KR102211884B1 - Method for providing skin care cosmetics recommendation serivce using bigdata based ai machine learning - Google Patents

Abstract

A method of providing skin care recommendation service using big data-based artificial intelligence machine learning is provided, and a Baumann skin type questionnaire (BSTQ) is transmitted to the user terminal, and the result of the survey on the Baumann skin type questionnaire from the user terminal Receiving data to diagnose skin type, collecting product data including at least one or a combination of at least one of basic care products, sun care products, cleansing products, and makeup products used by the user from the user terminal, user terminal Collecting environmental data including gender, age, weight, lifecycle, diet, and occupation from the data, and parsing the diagnosed skin type, product data, and environmental data as input values, skin type, product, and environment The step of deriving and classifying the identifier of, and extracting reference data having a preset similarity from pre-built big data based on the classified identifier, and extracting the product containing the recommended component previously mapped to the reference data and stored. It includes the step of recommending to the terminal.

Description

Method of providing skin care recommendation service using big data-based artificial intelligence machine learning {METHOD FOR PROVIDING SKIN CARE COSMETICS RECOMMENDATION SERIVCE USING BIGDATA BASED AI MACHINE LEARNING}

The present invention relates to a method for providing a skin care recommendation service using big data-based artificial intelligence machine learning, which learns patterns inherent in skin conditions, products used, and environmental information through unsupervised learning and recommends products based on big data. Provides a way.

Cosmetics are consumer goods that are used every day in the real life of ordinary consumers, and are products in which users' reactions appear immediately. Such contents can also be found on social networks due to the spread of mobile devices and the development of the Internet environment, and can be analyzed with big data through the collection of corresponding data, but human skin has a difference in condition for each individual. Depending on the condition of the skin, skin problems are also different. Therefore, in 2016, as the Ministry of Health and Welfare allowed the general public to perform genetic testing through private institutions, users with various skin concerns are activating the genetic testing in order to use cosmetics that are more suitable for them, and skin diagnosis along with big data And cosmetics recommendation platform.

At this time, the method of recommending cosmetics using big data is performed by analyzing trends. In this regard, in Korean Patent Application Publication No. 2018-0074323 (published on July 3, 2018), data related to cosmetics are collected by keywords for each type of cosmetics by accessing SNS (Social Network Service) and the web. , By connecting to a public institution's server to collect cosmetic-related standard data on public data, and storing big data by configuring cosmetic-related data collected by the data collection unit, standard data, and data collected through public APIs into multiple categories. , A configuration is disclosed that analyzes natural language with big data to analyze customer trends for cosmetics, and verifies and merges the analysis results using cosmetics-related data on public data.

However, even if the above-described big data is used, human skin conditions are different for each individual, and skin concerns are different depending on the individual skin condition, and general consumers' preferences for using cosmetics are my own and my skin. Since preferences appear in the order of matching cosmetics and detailed counseling, it is not very likely that the status of a group clustered into the same group is an individual status unless an accurate analysis is made, so the recommendation service using only big data is less accurate. At this time, since genetic testing is also available in private institutions, genetic analysis of skin is also active as described above.According to the situation, personal skin type and genetic information are considered and data is collected from social networks, and customized recommendations through big data analysis Service development is required.

An embodiment of the present invention collects and stores ingredients included in cosmetics, diagnoses an individual's skin condition through a Baumann skin type questionnaire (BSTQ), and unsupervised learning of the skin condition as an input value. Conduct (Unsupervised Learning) to learn patterns, characteristics, and structures inherent in the skin condition, and generate a query with reference data based on the learned result using pre-built big data, and output recommended cosmetics and management methods as the result. Therefore, a method of providing skin care recommendation service using artificial intelligence machine learning based on big data that can accurately diagnose the user's skin condition through machine learning and generate more accurate recommendation results by searching big data as a query for the learning results. Can provide. However, the technical problem to be achieved by the present embodiment is not limited to the technical problem as described above, and other technical problems may exist.

As a technical means for achieving the above-described technical problem, an embodiment of the present invention transmits a Baumann skin type questionnaire (BSTQ) to a user terminal, and surveys a Baumann skin type questionnaire from the user terminal. Receiving result data to diagnose skin type, collecting product data including at least one or a combination of at least one of the basic care products, sun care products, cleansing products, and makeup products used by the user from the user terminal, the user Collecting environmental data including gender, age, weight, lifecycle, diet, and occupation from the terminal, and parsing the diagnosed skin type, product data, and environmental data as input values, such as skin type, product, and Deriving and classifying the identifier of the environment, and extracting reference data having a preset similarity from the big data previously established based on the classified identifier, and extracting a product containing recommended components previously mapped to the reference data and stored. And recommending to a user terminal.

According to any one of the above-described problem solving means of the present invention, ingredients included in cosmetics are collected and stored, and an individual's skin condition is diagnosed through a Baumann skin type questionnaire (BSTQ), and the skin condition is evaluated. Unsupervised learning is performed as an input value to learn patterns, characteristics, and structures inherent in the skin condition, and a query is generated using reference data based on the learned result using pre-built big data, and the result is recommended. By outputting cosmetics and management methods, it is possible to accurately diagnose the user's skin condition by machine learning, and to generate more accurate recommendation results by searching big data by querying the learning results, and when there is negative feedback. , Reinforcement Learning can be performed to minimize analysis errors.

