KR102320003B1 - The user interface method of recommending cosmetic using two criteria in cosmetic information providing mobile application - Google Patents

Abstract

The present invention relates to a user interface method of a mobile application for providing cosmetic information.
The first criterion is the recent history of the user's direct previous cosmetic inquiry. The second criterion is the similarity of the review characteristics to the cosmetics screened by the first criterion. The similarity of this review attribute is quantified as a review attribute table in a review-based similar product database. According to the present invention, when a user enters the cosmetic recommendation screen after creating a cosmetic inquiry event, the above two criteria can be applied to recommend related products.

Description

A user interface method that recommends cosmetics by applying two criteria in a cosmetics information provision mobile application.

The present invention relates to a technology for providing cosmetic information using mobile application software.

Cosmetics are the representative consumer goods of human life. Although cosmetics have traditionally been used for color makeup purposes, in recent years, various types of cosmetics have been sold, including functional cosmetics such as UV protection, wrinkle improvement, and whitening. Basic cosmetics, hair cosmetics, color make-up products, body cleansing products, shaving products, and eye makeup products account for the largest proportion, with nail products, infant products, fragrance products, hair dyeing products, bath products, Industrial fields are expanding to products for preventing body odor.

However, cosmetics have the characteristic of being used directly on the user's body. Therefore, the fact that the user's physical characteristics are quite different from person to person cannot but be emphasized. It is also considered very important in this field that the purpose and preference of cosmetics use varies from person to person. Therefore, cosmetics users want to obtain considerable information based on their physical characteristics and personal interests before purchasing cosmetics. And usually listen to the experience of others. They purchase cosmetics based on the reactions of people with similar physical characteristics and the experiences of people with similar interests.

Therefore, in the cosmetic information provision mobile application technology, it is essential to create a user interface structure so that the experiences and opinions of others can be conveyed to the user. For this purpose, the applicant of the present invention has been hard at research and development as a leading company in the industry. Prior Patent Document 1 discloses a technique for searching cosmetic information based on ingredients of cosmetics, and Prior Patent Document 2 discloses a method of effectively visualizing review data of others. Prior Patent Document 1 had the advantage of facilitating the search for cosmetics suitable for the physical characteristics of cosmetics users. Prior Patent Document 2 had the advantage of being able to more easily obtain other people's experience by allowing a large number of review data to be visualized with a certain pattern.

However, in these prior patent documents, the user's own interests and the user's own behavior in the application are not considered. The inventors of the present invention explored these limitations further. Not all users browse products in the application and purchase immediately. Although the user has viewed a specific product through the application, it cannot be concluded that the product is the best product. You may be more drawn to the product you see today than the product you saw yesterday. The inventors of the present invention have completed the present invention after long research and efforts to find a user interface method that more actively considers the behavior patterns of these users.

Prior Patent Document 1: Korean Patent No. 1677128 (Birdview Co., Ltd.)

Prior Patent Document 2: Korean Patent No. 19553180 (Birdview Co., Ltd.)

An object of the present invention is to provide a method for recommending optimized cosmetic information to a user who uses a cosmetic information providing mobile application. The inventors of the present invention have perfected a user interface method in which both are considered simultaneously. First, the history of cosmetics inquiry by the mobile application user while using the application is taken as an essential element. Second, the review data of others managed as a database on the server is an essential element.

On the other hand, other objects not specified in the present invention will be additionally considered within the range that can be easily inferred from the following detailed description and effects thereof.

In order to achieve the above object, the first aspect of the present invention is a user interface method of a cosmetic information providing mobile application, wherein a mobile device executes the cosmetic information providing mobile application to perform N (N is an integer greater than 1) cosmetic inquiry event. After generating , entering the cosmetic recommendation screen of the cosmetic providing mobile application, displaying the inquiry cosmetic information extracted in response to the cosmetic inquiry event, and at the same time, the inquiry cosmetic in the review-based similar product database managed by the server It characterized in that it comprises the step of extracting the matching cosmetic information stored in advance by matching the product with similar review characteristics to the inquiry cosmetic displayed as information, and displaying the information together with the inquiry cosmetic information.

