KR102502135B1 - Computing apparatus for executing an artificial intelligence planner recommending consumption events - Google Patents

Abstract

According to one aspect of the present invention, provided is a computing device running an artificial intelligence planner which recommends a buying event using time-series consumption patterns of multiple users. The computing device includes: a memory for storing instructions for executing an artificial intelligence planner; a processor for generating output information for each of the multiple users; and a communication unit for transmitting the output information to a plurality of user terminals.

Description

A computing device that runs an artificial intelligence planner that recommends purchase events

The following description is an invention related to a computing device that executes an artificial intelligence planner that recommends purchase events, and in particular, learns an artificial neural network with a plurality of consumption patterns derived from the start of dating of users, and predicts purchases from the learned artificial neural network. The present invention relates to a device for recommending events customized to individual users.

Today, the culture of non-marriage living without marriage is rapidly spreading among young people in their 20s and 30s due to various factors such as a rapid rise in real estate prices and economic burdens such as childbirth. In addition, in the conventional marriage market, there is a culture in which information on marriage is obtained through a service job called a wedding planner in the consumption process related to marriage, and discount events are provided by being introduced to studios, makeup, and wedding dress companies that collaborate with a specific wedding planner. existed. However, in a reality where the marriage population is rapidly decreasing, the job of a wedding planner, which is provided based on people as in the prior art, is rapidly declining in efficiency.

The need for artificial intelligence-based computing devices that efficiently provide information on various purchase events, such as accessory purchases, travel reservations, and fashion product purchases within the large category of marriage, by receiving consumption patterns of people who share various values as input, is increasing. are doing

Korean Patent Registration No. 10-2334605 discloses a method of receiving comparative estimates from a plurality of service providers according to a request based on user information, and recommending information reflecting the additional points given to each service provider to the user. . However, the target patent does not disclose any content about a technology for automatically outputting information on events commonly applicable to various users, such as couples who are promised marriage or couples who are still in love.

According to one aspect, a computing device for executing an artificial intelligence planner that recommends purchase events using time-series consumption patterns of a plurality of users is provided. The computing device includes a memory for storing instructions for executing the artificial intelligence planner, a processor for generating output information for each of the plurality of users, and a communication unit for transmitting the output information to a plurality of user terminals. can do.

According to an embodiment, the processor uses pre-obtained time information from a dating start point of each of a plurality of consumers as input data, and uses consumption amount information and consumption item information for dating or marriage as output data Learning a supervised artificial neural network, inputting time information from the start of dating of each of a plurality of users to the learned artificial neural network, whereby the expected consumption vector of the plurality of users - the expected consumption vector for each consumption item Including expected consumption amount information corresponding to the information as an element, obtaining -, assigning a user to one of m arbitrary user groups - where m is a natural number - using the expected consumption vectors of the plurality of users. and outputting a purchase event having a correlation coefficient greater than or equal to a threshold value with a matching user group to which the user is assigned.

According to another embodiment, the assigning of the user may include comparing the center point of each cluster of the arbitrary m number of user groups with a Euclidean distance of an expected consumption vector of the user, thereby assigning the user to a matching user group. It may include the step of allocating to.

According to another embodiment, in the m user groups, the computing device allocates a plurality of users to m random clusters according to respective expected consumption vectors, and based on the center point of each of the m clusters. to reallocate each expected consumption vector to one of the m clusters, recalculate the center points of each of the m clusters to which the expected consumption vectors are reallocated, and determine whether the reallocation of each expected consumption vector is repeated. can be created by determining

According to another embodiment, the instruction may include selecting an i th cluster from among the m clusters, and when a silhouette value calculated for the i th cluster is equal to or greater than a predetermined threshold, each of the plurality of users It may further include stopping repetition of the reallocation step of the expected consumption vector of .

According to another embodiment, the outputting of the purchase event may include outputting information about a consumption item having a correlation coefficient equal to or greater than a threshold based on an average value of expected consumption amounts within each matching user group. .