1 is a view for explaining a system for providing a skin care recommendation service using artificial intelligence machine learning based on big data according to an embodiment of the present invention.
FIG. 2 is a block diagram illustrating a server for providing a skin care recommendation service included in the system of FIG. 1.
3 is a view for explaining an embodiment in which a skin care recommendation service using artificial intelligence machine learning based on big data is implemented according to an embodiment of the present invention.
4 is a diagram illustrating a process of transmitting and receiving data between components included in the system for providing a skin care recommendation service using artificial intelligence machine learning based on big data of FIG. 1 according to an embodiment of the present invention.
5 is a flowchart illustrating a method of providing a skin care recommendation service using artificial intelligence machine learning based on big data according to an embodiment of the present invention.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings so that those of ordinary skill in the art can easily implement the present invention. However, the present invention may be implemented in various different forms and is not limited to the embodiments described herein. In the drawings, parts irrelevant to the description are omitted in order to clearly describe the present invention, and similar reference numerals are assigned to similar parts throughout the specification.

Throughout the specification, when a part is said to be "connected" to another part, this includes not only "directly connected" but also "electrically connected" with another element interposed therebetween. . In addition, when a part "includes" a certain component, it means that other components may be further included, and one or more other features, not excluding other components, unless specifically stated to the contrary. It is to be understood that it does not preclude the presence or addition of any number, step, action, component, part, or combination thereof.

The terms "about", "substantially" and the like, as used throughout the specification, are used in or close to the numerical value when manufacturing and material tolerances specific to the stated meaning are presented, and are used in the sense of the present invention. To assist, accurate or absolute figures are used to prevent unfair use of the stated disclosure by unscrupulous infringers. As used throughout the specification of the present invention, the term "step (to)" or "step of" does not mean "step for".

In the present specification, the term "unit" includes a unit realized by hardware, a unit realized by software, and a unit realized using both. Further, one unit may be realized using two or more hardware, or two or more units may be realized using one hardware.

In this specification, some of the operations or functions described as being performed by the terminal, device, or device may be performed instead in a server connected to the terminal, device, or device. Likewise, some of the operations or functions described as being performed by the server may also be performed by a terminal, device, or device connected to the server.

In this specification, some of the operations or functions described as mapping or matching with the terminal means mapping or matching the unique number of the terminal or the identification information of the individual, which is the identification information of the terminal. Can be interpreted as.

Hereinafter, the present invention will be described in detail with reference to the accompanying drawings.

1 is a view for explaining a system for providing a skin care recommendation service using artificial intelligence machine learning based on big data according to an embodiment of the present invention. Referring to FIG. 1, a skin care recommendation service providing system 1 using big data-based artificial intelligence machine learning includes at least one user terminal 100, a skin care recommendation service providing server 300, and at least one piece of information. It may include a server 400. However, since the system 1 for providing a skin care recommendation service using artificial intelligence machine learning based on big data of FIG. 1 is only an embodiment of the present invention, the present invention is not limitedly interpreted through FIG. 1.

In this case, each component of FIG. 1 is generally connected through a network 200. For example, as shown in FIG. 1, at least one user terminal 100 may be connected to a skin care recommendation service providing server 300 through a network 200. In addition, the skin care recommendation service providing server 300 may be connected to at least one user terminal 100 and at least one information providing server 400 through the network 200. In addition, the at least one information providing server 400 may be connected to the skin care recommendation service providing server 300 through the network 200.

Here, the network refers to a connection structure in which information exchange is possible between respective nodes such as a plurality of terminals and servers, and examples of such networks include RF, 3rd Generation Partnership Project (3GPP) network, and Long Term (LTE). Evolution) network, 5GPP (5rd Generation Partnership Project) 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, NFC network, satellite broadcasting network, analog broadcasting network, Digital Multimedia Broadcasting (DMB) network, and the like, but are not limited thereto.

In the following, the term “at least one” is defined as a term including the singular number and the plural number, and even if the term “at least one” does not exist, each component may exist in the singular or plural, and may mean the singular or plural. It will be self-evident. In addition, it will be possible to change according to the embodiment that each component is provided in a singular or plural.

At least one user terminal 100 is a terminal that receives skin diagnosis and recommendations for cosmetics or skin care suitable for itself by using a web page, app page, program or application related to a skin care recommendation service using artificial intelligence machine learning based on big data. Can be In this case, the at least one user terminal 100 may be a terminal that receives a Baumann skin type questionnaire (BSTQ) from the skin care recommendation service providing server 300 and transmits a response thereto. Here, the at least one user terminal 100 may be a terminal that transmits the genetic test result to the skin care recommendation service providing server 300 when there is a genetic test result. In addition, the at least one user terminal 100 may be a terminal that divides and inputs cosmetics to be used for each purpose, and transmits environmental data including body data and occupation data to the skin care recommendation service providing server 300. Accordingly, the at least one user terminal 100 may be a terminal capable of receiving a recommendation from the skin care recommendation service providing server 300 for an accurate diagnosis and product or management suitable for one's skin type based on big data.