In the user interface method for recommending cosmetics by applying two criteria in the cosmetics information providing mobile application according to a preferred embodiment of the present invention, the inquiry cosmetic information relates to cosmetics corresponding to the recent cosmetic inquiry event of the mobile device. Information is good.

In addition, a second aspect of the present invention is a user interface method for recommending cosmetics by applying two criteria in a cosmetics information providing mobile application:

The server that provides cosmetic information providing mobile application extracts the pre-registered review representative expression from the product review of the user review database, generates a representative expression vector for each cosmetic, and creates a first vector, and separately from the representative expression vector, the same product A second vector is created for each cosmetic by calculating the probability that the words included in the review will appear, and each word is vectorized. After linking the first vector and the second vector, the cosmetic similarity between products is calculated and quantified as a review characteristic. After building a review-based similar product database,

After the mobile device executes the cosmetic information providing mobile application to generate a cosmetic inquiry event N (N is an integer greater than 1) times, and then enters the cosmetic recommendation screen of the cosmetic providing mobile application,

The server displays the inquiry cosmetic information extracted in response to the cosmetic inquiry event on the application screen, and at the same time extracts the matching cosmetic information in the order of the high similarity between the inquiry cosmetic information and the review characteristics from the review-based similar product database. It characterized in that it comprises the step of displaying together with the inquiry cosmetic information.

According to the present invention as described above, a user interface for recommending related products can be structured by applying two criteria: a recent history in which a user directly previously inquired a cosmetic product, and a similarity in review characteristics for a cosmetic selected based on a recent inquiry history. can By configuring the cosmetic recommendation user interface in this way, it is possible to achieve the effect that the recommendation element reflecting one's own interests and the recommendation factor reflecting the interests of others can be considered together.

On the other hand, even if it is an effect not explicitly mentioned herein, it is added that the effects described in the following specification expected by the technical features of the present invention and their potential effects are treated as described in the specification of the present invention.

1 is a diagram schematically showing a system configuration according to a preferred embodiment of the present invention.
Fig. 2 shows an example of the overall process of the user interface method according to any preferred embodiment of the present invention.
Fig. 3 schematically shows a process for building a review-based similar product database according to one preferred embodiment of the present invention.
4 shows a schematic configuration example of a process for generating a vector by managing review big data as a representative expression in the present invention.
5 is an example of a user screen according to a preferred embodiment of the present invention.
※ It is revealed that the accompanying drawings are exemplified as a reference for understanding the technical idea of the present invention, and the scope of the present invention is not limited thereby.

Hereinafter, with reference to the drawings, the configuration of the present invention guided by various embodiments of the present invention and effects resulting from the configuration will be described. In the description of the present invention, if it is determined that the subject matter of the present invention may be unnecessarily obscured as it is obvious to those skilled in the art with respect to related known functions, the detailed description thereof will be omitted.

1 schematically shows a system configuration according to a preferred embodiment of the present invention.

The service server 100 provides and manages the cosmetic information providing mobile application software 12 . The user mobile device 10 downloads the cosmetic information providing mobile application software 12 through the application platform and installs it in the device. And the mobile device 10 connects and communicates with the service server 100 through the communication network by executing the cosmetic information providing mobile application software 12 .

The cosmetic information providing mobile application software 12 of the present invention has, in addition to the function of accessing the service server to transmit and receive data, a function to search and read cosmetic information, a function to write and upload a review, a function to select and compare cosmetics, a cosmetic product It includes functions to purchase and pay for products, manage personal information, and recommend cosmetics. Information is displayed in data formats such as text, image, video, and voice, and it is structured as a user interface.

The service server 100 for providing and managing cosmetic information providing mobile application software (hereinafter, abbreviated as "cosmetic application") is composed of one or more server devices, and the server device includes a plurality of hardware/software equipment. It stores the resources necessary for the normal execution of the application and performs the functions of managing and communicating data.

The service server 100 has a plurality of databases. The product information database 111 may store product information to be provided to the user device through the cosmetic application. It can store product information based on text/image/voice, and includes cosmetic ingredient information or functional information suitable for cosmetic applications.

The user information database 112 may record user information. In particular, user profile information such as the user's age, gender, skin type (dry, oily, neutral, combination), skin concerns (atopic, acne, sensitive skin), and having children can be stored and managed for each user account.