1A is a diagram for explaining a deep learning operation method using an artificial neural network.
1B is a diagram for describing a computing device that executes an artificial intelligence planner that recommends a purchase event, according to an exemplary embodiment.
2 is a diagram for explaining a learning process of an artificial intelligence planner according to an embodiment.
3 is a flowchart illustrating a method of recommending a purchase event using an artificial neural network according to an exemplary embodiment.
4 is a flowchart illustrating a process of clustering users based on an expected consumption vector according to an embodiment.
5 is an exemplary diagram illustrating a process of clustering users based on an expected consumption vector according to an embodiment.

Specific structural or functional descriptions of the embodiments are disclosed for illustrative purposes only, and may be modified and implemented in various forms. Therefore, the embodiments are not limited to the specific disclosed form, and the scope of the present specification includes changes, equivalents, or substitutes included in the technical spirit.

Although terms such as first or second may be used to describe various components, such terms should only be construed for the purpose of distinguishing one component from another. For example, a first element may be termed a second element, and similarly, a second element may be termed a first element.

It should be understood that when an element is referred to as being “connected” to another element, it may be directly connected or connected to the other element, but other elements may exist in the middle.

Singular expressions include plural expressions unless the context clearly dictates otherwise. In this specification, terms such as "comprise" or "have" are intended to designate that the described features, numbers, steps, operations, components, parts, or combinations thereof exist, but one or more other features, numbers, or steps However, it should be understood that it does not preclude the possibility of existence or addition of operations, components, parts, or combinations thereof.

Unless defined otherwise, all terms used herein, including technical or scientific terms, have the same meaning as commonly understood by one of ordinary skill in the art. Terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with the meaning in the context of the related art, and unless explicitly defined in this specification, it should not be interpreted in an ideal or excessively formal meaning. don't

Hereinafter, embodiments will be described in detail with reference to the accompanying drawings. In the description with reference to the accompanying drawings, the same reference numerals are given to the same components regardless of reference numerals, and overlapping descriptions thereof will be omitted.

1A is a diagram for explaining a deep learning operation method using an artificial neural network.

Artificial intelligence (AI) algorithms including deep learning, etc. input input data (10) to an artificial neural network (ANN), and learn output data (30) through operations such as convolution And, features can be extracted using the learned artificial neural network. An artificial neural network may refer to a computer scientific architecture modeling a biological brain. In an artificial neural network, nodes corresponding to brain neurons are connected to each other and collectively operate to process input data. Examples of various types of neural networks include Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs), Deep Belief Networks (DBNs), and Restricted Boltzman Machines (Restricted Boltzman Machines). Machine, RBM) method, etc., but is not limited thereto. In a feed-forward neural network, neurons in the neural network have links to other neurons. Such connections may extend through the neural network in one direction, for example in a forward direction.

1A shows the structure of an artificial neural network (eg, a Convolution Neural Network (CNN) 20) that receives input data 10 and outputs output data 30. The artificial neural network may be a deep neural network having two or more layers.

A convolutional neural network 20 may be used to extract “features” such as borders, line colors, etc. from input data 10 . The convolutional neural network 20 may include a plurality of layers. Each layer may receive data, and may process data input to the corresponding layer to generate data output from the corresponding layer. Data output from the layer may be a feature map generated by convolving an image or a feature map input to the convolutional neural network 20 with a filter weight value. Initial layers of convolutional neural network 20 may be operated to extract low-level features such as edges or gradients from the input. Subsequent layers of the convolutional neural network 20 may extract progressively more complex features such as eyes, nose, etc. in the image.

1B is a diagram for describing a computing device that executes an artificial intelligence planner that recommends a purchase event, according to an exemplary embodiment.

According to an embodiment, the computing device may receive time information from when a plurality of users start dating, and output expected consumption vectors of the plurality of users. In addition, the computing device may transmit various promotional information for each matching user group clustered according to the expected consumption vector as a purchase event.

Specifically, the expected consumption vector may represent a vector including expected consumption amount information corresponding to each consumption item information as an element. For example, if there is a couple who has been dating for 100 days, the computing device provides expected consumption amount information for each expected consumption item (accessories, food and beverage, fashion, travel ??) (1,000,000 won, 250,000 won). Won , 1,500,000 won, 500,000 won , ??) can be output. Based on the above expected consumption vector, when the couple goes on a trip to commemorate the 100th day of dating, the possibility of spending around KRW 500,000 is the highest, and when a promotion of that amount is proposed, the most It contains information that can be effective.