Here, the at least one user terminal 100 may be implemented as a computer capable of accessing a remote server or terminal through a network. Here, the computer may include, for example, a navigation system, a notebook equipped with a web browser, a desktop, a laptop, and the like. In this case, the at least one user terminal 100 may be implemented as a terminal capable of accessing a remote server or terminal through a network. At least one user terminal 100, for example, as a wireless communication device that is guaranteed portability and mobility, navigation, PCS (Personal Communication System), GSM (Global System for Mobile communications), PDC (Personal Digital Cellular), PHS(Personal Handyphone System), PDA(Personal Digital Assistant), IMT(International Mobile Telecommunication)-2000, CDMA(Code Division Multiple Access)-2000, W-CDMA(W-Code Division Multiple Access), Wibro(Wireless Broadband Internet) ) All types of handheld-based wireless communication devices such as terminals, smartphones, smartpads, and tablet PCs may be included.

The skin care recommendation service providing server 300 may be a server that provides a skin care recommendation service web page, an app page, a program, or an application using big data-based artificial intelligence machine learning. In addition, the skin care recommendation service providing server 300 collects raw data including public data, weather data, ingredient data, cosmetics data, and social media data from at least one information providing server 400, and , It may be a server that builds big data based on the collected data. In addition, the skin care recommendation service providing server 300 may be a server for learning by relating a skin type as a result of skin diagnosis to a product, an ingredient, and the like. In addition, the skin care recommendation service providing server 300 transmits a Bowman skin type questionnaire to at least one user terminal 100, receives a response from at least one user terminal 100, and receives at least one user terminal 100. ), based on the skin type and genetic test, machine learns the patterns, characteristics, and structures inherent in the user's data through unsupervised learning, extracts reference data for this, and generates a query. It may be a server that inputs the query into big data and extracts recommendation data, which is a result, and transmits it to the user terminal 100. To this end, skin type, genetic test, product information, etc. must be previously mapped and stored, and this data can be continuously updated and re-learned to update big data. In addition, the skin care recommendation service providing server 300 receives feedback from at least one user terminal 100, and if the feedback is negative, it evolves in the direction of correcting errors by performing reinforcement learning. It could be a server. To this end, the skin care recommendation service providing server 300 constructs big data using at least one information providing server 500 and previously stored history log data, and classifies big data through collection, pre-processing, and analysis. And a server for learning after clustering. In addition, the skin care recommendation service providing server 300 performs data learning using a deep learning artificial neural network algorithm for tagging an identifier from image data and image data in order to extract an identifier from image data or image data that is unstructured data. It can be an ongoing server. In addition, the skin care recommendation service providing server 300 may be a server for tagging or extracting an identifier from an image, image, etc. that are subsequently input according to a learning result.

Here, the skin care recommendation service providing server 300 may be implemented as a computer that can access a remote server or terminal through a network. Here, the computer may include, for example, a navigation system, a notebook equipped with a web browser, a desktop, a laptop, and the like.

At least one information providing server 400 provides information to the skin care recommendation service providing server 300 using a web page, app page, program or application related to a skin care recommendation service using big data-based artificial intelligence machine learning. It can be a server that does. At this time, the at least one information providing server 400 may be a server of a public institution or a server of a private company. In addition, the server of a private company may also be a server of a cosmetics company, a social media related company, a portal service company, etc., the type is not limited to the above, and the list may increase or decrease according to the embodiment or as the data to be collected is changed. It is obvious that it can be transformed. Here, the skin care recommendation service providing server 300 may collect information through web crawling when it is difficult to directly receive feedback from at least one information providing server 400, and in this case, provide at least one information The server 400 may be deleted.

Here, the at least one information providing server 400 may be implemented as a computer that can access a remote server or terminal through a network. Here, the computer may include, for example, a navigation system, a notebook equipped with a web browser, a desktop, a laptop, and the like. In this case, the at least one information providing server 400 may be implemented as a terminal capable of accessing a remote server or terminal through a network.

FIG. 2 is a block diagram illustrating a server for providing a skin care recommendation service included in the system of FIG. 1, and FIG. 3 is a skin care recommendation service using artificial intelligence machine learning based on big data according to an embodiment of the present invention. A diagram for explaining an implemented embodiment.

Referring to FIG. 2, the skin care recommendation service providing server 300 includes a diagnosis unit 310, a product collection unit 320, an environment collection unit 330, a classification unit 340, a recommendation unit 350, and a management unit. (360), a big data unit 370, a learning unit 380, and a verification unit 390 may be included.

The skin care recommendation service providing server 300 according to an embodiment of the present invention or another server (not shown) operated in conjunction with the at least one user terminal 100 and at least one information providing server 400 are used to provide big data. When transmitting a skin care recommendation service application, program, app page, web page, etc. using machine learning based artificial intelligence, at least one user terminal 100 and at least one information providing server 400 You can install or open skin care recommendation service applications, programs, app pages, and web pages using intelligent machine learning. In addition, a service program may be driven in at least one user terminal 100 and at least one information providing server 400 by using a script executed in a web browser. Here, the web browser is a program that enables you to use the web (WWW: world wide web) service, and refers to a program that receives and displays hypertext described in HTML (hyper text mark-up language). For example, Netscape , Explorer, chrome, etc. In addition, the application means an application on the terminal, and includes, for example, an app that is executed on a mobile terminal (smartphone).