The review database 113 stores review data collected from other users connected through a network. The present invention is particularly relevant to a technique for refining review data stored in the review database 113 by language processing in several steps.

The review-based similar product database 114 processes the review data recorded in the review database 113 to quantify the review characteristics for each product and stores the recorded data. Matching cosmetic information in which the cosmetic similarity between products is quantified is recorded in this database 114 .

Meanwhile, although not shown, many more diverse databases may be included. A database for recording logs, a database for recording billing information, a database for recording advertisement information, etc. are included as necessary.

The service server 100 may include a cosmetic recommendation module 102 and a review characteristic module 104 . The cosmetic recommendation module 102 recommends optimal cosmetics according to a predetermined rule to the mobile device through the cosmetic application. In particular, in the present invention, the cosmetic recommendation module 102 recommends cosmetics by applying two criteria. The review characteristic module 104 provides a function of analyzing and linguistically processing review data stored in the review database 113 so that it can be provided to the user. For example, various step-by-step language processing functions shown in FIGS. 3 and 4 may be all or selectively included. The review characteristics module 104 includes a language processing module. Stemmers for example. It includes a parser, topic extractor, semantic class generator, polarity classifier, and set matching module for converting to representative expressions. In addition, language processing is performed using various pre-built language dictionaries. Then, the processed and processed data is managed as matching cosmetic information in the review-based similar product database 114 .

Fig. 2 schematically shows the overall process of a user interface method according to a preferred embodiment of the present invention.

First, the mobile device executes the cosmetic application and generates N (N is an integer greater than 1) cosmetic inquiry event on the application screen (S100). The cosmetic inquiry event means that the mobile device executed the cosmetic application to access the service server, and clicked on a specific cosmetic to view detailed information. Therefore, the service server can specify the corresponding cosmetic product.

Then, the mobile device enters the cosmetic recommendation screen through the cosmetic application user interface (S110). Preferably, the phrase/symbol of "recommendation" is configured as a selectable button in the user interface, so that the input means (means for generating a touch event) of the mobile device selects the recommendation button, so that step S110 can be executed.

Then, steps S120 and S130 are simultaneously executed. On the other hand, inquiry cosmetic information corresponding to the cosmetic inquiry event executed in step S110 is displayed through a partial area of the cosmetic recommendation screen (S120). At the same time, on the other hand, the displayed cosmetic information and the review characteristics are matched with similar products, and matching cosmetic information stored in advance is extracted and displayed on a partial area of the cosmetic recommendation screen (S130). Preferably, the matching cosmetic information is calculated in advance and registered for each cosmetic, or a plurality of cosmetic products having a higher ranking among the rankings of similarities determined for each cosmetic that are calculated in real time by the review characteristic module are extracted and displayed. .

By configuring the cosmetic recommendation user interface through such a process, it is possible to achieve the effect that the recommendation element reflecting one's own interests and the recommendation element reflecting the interests of others can be considered together.

Now, let's take a look at a preferred embodiment of how the review characteristics are quantified as cosmetic similarity between products. 3 shows a process in which the review characteristics module aligns the review characteristics based on the similarity of each product according to a preferred embodiment of the present invention.

Steps S201 and S203 and steps S202 and S204 may be executed, respectively.

First, a representative expression for each review is extracted (S201). And the extracted representative expression is vectorized (S203). In this regard, it is necessary to examine in detail what a representative expression is and how language processing is performed. See Figure 4. The review source uploaded from the cosmetic application user device stored in the review resource database 121 and the representative expression registered in advance in the representative expression database 123 are matched. In order to actually perform this matching, the dictionary base 160 on which several dictionaries are built is used step by step.

The dictionary base 160 is preferably a topic dictionary 161, a topic-semantic class dictionary 162, a semantic class-morpheme-polar dictionary 163, a negative word dictionary 164, a topic-polarity-representative expression dictionary ( 165) may be included.

A topic set is generated by extracting a topic from a review source using data registered in the topic dictionary 161 . The set matching module 120 matches the topic set and the representative expression set registered in the representative expression database. To this end, language processing such as morpheme analysis, syntax analysis, and semantic classification analysis is performed using the above-mentioned dictionary. In addition, since polarity is given to the representative expression in the present invention, even if the same word is included, it is structured to have a completely different meaning depending on the polarity. Therefore, the aforementioned semantic class-morpheme-polarity dictionary 163, negative word dictionary 164, and topic-polarity-representative expression dictionary 165 are used for polarity determination. The specific method will be described again below.

