level of interest of the user, and a first user having the highest similarity by category between the user and the first user. 1 The step of determining a category, and assigning a weight to the first category to first select at least one or more other users in the order of the highest similarity average by category for the user according to the similarity by category between the user and the other user Extracting, and selecting a similar user again based on the similarity of the purchase preference between the first extracted other user and the user,
Further comprising the step of providing a user interface for performing a follow application function for the selected similar user to the user,
Providing a user interface for performing a follow application function for the selected similar user to the user includes: following a similar user according to at least one of the user's similar users, and following a user who is followed. When a user purchases the product at the online shopping mall, an alarm notifying that the following user has purchased the product, information on the product, a following user purchase including a product review and a wearing shot of the following user for the product Providing the information to a follower user who is a user who followed the information.
A method of providing product recommendation service based on deep learning neural networks. delete delete delete The method of claim 1,
The preferred fit numerical information,
A random number obtained based on the average value of the fit value information of at least one product previously purchased by the user in the online shopping mall.
A method of providing product recommendation service based on deep learning neural networks. The method of claim 5,
The preferred clothing style category information above,
Information for selecting at least one uppermost clothing style category according to the counted value by counting the clothing style categories to which the at least one or more products previously purchased by the user in the online shopping mall are classified and belonged to each
A method of providing product recommendation service based on deep learning neural networks. delete delete delete delete The method of claim 1,
The above interest level information,
Generated based on at least one or more of the number of times the user has selected and purchased the recommended product
A method of providing product recommendation service based on deep learning neural networks. delete delete delete The method of claim 1,
Providing a discount means that can be used when purchasing the product in the online shopping mall to the following user having a predetermined number or more of the follower users.
A method of providing product recommendation service based on deep learning neural networks.

Images