Referring to FIG. 1B , an artificial intelligence planner learning device 100 and a computing device 150 executing the artificial intelligence planner are shown. The artificial intelligence planner learning apparatus 100 according to an embodiment provides various functions such as generating a neural network, training (or learning) a neural network, or retraining a neural network. Corresponds to a computing device having processing functions. For example, the artificial intelligence planner learning apparatus 100 may be implemented in various types of devices such as a personal computer (PC), a server device, and a mobile device.

The artificial intelligence planner learning apparatus 100 may generate a trained neural network 110 by repeatedly training (learning) a given initial neural network. Creating the trained neural network 110 may mean determining neural network parameters. Here, the parameters may include, for example, various types of data input/output to the neural network, such as input/output activations, weights, and biases of the neural network. As the iterative training of the neural network progresses, the parameters of the neural network can be tuned to compute a more accurate output for a given input.

The trained neural network 110 according to an embodiment may include a plurality of neural networks.

The artificial intelligence planner learning device 100 may transfer the trained neural network 110 to the computing device 150 that executes the artificial intelligence planner. The computing device 150 executing the artificial intelligence planner may be included in a mobile device, an embedded device, or the like. Also, as another embodiment, the computing device 150 executing the artificial intelligence planner may be dedicated hardware for driving the neural network.

The computing device 150 executing the artificial intelligence planner may drive the trained neural network 110 as it is or the neural network 160 in which the trained neural network 110 is activated (eg, quantized). there is. The computing device 150 that executes the artificial intelligence planner that drives the processed neural network 160 may be implemented in an independent device separate from the artificial intelligence planner learning apparatus 100 . However, it is not limited thereto, and the computing device 150 executing the AI planner may be implemented in the same device as the AI planner learning device 100 .

2 is a diagram for explaining a learning process of an artificial intelligence planner according to an embodiment. The contents described with reference to FIGS. 1A to 1B may be equally applied to FIG. 2 , and duplicated contents may be omitted.

Referring to FIG. 2 , the artificial intelligence planner 200 may learn an artificial neural network by using time-series consumption patterns 210 of a plurality of users as learning data. More specifically, the artificial intelligence planner 200 uses pre-obtained time information from the start of love for each of a plurality of consumers as input data, and consumes amount information and consumption item information for love or marriage as output data. It can be used for supervised learning. Specifically, the artificial intelligence planner 200 receives time information from the start of love for the user, and outputs an expected consumption vector for the user according to an expected probability value based on a pre-learned consumption pattern of a plurality of consumers. It may include a neural network 110 learned to do. Specifically, the expected consumption vector may include expected consumption amount information corresponding to each consumption item information as an element.

For example, if time information (eg, D+100, D+200 ??, etc.) from the start of dating is input in the process of learning the consumption pattern of a specific user A, the neural network 110 determines each consumption item (eg, accessories, food and beverage, fashion, travel, etc.) may be learned to output consumption amount information.

Learning data delivered to the neural network 110 included in the artificial intelligence planner 200 may be labeled as shown in Table 1 below.

time information consumption items consumption history consumption amount D+100 accessory buy a coupling KRW 1,000,000 D+200 food and drink wedding hall down payment KRW 1,000,000 D+230 accessory down payment 6,000,000 won D+270 fashion dress down payment KRW 2,000,000 D+290 travel bridal shower 500,000 won

The artificial intelligence planner 200 learned as described above may generate an output indicating which amount of expected consumption will be generated for a certain consumption item according to time information from each user's dating start point. Specifically, the artificial intelligence planner 200 horizontally receives time information 220 from the start of dating of a plurality of users and outputs an expected consumption vector 230, so that consumption for similar items at a specific time point occurs. The effect of vertically outputting potential customer groups with high potential as matching user groups can be expected.