Referring to FIG. 2, the diagnosis unit 310 transmits a Baumann skin type questionnaire (BSTQ) to the user terminal 100, and the result of a survey on the Baumann skin type questionnaire from the user terminal 100 By receiving data, you can diagnose your skin type. Here, the Bowman skin type questionnaire is oily or dry, sensitive or resistant, pigmented or nonpigmented, wrinkled, or tight. The purpose is to classify skin types according to the Bowman skin type index so that you can identify your own skin type among the 16 skin types by evaluating, and the skin types classified by the questionnaire are shown in Table 1 below, and (3) The four uppercase letters marked from ,3) are an acronym for the above-described characteristics. For example, OSPW is a skin type that has characteristics of oily, sensitive, pigmented, and wrinkled, that is, oily, sensitive, pigmented, and wrinkled. In addition, here, since the Bowman skin type questionnaire is the same as the disclosed technology, a detailed description will be omitted in the specification of the present invention.

OILY DRY PIGMENTED NONPIGMENTED PIGMENTED NONPIGMENTED Wrinkled OSPW OSNW DSPW DSNW Sensitive Tight OSPT OSNT DSPT DSNT Sensitive Wrinkled ORPW ORNW DRPW DRNW Resistant Tight ORPT ORNT DRPT DRNT Resistant

At this time, in addition to the Bowman questionnaire, genetic test results may be further added, and test results such as DNA and hair minerals may be added. This may be used later to analyze the user's skin pattern by calculating the user's skin type analysis result and genetic information. Here, the genetic information can be classified into good/careful/management by score by converting 100 points into a perfect score based on the results of a genetic analysis agency, and genetic testing is the basis for judging the current skin condition. Based on, the current skin care condition can be checked and directions can be presented, and examples thereof are shown in Table 2 below.

division Skin condition Criterion Good genetic index
(No variation) Good Excellent skin care Problem Lack of skin care Genetic index variation
(Including some variations) Good Excellent skin care before expression of mutant genes Problem Lack of secretion management for expression of mutant genes

The product collection unit 320 may collect product data including at least one or a combination of at least one of a basic care product, a sun care product, a cleansing product, and a makeup product used by the user from the user terminal 100. At this time, basic care may include skin, toner, lotion, essence, cream, face oil, no care, neck care, mist, mask, pack, and sun care, sun cream, sun spray, sun stick, after sun care, tanning May include, and the cleansing may include eye remover, lip remover, cleansing oil, cleansing oil, lotion, cleansing foam, cleansing gel, cleansing water, cleansing tissue, soap, scrub, peeling, and makeup, facial They can be classified by division. At this time, the ingredients contained in each product can be sorted by ingredient name, organized and analyzed. Here, it will be apparent that the collected product data is not limited to those listed. The environment collection unit 330 includes environmental data including gender, age, weight, lifecycle, diet, and occupation from the user terminal 100. Can be collected. At this time, the environmental data may collect physical data including gender, age, weight, height, and the like, and data such as occupation, lifestyle, and diet to grasp an environment such as external environment and exposure to electromagnetic waves. Here, it will be apparent that the collected environmental data is not limited to those listed.

The classification unit 340 may parse the diagnosed skin type, product data, and environmental data to derive and classify the skin type, product, and environment identifiers as input values. For example, assuming that skin type is classified as A type, product data is B product, and environmental data is classified as C data, X, Y, Z, which are pre-stored identifiers corresponding to A, B, C, are extracted, and user XYZ, which is an identifier for the skin, products used, and environment, can be assigned. This can then be used as an input value (reference value, reference data) to find a group closest to the user in big data. Here, in an embodiment of the present invention, unsupervised learning may be performed to find a pattern embedded in the data using the user's skin type, product data, and environmental data as input values. Here, machine learning may be defined as a methodology that gives a computer ability to learn by itself even if it is not explicitly programmed, and this is a method in which a program learns patterns of data by itself based on data. Machine learning is classified into supervised learning and unsupervised learning depending on whether the correct answer is specified in the data required for learning, and classification, which divides the data into finite categories according to the purpose of use, and continuous It is divided into regression that maps to values, clustering that combines similar data, and dimension reduction methodology that maps multidimensional data to a representative lower dimension.

Deep learning is one of the machine learning techniques that has dramatically improved the performance of machine learning, which has been stagnant for a while. Deep learning is a methodology based on an artificial neural network (ANN) algorithm that mimics the overlap of synapses in the human brain structure. , As a deep learning structure, a deep neural network (DNN) in which multiple hidden layers exist between an input layer and an output layer, and a filter required for factor extraction in front of the hidden layer. It may include a convolutional neural network (CNN) that learns filters together with a helical neural network (CNN), and a recurrent neural network (RNN) capable of processing time series data by stacking artificial neural networks of each time. Here, the high performance of the deep learning model is explained in two ways. First, the neural network is a universal approximator that can approximate all kinds of functions by overlapping the weighted sums of functions in each layer. Data can be simulated with high accuracy. Second, it is important to properly extract the factors representing the data in order to classify the data well, and the optimal factor can be extracted through filter learning using a spiral neural network. In addition, deep learning is a form in which a model in the field of artificial intelligence called a neural network is developed, and is a structure in which a hidden layer of an artificial neural network consisting of a hierarchical structure consists of several stages. In recent deep learning models, the number of weights (meaning connection strength) that connects nodes can reach up to billions due to the increased number of internal layers.