One or more representative expressions may be extracted by executing the set matching module 120 for a review uploaded by a certain reviewer. At this time, it is the equivalent of a plurality of representative expressions extracted. However, when P (P is an integer greater than or equal to 1) reviewers upload a review, the representative expressions extracted through each event are not equivalent in terms of frequency in the entire review big data.

The frequency calculation module 122 calculates a frequency for all the extracted representative expressions. The vector generation module 124 generates a vector for the representative expression. Also, preferably, the vector generation module 124 vectorizes and connects the product review document as described below.

Now, based on the above configuration and principle, we will look at the vector generation in step S203 for the representative expression. Assume that a reviewer mobile device executes the application of the present invention and inputs the following text through the review interface.

The whiteness is really bad ㅠㅠ

Since it doesn't spread well, I think it's better to mix it with oil or essence. I use it considering that it has minimized harmful ingredients, but I really don't know what to do with the white turbidity...

Then, morphological analysis and syntax analysis are performed. For example:

Cloudy _NNG Real_MAG Severe_VA_EF ㅠㅠ_UNKNOWN Application_NNG Castle_XSN Degree_JX Not So_MAG_VCP_EC Oil_NNG or _JC Essence_NNG Mixed_VV_ETM_ NNB _JKS ok_VA_ETM _NNB _VA _EF harmful_XR _XSA _ETM ingredient_NNG_JKO MIN_NNG_XSN _XSVN_ETM _NNB Considering _VCP ni_EC _NNG ha_XSV high _EC writing _VV but _EC cloudy_NNG Eun_JX really_MAG how_MAG ha_VV aya_EC ha_VXd_EF._SF.._SY

Extract the topics included in the review using the topic dictionary. As a result of the above morphological analysis, a topic set called {white turbidity, balm} was extracted.

Then, the topic set and the representative expression set are matched. In the representative expression set, for example, {with cloudiness, no cloudiness, good application, poor application, good moisture, bad moisture, oily, oily, less oily, bad oiliness, good moisture, bad moisture} Representative expressions are included.

Extracting the syntax related to the topic set is as follows.

(One) The whiteness is really bad ㅠㅠ

(2) Because the spreadability is not good

Also, if a semantic class set is prepared using the topic-semantic class dictionary, 'white turbidity' in phrase (1) becomes [presence], and 'balim' in phrase (2) becomes [presence, positive or negative]. The morpheme set to be searched based on the semantic class is as follows.

(One) [Yes_VV, Deep_VA, No_VV, Gone_VV]

(2) [Yes_VV, severe_VA, no_VV, disappear_VV, okay_VA, good_VA, bad_VA, not so_MAG]

The semantic class-morpheme-polarity dictionary is shown in Table 1 below, and the negative word dictionary is shown in Table 2 below.

semantic class morpheme polarity positive negative ok_VA One positive negative good_VA One positive negative bad_VA -One positive negative not much_MAG -One existence and nonexistence have_VV One existence and nonexistence deep _VA One existence and nonexistence no_VV -One existence and nonexistence disappear_VV -One

negative not_VX Ahn_MAG no_VCN

If morphemes and negative words are searched for in the above-mentioned phrases, (1) becomes [deep_VA], (2) becomes [star_MAG], and when morpheme polarity is identified in Table 1, phrase (1) becomes [1] ] and (2), the polarity of [-1] is determined.

Then, the topic can be converted to <white cloudy> and <bad spreadability> in the representative expression set.

The process discussed above was for a single review source. However, considering the fact that many review sources are uploaded through the network as described above (for example, there are P number of review sources), it is inevitable to consider the frequency. Therefore, even if there is one review topic converted to the representative expression of <bad applicationability>, if the expression <bad applicationability> is not detected in P-1 review sources, the representative expression frequency of <bad applicationability> is 1.