3 is a flowchart illustrating a method of recommending a purchase event using an artificial neural network according to an exemplary embodiment. Referring to FIG. 3 , a method 300 of recommending a purchase event using an artificial neural network by a computing device includes obtaining an expected consumption vector for a user (310), the user's expected consumption vector and an arbitrary m (where, m is a natural number) allocating a user to one user group by comparing Euclidean distances of center points of the number of user groups (320) and outputting a purchase event having a correlation coefficient greater than or equal to a threshold value with a matching user group to which the user is assigned ( 330) may be included. Although not shown in FIG. 3, a computing device that executes an artificial intelligence planner that recommends a purchase event using an artificial neural network includes a memory storing instructions for executing the artificial intelligence planner, and a memory for each of the plurality of users. It may include a processor that generates output information about the output information, and a communication unit that transmits the output information to a plurality of user terminals.

In step 310, the computing device recommending the purchase event may receive time information from the start of dating from the user terminal. Specifically, the process of receiving time information from the start of dating may be performed by a communication unit included in the computing device. The communication unit may include a communication interface. For example, communication interfaces include WLAN (Wireless LAN), WiFi (Wireless Fidelity) Direct, DLNA (Digital Living Network Alliance), Wibro (Wireless broadband), Wimax (World Interoperability for Microwave Access), HSDPA (High Speed Downlink Packet Access), etc. Bluetooth™, Radio Frequency Identification (RFID), Infrared Data Association (IrDA), UWB (Ultra Wideband), ZigBee, NFC (Near Field Communication), etc. can In addition, the communication interface may represent any interface capable of communicating with the outside (eg, a wired interface).

In step 310, the computing device may acquire an expected consumption vector by inputting time information from the start of love for a specific user to the pre-learned artificial intelligence planner. For the learning process and operation process of the artificial intelligence planner, since the descriptions of FIGS. 1A, 1B, and 2 described above can be applied as they are, redundant descriptions will be omitted.

In step 320, the computing device may assign the user to one of m arbitrary user groups (where m is a natural number) using the expected consumption vector of the user. A process of generating a cluster for arbitrary m user groups by a computing device and allocating a specific user to one cluster will be described in detail with additional drawings.

In step 330, the computing device may output a purchase event having a correlation coefficient equal to or greater than a threshold value with a matching user group to which the user is assigned. Specifically, the computing device may output information about consumption items having a correlation coefficient equal to or greater than a threshold value based on an average value of expected consumption amounts within each matching user group. For example, when the expected consumption amount of 500,000 won for the accessory item of the first matching user group is calculated as an average value, the computing device transmits a purchase event in which a promotion in the “accessory” field corresponding to the same consumption item exists. can Specifically, the computing device may transmit discount information on rings of brand A and discount information on necklaces of brand B to a plurality of user terminals included in the first matching user group. More specifically, the computing device may output purchase event information on a consumption item associated with a specific matching user group based on an average value of expected consumption amounts within each matching user group.

4 is a flowchart illustrating a process of clustering users based on an expected consumption vector according to an embodiment. Referring to FIG. 4 , a method 400 for clustering users is shown. The clustering method 400 of users includes a step 410 of allocating a plurality of users to m clusters according to respective expected consumption vectors, and assigning each expected consumption vector to k clusters based on the center point of each of the m clusters. Re-assignment to one of (420), re-calculating the center points of each of the k clusters to which the expected consumption vector is reallocated (430), and determining whether to repeat the reallocation of the expected consumption vector (440). can include

In step 410, the computing device may obtain expected consumption vectors for a plurality of users. In addition, the computing device may allocate the expected consumption vector of the user to arbitrary m (where m is a natural number) number of clusters, and initialize center points corresponding to each of the m number of clusters. Exemplarily, the expected consumption vector may represent a vector including, as elements, an expected consumption amount of each of the user's consumption items based on a specific time point.

In step 420, the computing device may reallocate each expected consumption vector to one of the m clusters based on the center points of each of the m clusters. For example, the center points of each of the m clusters may represent the center points initialized in step 410 . The computing device may reallocate the expected consumption vectors of each of the plurality of users so that the expected consumption vectors are allocated to a cluster including the closest central point.

In step 430, the computing device may re-select the center point of each of the m clusters using expected consumption vectors newly reassigned to each of the m clusters.