At this time, a convolutional neural network (CNN) is a model that has the advantage of being able to recognize even if the size or location of a pattern changes by imitating the visual processing process of an organism. It has three characteristics. receptive field), shared weight, and subsampling. First, the local receptive field means that each node of the lower layer is connected to all nodes of the upper layer by weights (filters), unlike in the existing model. When the filter is applied to all regions of the input pattern and the weight is multiplied by the region to be summed, transmission of the filter to the node of the panel of the upper node is called convolution. Second, weight sharing means that the weight is shared, meaning that the filter applied to the node and the input pattern (or lower layer) in one panel is the same. Since one filter can recognize one feature (or small pattern), one panel can find the feature that exists at a specific location of the input pattern. Third, sub-sampling is a process that occurs in the second and third layers, compressing the panels existing in the second layer, and is also called pooling. At this time, average pooling, which averages and multiplies the weights, is also often used, but in an embodiment of the present invention, maximum pooling using the largest value among four values in the 2X2 area is used. . Sub-sampling serves to reduce the size of the input pattern and to make it less sensitive to variations in the position of the found feature. This convolution and sub-sampling process may be repeated a plurality of times according to an embodiment, and after the sub-sampling is finished, the neural network is connected.

At this time, when learning the weights of the convolutional neural network, since the convolutional neural network has a very large number of inner layers, the number of nodes increases and the weights connecting the nodes increase. In general, when the number of weights in the model increases, the size of the first-order partiality function is repetitively calculated when the error backpropagation algorithm is applied. However, in the case of a convolutional neural network, there is a feature of weight sharing, so the number of weights is not large enough to cause such a problem, so an error backpropagation algorithm can be applied without difficulty. However, since the feature of weight sharing and the process of subsampling are different from that of a general neural network model, the algorithm may be modified. For example, for shared weights, after calculating the linear partial derivative in the same way as when the weights are different, a process of adding all the values of the linear partial derivatives to the same weight is applied. When a pattern is compressed through sub-sampling, an operation in the opposite direction must be performed when backpropagating an error, so a process of expanding through up-sampling may be further added. However, it will be apparent that it is not limited to the above-described method and that various methods can be used.

Meanwhile, a recurrent neural network (RNN) model is suitable for processing sequential information such as speech recognition or language recognition. The RNN receives and processes the elements constituting sequential data one at a time, and stores the processed information in the internal node.It is necessary after creating the corresponding internal node according to the input time and storing the processed information up to that point in time. This is a method of outputting through the output node. In addition, the circulatory neural network model may be input/output non-sequential data in some cases, and can be applied in various ways depending on the type of input/output data.If input/output data is sequential, it can be applied to machine translation or annotating videos. In addition, if the output is out of sequence, it may be applied to emotional analysis, and when the input is out of sequence, it may be applied to annotate a picture, but is not limited thereto. Lastly, a deep belief network is an unsupervised learning model, characterized by the fact that humans and animals learn by themselves through observation. In this case, it will be apparent that it is not limited to the listed ones and that various machine learning methods may be used, and may be changed according to embodiments.

The recommendation unit 350 extracts reference data having a preset similarity from pre-built big data based on the classified identifier, and extracts a product including a recommended component previously mapped to the reference data and stored, and the user terminal 100 ) Can be recommended. The recommendation unit 350 extracts reference data having a preset similarity from pre-built big data based on the classified identifier, and extracts a product including a recommended component previously mapped to the reference data and stored, and the user terminal 100 ), based on the classified identifier, automatically derives a query to search for reference data having a preset similarity, and uses the derived query to search for reference data having a preset similarity, As a result of the search, reference data having a preset similarity may be extracted.

At this time, when using big data for recommendation, collaborative filtering may be used. Here, Collaborative Filtering creates a profile based on information on the user's skin condition, environmental data, and product data, and based on this, other users with similar purchase preferences, skin conditions, and environmental data It's a way to recommend items you buy or prefer. For example, after creating a single profile or multiple profiles based on the purchase preference, skin condition and environmental data similar to the target customer, and ensemble data mining techniques such as artificial neural networks and decision trees and collaborative filtering techniques, the recommendation performance is improved. It can be improved. Alternatively, a fusion technique that creates one profile using skin conditions and combines the prediction results of the k-NN model between users and the k-NN model between products can be used, or purchase data and review data of online stores Using a user-product matrix-based profile, a logistic regression analysis, decision tree, and artificial neural network model can be constructed and ensembled. However, it is not limited to the above-described method, and it will be apparent that a method of making a recommendation based on a user profile can be used in various ways.

The management unit 360 parses the skin type, product data, and environmental data diagnosed by the classification unit 340 to derive and classify the skin type, product, and environment identifiers as input values, Weather data including any one or at least one of temperature, humidity, ultraviolet rays, fine dust, ultrafine dust, and yellow dust may be collected using a web crawler. In addition, the management unit 360 extracts reference data having a preset similarity from pre-built big data based on the classified identifier, extracts a skin care method previously mapped to the reference data, and stores the stored skin care method, and sends it to the user terminal 100. I can recommend it. For example, in autumn, where dry weather continues, itching and skin tension appear as the moisture in the skin decreases, and many women try to keep moist by applying skin moisturizers. However, people who have a lot of skin problems and who show oily skin even in the fall are worried about choosing a moisturizing product, because if you choose the wrong one, it can cause greasy and trouble. Accordingly, if the user has oily skin, it may be recommended to use a moisturizing cream or moisturizing product without oil, or a product that helps to suppress wetting and trouble, such as glycerin or tea tree ingredients. In addition, after cleansing, we will provide recommended products extracted according to the skin condition identified by the diagnosis unit 310 so that if you apply a moisturizing cream or essence that has less oil but can increase moisture supply and retention, the inside of the skin can be prevented from drying out. I can.