The above process can be exemplified as in [Table 3], including the extraction and frequency of representative expressions for <Sunscreen 2> and <Sunscreen 3>, for example, if the extraction and frequency of representative expressions for <Sunscreen 1> are involved.

product name review Representative Expression Extraction Representative expression frequency calculation sun cream 1 The whiteness is really bad ㅠㅠ
I think it's better to mix it with oil or essence because it doesn't apply well.
I use it considering that it has minimized harmful ingredients, but what should I really do about the white turbidity?? cloudy One poor fit One sun cream 2 There is a little white turbidity.
Immediately after application, the hair sticks slightly!
As time passes, it becomes less sticky and rather holds sebum, so if you have dry skin, it may feel dry.
It's an inorganic sunscreen and the spreadability is fine. cloudy One good spreadability One sun cream 3 I was worried that there was a white cloudiness, but surprisingly it wasn't too bad.
The spreadability is also much better than I expected! No irritation, no trouble, good cloudless One good spreadability One

Next, after calculating the frequency of the representative expression results extracted from the product review, it is normalized to a value between 0 and 1 (min-max normalized) to convert the representative expression frequency for each review into a representative expression vector.

product name cloudy cloudless good spreadability poor fit … sun cream 1 0.1 0 0 0.2 0 sun cream 2 0.1 0 0.1 0 0.4 sun cream 3 0 0.1 0.1 0 0.2

product name representative expression vector sun cream 1 [0.1, 0, 0, 0.2, 0, … ] sun cream 2 [0.1, 0, 0.1, 0, 0.4, … ] sun cream 3 [0, 0.1, 0.1, 0, 0.2, … ]

In this way, the review characteristic module extracted the pre-registered review representative expression from the product review of the review database, and generated the representative expression vector for each cosmetic product to create the first vector.

Next, a second vector is generated by vectorizing each word by calculating the probability of occurrence of words in the product review. The well-known Doc2Vec modeling technique was used as an algorithm for vectorizing a document by averaging vectors of words appearing in a document. As a result, it is shown in Table 6.

product name review Doc2Vec sun cream 1 The whiteness is really bad ㅠㅠ
I think it's better to mix it with oil or essence because it doesn't apply well.
I use it considering that it has minimized harmful ingredients, but what should I really do about the white turbidity?? [0.3, 0.45, 0.5, … ] sun cream 2 There is a little white turbidity.
Immediately after application, the hair sticks slightly!
As time passes, it becomes less sticky and rather holds sebum, so if you have dry skin, it may feel dry.
It's an inorganic sunscreen and the spreadability is fine. [0.8, 0.45, 0.2, … ] sun cream 3 I was worried that there was a white cloudiness, but surprisingly it wasn't too bad.
The spreadability is also much better than I expected! No irritation, no trouble, good [0.64, 0.23, 0.38, … ]

The product review is vectorized by connecting the first vector (representative expression vector) and the second vector (review document vector) (S210). The final result is shown in Table 7.

product name final vector sun cream 1 [0.3, 0.45, 0.5, … , 0.1, 0, 0, 0.2, 0] sun cream 2 [0.8, 0.45, 0.2, … , 0.1, 0, 0.1, 0, 0.4] sun cream 3 [0.64, 0.23, 0.38, … , 0, 0.1, 0.1, 0, 0.2]

Next, the similarity between products is calculated (S220). Preferably, the cosine similarity is calculated using Equation 1.

(식 1)

(Equation 1)

The cosine similarity is calculated and sorted based on the similarity (S230). The results are shown in Table 8 below.

product Products subject to calculation Similarity
(Cosine similarity) Similar ranking
(review characteristics) sun cream 1 sun cream 3 0.7 One sun cream 1 sun cream 2 0.5 2 sun cream 2 sun cream 3 0.8 One sun cream 2 sun cream 1 0.5 2 sun cream 3 sun cream 2 0.8 One sun cream 3 sun cream 1 0.7 2

In this way, by quantifying the cosmetic similarity between products for all cosmetics, review characteristics for individual cosmetics can be managed as a table, and it can be stored as a review-based similar product database. Theoretically, if K (K is an integer greater than 1) cosmetic products are registered in the system, a review characteristic table can be created while K-1 other products are given cosmetic similarity between products for each cosmetic product. In the review characteristic table, similar rankings are given to each other product, and matching cosmetic information may be selected in the order of high similarity rankings.