Also, in step 440, the computing device may determine whether to repeat the step of reallocating the expected consumption vectors. Specifically, in step 440, the computing device may be determined according to a silhouette value corresponding to each of the m clusters. More specifically, the computing device may determine the silhouette value using Equation 1 below.

When the i-th cluster is selected among the m clusters, a(i) may represent an average value of dissimilarity of expected consumption vectors existing in the same i-th cluster as the first expected consumption vector. In addition, b(i) may represent a minimum value of an average dissimilarity of expected consumption vectors existing in a second cluster different from the first expected consumption vector. More specifically, each of a(i) and b(i) may be calculated based on a Euclidean distance. When the calculated silhouette value is greater than or equal to a predetermined threshold, the computing device determines that clustering of expected consumption vectors of users has been appropriately performed, and may stop repeating the reassignment step.

Accordingly, the computing device compares the center point of each cluster of m user groups with the Euclidean distance of the expected consumption vector of the user, and the user group to which the center point is closest to the expected consumption vector of the specific user. may be determined as a matching user group. Also, the computing device may assign a specific user group as a matching user group. Accordingly, the central point of the matching user group may be re-selected based on the principle described above.

5 is an exemplary diagram illustrating a process of clustering users based on an expected consumption vector according to an embodiment. Referring to FIG. 5 , a graph 500 in which a computing device clusters a plurality of users into three matching user groups 510 , 520 , and 530 is shown. The description of the clustering process of the computing device may be applied to the description described in conjunction with FIG. 4 above, so duplicate descriptions will be omitted.

In FIG. 5 , as an example, an expected consumption vector (blue circle) of users 100 days after starting dating and an expected consumption vector (green circle) of users 365 days after dating start are graphed. Samples that can be said to be in the early stages of love are analyzed for consumption patterns with similarities in the first consumption item, accessories, and the second consumption item, fashion, within 1.2 million won, and are clustered into one matching user group. In this case, the computing device may output purchase event information about a product related to accessories and fashion and worth around 1.2 million won for the first matching user group 510 . Specifically, the purchase event information may include brand information of a specific product, product information under discount, and discount period information.

On the other hand, different consumption patterns may be analyzed for the first consumption item and the second consumption item of the sample in which time information from the start of dating is about one year. In this case, consumption patterns having a similarity of about 3.3 million won or less in the field of accessories, which is the first consumption item, may be analyzed and clustered into the second matching user group 520 . For the second matching user group 520 , the computing device may output purchase event information about a product associated with an accessory and worth around 3.2 million won.

In addition, in the fashion field, which is the second consumption item, for a sample whose time information from the start of dating is about 1 year, consumption patterns having a similarity of about 4.8 million won can be analyzed and clustered into a third matching user group 530. there is. The third matching user group 530 may be presumed to have used a package product such as wedding dress custom-made or wedding dress rental. The computing device may output purchase event information on products of around 4.8 million won related to the fashion field including formal wear and wedding dresses to the third matching user group 530 .

The computing device according to the present embodiment integrates horizontal information about each user's romance using an artificial intelligence planner in which consumption patterns of a plurality of buyers are analyzed to provide vertical information (eg, matching user group) that can be used as a marketing element. By deriving it, the effect of increasing marketing efficiency in the non-face-to-face era can be expected. In addition, based on clustering through the expected consumption vector, it is possible to efficiently transmit information suitable for the expected consumption amount and consumption items of target customers, thereby reducing the amount of information transmission and reducing labor costs for hiring wedding planners. there is.

The embodiments described above may be implemented as hardware components, software components, and/or a combination of hardware components and software components. For example, the devices, methods and components described in the embodiments may include, for example, a processor, a controller, an arithmetic logic unit (ALU), a digital signal processor, a microcomputer, a field programmable gate (FPGA). array), programmable logic units (PLUs), microprocessors, or any other device capable of executing and responding to instructions. A processing device may run an operating system (OS) and one or more software applications running on the operating system. A processing device may also access, store, manipulate, process, and generate data in response to execution of software. For convenience of understanding, there are cases in which one processing device is used, but those skilled in the art will understand that the processing device includes a plurality of processing elements and/or a plurality of types of processing elements. It can be seen that it can include. For example, a processing device may include a plurality of processors or a processor and a controller. Other processing configurations are also possible, such as parallel processors.