The big data unit 370 transmits a Baumann skin type questionnaire (BSTQ) from the diagnosis unit 310 to the user terminal 100, and conducts a survey on the Baumann skin type questionnaire from the user terminal 100. Before receiving the result data and diagnosing the skin type, cosmetic ingredient information is collected from at least one cosmetic manufacturer server, review data and ingredient data for at least one type of cosmetic product are collected through web crawling, and at least one Harmful ingredient data is collected from public servers, and raw data including collected cosmetic ingredient information, review data, ingredient data, and harmful ingredient data are stored in parallel and distributed, and unstructured ( Unstructed data, structured data and semi-structured data are refined, pre-processing including classification as metadata, extracting pre-processed data, and analysis including data mining You can build big data by performing a method and visualizing and outputting the analyzed data. In this case, it may be stored to be classified according to components, items, classifications, keywords, etc., and classification analysis by component, user cluster analysis, trouble association analysis, product prediction analysis, and the like may be performed.

Here, in data mining, classification (Classification), which predicts a class of new data by learning a training data set with a known class by searching for an intrinsic relationship between preprocessed data, or clustering that group data based on similarity without class information ( Clustering) may be included. Of course, there may be various other mining methods, and may be mined differently according to the type of big data to be collected and stored or the type of query to be requested later. The big data constructed in this way may be verified through deep learning or machine learning of artificial neural networks. At this time, the artificial neural network can use a CNN (Convolutional neural network) structure, which is a network structure using a convolutional layer, which is suitable for image processing, and can classify images based on features in the image by inputting image data. Because it can. In addition, text mining is a technology that aims to extract and process useful information from non/semi-structured text data based on natural language processing technology. Through text mining technology, meaningful information can be extracted from a vast bundle of texts, linkages with other information can be grasped, and the results can be obtained beyond finding a category of text or simply searching for information. Using this, in the free travel integrated reservation service according to an embodiment of the present invention, large-capacity language resources and statistical and regular algorithms can be used to analyze an identifier or natural language input as a query and discover hidden information therein. have. In addition, cluster analysis can be used to finally discover a group of similar characteristics while combining objects with similar characteristics, and when a recommendation consultation is requested afterwards by grasping the characteristics of the user cluster, the object of the recommendation consultation is determined. Can be classified.

The learning unit 380 extracts reference data having a preset similarity from the big data previously constructed based on the identifier classified by the recommendation unit 350, and selects a product including the recommended component previously mapped to the reference data and stored. After extracting and recommending it to the user terminal 100, feedback data on whether the recommended product is appropriate may be received from the user terminal 100, and reinforcement learning may be performed using the received feedback data. . At this time, reinforcement learning is a field of artificial intelligence that learns through compensation for actions, and various algorithms to implement it have been developed and applied in various fields such as robot control, communication networks, and board games. In this case, in the machine learning problem, reinforcement learning assumes that an agent (or robot) in an arbitrary environment recognizes its current state and acts. Then, the agent gets a positive or negative reward from the environment. The reinforcement learning algorithm finds a policy that maximizes the reward value accumulated in the agent during the problem solving process. In the reinforcement learning method, the environment is formalized by finitestate Markov decision processes (MDPs). In addition, reinforcement learning algorithms are directly related to dynamic programming techniques. Here, dynamic programming refers to dividing a given problem into several problems, recording the results previously performed, and then using them again later. In addition, it can be said that it is important to find a balance between exploration of an area that has not yet been investigated and exploration of already known knowledge in the reinforcement learning method.

The verification unit 390 extracts reference data having a preset similarity from the big data previously established based on the identifier classified by the recommendation unit 350, and selects a product including the recommended component previously mapped to the reference data and stored. After extracting and recommending it to the user terminal 100, if negative feedback data on whether the recommended product is appropriate or not is received from the user terminal 100, authentication is performed to verify the error of the product containing the recommended component, After identifying the part and the cause of the error, and updating the identified part of the error and the cause of the error in big data, deep learning using an artificial neural network is performed and deep learning to find the pattern of the error and predict the error rate through classification. It can be reflected in the data derived from the big data. In this case, deep learning, artificial neural networks, and reinforcement learning are the same as the above-described methods, and thus description thereof will be omitted.

Hereinafter, the operation process according to the configuration of the above-described skin care recommendation service providing server of FIG. 2 will be described in detail with reference to FIG. 3 as an example. However, it will be apparent that the embodiment is only any one of various embodiments of the present invention, and is not limited thereto.

Referring to FIG. 3A, (a) the skin care recommendation service providing server 300 extracts review data and cosmetic ingredient data through online, and collects public data to identify harmful ingredients, extract ingredient information, and retrieve natural environment information. It is extracted, and cosmetic information of the manufacturer can be collected through the cosmetic manufacturer. And, the skin care recommendation service providing server 300 builds big data and does not recommend products to users only based on big data, but performs the user's skin type including genetic testing, and (b) By using an unsupervised learning method and a convolutional (convolution) network, sharing weights, and zooming out with maximum pooling, only the user's information is input to discover patterns, characteristics, and structures inherent in the data through learning. Compose and operate only with the input properties of the object for the training data. In addition, (c) personal information is collected in three levels (third order), and reinforcement learning is performed through feedback to provide a recommendation service in consideration of consumer behavior and observation range of state transition rules.