5 shows an example of a cosmetic recommendation screen 150 according to a preferred embodiment of the present invention.

The mobile device launches the cosmetics application to generate cosmetic inquiry events multiple times. Then, as shown in FIG. 5 , the recommendation button 151 provided on the application screen was pressed to enter the cosmetic recommendation screen 150 .

In the cosmetic recommendation screen 150, the search cosmetic information and the matching cosmetic information are simultaneously visualized and displayed through the query cosmetic and matching cosmetic simultaneous display area 152 in which the phrase “recommended recommendation of this product” is displayed. The query cosmetic and matching cosmetic display area 152 includes a query cosmetic display area 153 and matching cosmetic display areas 154 , 155 , 156 .

The simultaneous display area 152 for inquiry cosmetics and matching cosmetics may be designed in various formats, such as text, video, and audio. In the embodiment of Fig. 5, only one inquiry cosmetic was selected and displayed as text. In general, it may be indicated by a cosmetic name, such as "Round Lab - 1025 Dokdo Toner". Also, in the example screen, three matching cosmetics are displayed. In general, the cosmetic name and packaging image may be displayed together.

In the cosmetic recommendation screen 150 , recommended cosmetics of different types may be further displayed in the upper area 158 and the lower area 159 of the simultaneous display area 152 for the searched cosmetics and the matching cosmetics.

As described above, the user interface method for recommending cosmetics by applying two criteria has been described. The first criterion is the user's recent history of inquiring about cosmetics before. The second criterion is the similarity of the review characteristics for the cosmetics selected by the first criterion. The similarity of the review characteristics is digitized in the review characteristics table and registered by rank as described above.

For reference, the user interface method for recommending cosmetics by applying two criteria in the cosmetics information providing mobile application according to an embodiment of the present invention is implemented in the form of a program command that can be executed through various computer means and is computer readable may be recorded on the medium. The computer-readable medium may include program instructions, data files, data structures, etc. alone or in combination. The program instructions recorded on the medium may be specially designed and configured for the present invention, or may be known and available to those skilled in the art of computer software.

Examples of computer readable media include hard disks, magnetic media such as floppy disks and magnetic tapes, optical recording media such as CD-ROMs and DVDs, magneto-optical media such as floppy disks, and ROM, RAM, A hardware device specifically configured to store and execute program instructions, such as flash memory, may be included. Examples of program instructions include not only machine code such as generated by a compiler, but also high-level language code that can be executed by a computer using an interpreter or the like. The hardware devices described above may be configured to operate as one or more software modules to perform the operations of the present invention, and vice versa.

The protection scope of the present invention is not limited to the description and expression of the embodiments explicitly described above. In addition, it is added once again that the protection scope of the present invention cannot be limited due to obvious changes or substitutions in the technical field to which the present invention pertains.

Claims (3)

As a user interface method of a cosmetic information providing mobile application,
(1) The mobile device executes the cosmetic information provision mobile application to generate N (N is an integer greater than 1) cosmetic inquiry event, and then selects a selectable button on the user interface of the cosmetic information provision mobile application. The step of entering the cosmetic recommendation screen,
(2) Products with similar review characteristics to the inquiry cosmetics displayed as the inquiry cosmetic information in the review-based similar product database managed by the server while displaying the inquiry cosmetic information extracted in response to the cosmetic inquiry event on the mobile application screen extracting the matching cosmetic information stored in advance by matching with , and displaying the information on the cosmetic recommendation screen together with the inquiry cosmetic information, so that the self-interest recommendation element and the other person's interest recommendation element are simultaneously displayed on the mobile application screen,
The review-based similar product database is represented as a result of the server matching the topic set included in the product review of the user review database and the pre-registered review representative expression set using the dictionary base including the topic-polarity-representative expression dictionary An expression vector is generated for each cosmetic to create a first vector, and the probability of occurrence of words included in the same product review is calculated separately from the representative expression vector, and each word is vectorized to create a second vector for each cosmetic, After connecting the first vector and the second vector, the cosmetic similarity between products is calculated and quantified as a review characteristic for each product.
The interest recommendation element of others displayed on the cosmetic recommendation screen is cosmetic information determined by the quantified similarity of the review characteristics,
A user interface method of a mobile application for providing cosmetic information.
delete delete

Images