Software may include a computer program, code, instructions, or a combination of one or more of the foregoing, which configures a processing device to operate as desired or processes independently or collectively. The device can be commanded. Software and/or data may be any tangible machine, component, physical device, virtual equipment, computer storage medium or device, intended to be interpreted by or provide instructions or data to a processing device. , or may be permanently or temporarily embodied in a transmitted signal wave. Software may be distributed on networked computer systems and stored or executed in a distributed manner. Software and data may be stored on one or more computer readable media.

The method according to the embodiment may be implemented in the form of program instructions that can be executed through various computer means and recorded on a computer readable medium. Computer readable media may include program instructions, data files, data structures, etc. alone or in combination. Program instructions recorded on a computer readable medium may be specially designed and configured for the embodiment or may be known and usable to those skilled in computer software. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks and magnetic tapes, optical media such as CD-ROMs and DVDs, and magnetic media such as floptical disks. - includes hardware devices specially configured to store and execute program instructions, such as magneto-optical media, and ROM, RAM, flash memory, and the like. Examples of program instructions include high-level language codes that can be executed by a computer using an interpreter, as well as machine language codes such as those produced by a compiler. The hardware devices described above may be configured to operate as one or more software modules to perform the operations of the embodiments, and vice versa.

As described above, although the embodiments have been described with limited drawings, those skilled in the art can apply various technical modifications and variations based on the above. For example, the described techniques may be performed in an order different from the method described, and/or components of the described system, structure, device, circuit, etc. may be combined or combined in a different form than the method described, or other components may be used. Or even if it is replaced or substituted by equivalents, appropriate results can be achieved.

Claims (5)

A computing device that executes an artificial intelligence planner that recommends a purchase event using time-series consumption patterns of a plurality of users,
a memory for storing instructions for executing the artificial intelligence planner;
a processor generating output information for each of the plurality of users; and
Communication unit for transmitting the output information to a plurality of user terminals
including,
The processor;
Learning an artificial neural network that is supervised by using pre-obtained time information from the start of dating of each of a plurality of consumers as input data, and using consumption amount information and consumption item information for dating or marriage as output data. ;
The expected consumption vector of the plurality of users for a specific time point corresponding to the time information by inputting the time information labeled to increase in time series order from the start of dating of each of the plurality of users as learning data to the learned artificial neural network. obtaining, as an element, the expected consumption vector includes, as an element, information on an expected consumption amount corresponding to each item of consumption information based on a specific time point;
allocating a user to one of m user groups, where m is a natural number, using the expected consumption vectors of the plurality of users; and
Outputting a purchase event having a correlation coefficient greater than or equal to a threshold value with a matching user group to which the user is assigned
run,
In the step of outputting the purchase event,
Outputting information on consumption items having a correlation coefficient equal to or greater than a threshold value based on the average value of expected consumption amounts within each matching user group
including,
The information on the consumption item may include a purchase event in which a promotion exists for products of a plurality of brands related to the consumption item,
The step of allocating the user is,
Allocating the user to a matching user group by comparing the center point of each cluster of the arbitrary m number of user groups and the Euclidean distance of the expected consumption vector of the user.
include,
The arbitrary m user groups,
The computing device allocates a plurality of users to arbitrary m clusters according to respective expected consumption vectors, and reassigns each expected consumption vector to one of the m clusters based on the center point of each of the m clusters. and recalculating the center points of each of the m clusters to which the expected consumption vectors are reallocated, and determining whether or not to repeat the reallocation of each expected consumption vector,
The instruction is
selecting an arbitrary ith cluster from among the m clusters; and
When the silhouette value calculated using Equation 1 for the i-th cluster is equal to or greater than a predetermined threshold, stopping repetition of the reassignment step of the expected consumption vector of each of the plurality of users.
Including more,
Equation 1 above is

, a (i) represents the average value of dissimilarity of expected consumption vectors existing in the i-th cluster identical to the first expected consumption vector, and b (i) is in a second cluster different from the first expected consumption vector. A computing device, characterized in that it represents the minimum value of the mean of the dissimilarity of the expected consumption vectors that exist. delete delete delete delete

Images