Referring to FIG. 3B, the method of collecting data and the type of data include web documents, contents, OPNE API, real-time collection data, and data collection logs, and the collection point is open such as KFDA public data, ICID, and WIKI. It may include a database, a web database, a social media, and a keyword database of Google search, but is not limited to the above. In addition, data pre-processing can go through the process of filtering, conversion, purification, integration, and reduction, and data can be stored according to various factors including items, components, classifications, and keyword effectiveness, classification analysis by component, and user clustering. This result can be visualized on the platform by performing analysis, trouble-related analysis, and product prediction analysis.

The matters that are not described about the method of providing a skin care recommendation service using artificial intelligence machine learning based on big data of FIGS. 2 and 3 are described above. The method of providing a skin management recommendation service using artificial intelligence machine learning based on big data through FIG. 1 Since the contents are the same as those described for or can be easily inferred from the described contents, the following description will be omitted.

4 is a diagram illustrating a process of transmitting and receiving data between components included in the system for providing a skin care recommendation service using artificial intelligence machine learning based on big data of FIG. 1 according to an embodiment of the present invention. Hereinafter, an example of a process in which data is transmitted/received between each component will be described with reference to FIG. 4, but the present application is not limitedly interpreted as such an embodiment, and is illustrated in FIG. 4 according to various embodiments described above. It is apparent to those skilled in the art that the process of transmitting and receiving data may be changed.

Referring to FIG. 4, the skin care recommendation service providing server 300 collects data from at least one information providing server 400 (S4100), analyzes the data through data preprocessing (S4200), and visualizes the result. In this way, big data is built (S4300).

On the other hand, when a diagnosis request is received from the user terminal 100 (S4400), the skin care recommendation service providing server 300 transmits a Bowman questionnaire to the user terminal 100 to receive the answer (S4500, S4600), and The type is diagnosed (S4700), used product and environment data input from the user terminal 100 is integrated (S4800), and patterns and characteristics of user data are learned by unsupervised learning (S4810). And, the skin care recommendation service providing server 300 extracts the identifier by parsing the data (S4820), generates a query with the identifier (S4830), extracts a recommended product that is an answer to the query (S4840), and the user terminal The recommendation is made by transmitting to (100) (S4850).

At this time, the skin care recommendation service providing server 300 performs reinforcement learning if there is feedback from the user terminal 100 (S4860), if it is negative feedback (S4900, S4910), more preferably through deep learning of an artificial neural network. The error is verified (S4920), the error pattern is extracted (S4930), and big data is retrained and updated based on this.

The order between the above-described steps S4100 to S4930 is only an example and is not limited thereto. That is, the order between the above-described steps (S4100 to S4930) may be mutually changed, and some of the steps may be executed or deleted at the same time.

The matters that are not described about the method of providing a skin care recommendation service using artificial intelligence machine learning based on big data of FIG. 4 are described above with reference to FIGS. 1 to 3 as a method of providing a skin management recommendation service using artificial intelligence machine learning based on big data. Since the contents are the same as those described for or can be easily inferred from the described contents, the following description will be omitted.

5 is a flowchart illustrating a method of providing a skin care recommendation service using artificial intelligence machine learning based on big data according to an embodiment of the present invention. Referring to FIG. 5, the skin care recommendation service providing server transmits a Baumann skin type questionnaire (BSTQ) to a user terminal, and receives survey result data for the Baumann skin type questionnaire from the user terminal, The type is diagnosed (S5100).

In addition, the skin care recommendation service providing server collects product data including at least one or a combination of at least one of a basic care product, a sun care product, a cleansing product, and a makeup product used by the user from the user terminal (S5200), and the user Environmental data including gender, age, weight, life cycle, diet, and occupation are collected from the terminal (S5300).

In addition, the skin care recommendation service providing server parses the diagnosed skin type, product data, and environment data to derive and classify skin type, product, and environment identifiers as input values (S5400), and classified Based on the identifier, reference data having a predetermined similarity is extracted from pre-established big data, and a product including the recommended component previously mapped to the reference data is extracted and recommended to the user terminal (S5500).

The matters that are not described about the method of providing a skin care recommendation service using artificial intelligence machine learning based on big data of FIG. 5 are described above with reference to Figs. 1 to 4, and a method of providing a skin management recommendation service using artificial intelligence machine learning based on big data. Since the contents are the same as those described for or can be easily inferred from the described contents, the following description will be omitted.

A method for providing a skin care recommendation service using artificial intelligence machine learning based on big data according to an embodiment described with reference to FIG. 5 is a recording medium including instructions executable by a computer such as an application or program module executed by a computer. It can also be implemented in the form of. Computer-readable media can be any available media that can be accessed by a computer, and includes both volatile and nonvolatile media, removable and non-removable media. Further, the computer-readable medium may include all computer storage media. Computer storage media includes both volatile and nonvolatile, removable and non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data.

The method for providing a skin care recommendation service using artificial intelligence machine learning based on big data according to an embodiment of the present invention described above includes an application basically installed in a terminal (this includes a program included in a platform or an operating system basically installed in the terminal). It may be executed by an application store server, an application, or an application (that is, a program) directly installed on the master terminal through an application providing server such as a web server related to the service. In this sense, the method for providing a skin care recommendation service using artificial intelligence machine learning based on big data according to an embodiment of the present invention described above is implemented as an application (i.e., a program) installed in a terminal or directly installed by a user. It can be recorded on a computer-readable recording medium such as a terminal.

The above description of the present invention is for illustrative purposes only, and those of ordinary skill in the art to which the present invention pertains will be able to understand that it can be easily modified into other specific forms without changing the technical spirit or essential features of the present invention. will be. Therefore, it should be understood that the embodiments described above are illustrative in all respects and not limiting. For example, each component described as a single type may be implemented in a distributed manner, and similarly, components described as being distributed may also be implemented in a combined form.

The scope of the present invention is indicated by the claims to be described later rather than the detailed description, and all changes or modified forms derived from the meaning and scope of the claims and their equivalent concepts should be interpreted as being included in the scope of the present invention. do.

Claims (7)

delete delete delete delete delete delete In the skin care recommendation service providing method executed in the skin care recommendation service providing server,
The user terminal evaluates the presence of oily or dry, sensitive or resistant, pigmented or nonpigmented, wrinkled, or tight 16 types Sends a Baumann skin type questionnaire (BSTQ) that can identify one's skin type among skin types classified as, and receives data from the survey result on the Baumann skin type questionnaire from the user terminal to determine the skin type. Diagnosing a user's skin type by further adding a genetic test result including DNA and hair minerals;
A basic care product including skin, toner, lotion, essence, cream, face oil, no care, neck care, mist, mask, and pack, sun cream, sun spray, sun stick, after sun care from the user terminal , At least one or at least one of makeup products consisting of sun care products including tanning, eye remover, lip remover, cleansing oil, lotion, cleansing foam, cleansing gel, cleansing water, cleansing tissue, soap, scrub, and cleansing products including peeling Collecting product data including one combination;
Collecting environmental data including lifecycle, diet, and occupation in order to grasp physical data including gender, age, weight, and height from the user terminal and an environment including external environments and exposure to electromagnetic waves;
Parsing the diagnosed skin type, product data, and environment data to derive and classify the skin type, product, and environment identifiers as input values; And
Extracting reference data having a predetermined similarity from pre-established big data based on the classified identifier, extracting a product including a recommended component previously mapped to the reference data and stored, and recommending it to the user terminal;
Including,
After the step of parsing the diagnosed skin type, product data, and environmental data to derive and classify identifiers of the skin type, product, and environment as input values,
Collecting meteorological data including any one or at least one of temperature, humidity, ultraviolet rays, fine dust, ultrafine dust, and yellow dust on a daily basis using a web crawler; And
Diagnosis unit 310, product collection unit 320, environment collection unit 330, classification unit 340, recommendation unit 350, management unit 360, big data unit 370, learning unit 380, And, through the skin care recommendation service providing server 300 composed of the verification unit 390, the reference data having a preset similarity is extracted from the pre-built big data based on the classified identifier, and pre-mapped to the reference data and stored. Extracting a skin care method and recommending it to the user terminal;
It further includes,
Before the step of diagnosing skin type by transmitting a Baumann skin type questionnaire (BSTQ) to the user terminal and receiving survey result data for the Baumann skin type questionnaire from the user terminal,
Collecting cosmetic ingredient information from at least one cosmetic manufacturer server;
Collecting review data and ingredient data for at least one kind of cosmetics through web crawling;
Collecting harmful ingredient data from at least one public server;
Storing the collected cosmetic ingredient information, review data, ingredient data, and raw data including harmful ingredient data in parallel and distributed;
Purifying unstructured data, structured data, and semi-structured data included in the stored raw data, and performing pre-processing including classification as metadata;
Performing an analysis including data mining on the pre-processed data; And
Visualizing and outputting the analyzed data;
Including more,
After the step of extracting reference data having a predetermined similarity from pre-built big data based on the classified identifier, extracting a product containing recommended components previously mapped to the reference data, and recommending it to the user terminal ,
Receiving feedback data on whether the recommended product is appropriate from the user terminal; And
Performing reinforcement learning using the received feedback data;
Including,
In the data mining, classification (Classification) for predicting a class of new data by searching for an intrinsic relationship between the pre-processed data and learning a training data set with a known class, or clustering that group data based on similarity without class information ( Clustering),
After the step of extracting reference data having a predetermined similarity from pre-built big data based on the classified identifier, extracting a product containing recommended components previously mapped to the reference data, and recommending it to the user terminal ,
When receiving negative feedback data on whether or not the recommended product is appropriate from the user terminal, performing authentication for error verification of the product including the recommended component to determine the error part and the cause of the error;
Performing deep learning using an artificial neural network in order to find the pattern of the error and predict the error rate through classification after updating the identified error part and the cause of the error in the big data; And
Reflecting the data derived by the deep learning to the big data;
It further includes,
The step of extracting reference data having a preset similarity from pre-built big data based on the classified identifier, extracting a product containing a recommended component previously mapped to the reference data and storing the recommended component, and recommending it to the user terminal,
Deriving a query for automatically searching for reference data having the preset similarity based on the classified identifier;
Searching for reference data having the predetermined similarity by using the derived query; And
Extracting reference data having the predetermined similarity as a result of the search;
Is executed by performing
The skin care recommendation service providing server,
Build big data using at least one information providing server and previously stored history log data, classify and cluster big data through collection, preprocessing, and analysis, and then proceed with learning,
Data learning is performed using a deep learning artificial neural network algorithm for tagging an identifier from image data and image data in order to extract an identifier from unstructured data, such as image data or image data,
A method of providing a skin care recommendation service using big data-based artificial intelligence machine learning to extract and tag an identifier from an image or image that is subsequently input according to the learning result.

Images