KR20210022873A - Method and system for providing improved review information to users by analyzing cosmetic review information collected through the Internet - Google Patents

Abstract

The present invention relates to a method for providing improved review information to a user by analyzing cosmetic review information collected through the internet. According to one aspect of the present invention, the method comprises: a step of primarily collecting a plurality of reviews for a specific cosmetic product posted on a specific website; a first analysis step of dividing and quantifying image information in the primarily collected review into categories including at least color development, product container, and packaging, and quantifying text information by dividing the text information into at least positive and negative; a step of selecting a priority exposure target review having a high numerical value among the quantified image information and the quantified text information; and a step of rearranging the reviews for the specific cosmetic product posted on the specific website so that the priority exposure target review is shown to the user first or shown with a relatively greater weight compared to other reviews.

Description

{Method and system for providing improved review information to users by analyzing cosmetic review information collected through the Internet}

The present invention relates to a method and system for providing improved review information to a user by analyzing cosmetic review information collected through the Internet.

As interest in beauty increases, consumption of various cosmetics is increasing. In addition, with the development of e-commerce, the demand to purchase cosmetics online is increasing.

However, cosmetics are products that are used while being applied directly to the user's body, and when used, they have a direct/indirect effect on their appearance, so consumers who want to purchase cosmetics online should review cosmetics before purchasing cosmetics. They tend to make purchase decisions by referring to them.

For example, a consumer who wants to purchase a cosmetic product can use the Internet by other users to find out whether the color of the cosmetic product to be actually purchased is the same as that of the cosmetic product shown in the online product advertisement, and what advantages and disadvantages are compared to other products. You can make a purchase decision by referring to the review uploaded above.

However, since reviews uploaded on the Internet have many repetitive expressions and are displayed in the order of recent uploads, regardless of the quality of the reviews, it takes a long time to find a review that can help to grasp product information or purchase a product. There is a drawback of being caught.

In addition, there is a problem in that the seller cannot find a review containing the information that the user wants, and thus it does not lead to product purchase, resulting in a decrease in sales revenue.

Patent Document: Korean Patent Registration No. 10-1677128

Embodiments of the present invention have been proposed to solve the above problems, and are collected through the Internet that can improve customer service experience by providing high-quality, high-quality reviews to customers, and ultimately contribute to sales improvement of sellers. It is intended to provide a method and system for providing improved review information to users by analyzing the cosmetic review information.

According to an embodiment of the present invention, a method comprising: first collecting a plurality of reviews for a specific cosmetic product posted on a specific website by at least one of a text review collection unit and an image review collection unit; In the first collected reviews, image information is divided into categories including at least color development, product container, and packaging by the image review analysis unit and quantified, and text information is classified to include at least affirmations and negatives by the text review analysis unit. A first analysis step of digitizing; Selecting a priority exposure target review having a high numerical value from among the image information digitized by a review comparison unit and the digitized text information; And rearranging the reviews for the specific cosmetics posted on the specific website such that the first exposure target review is first displayed to the user or a relatively larger weight compared to other reviews by a review rearrangement unit. A method of analyzing cosmetic review information collected through the Internet and providing improved review information to a user may be provided.

In addition, after the step of rearranging the reviews, secondary collection of a plurality of reviews for the specific cosmetics posted on the specific website by at least one of a text review collection unit and an image review collection unit; A second analysis step of storing text information included in the secondly collected review in a database, analyzing the data stored by the text review analysis unit, and dividing the text information to include at least positives and negatives and quantifying them; And comparing the digitized result value of the text information obtained in the first analysis step with the digitized result value of the text information obtained in the second analysis step by a review comparison unit and providing it to the administrator of the specific website. A method of providing improved review information to a user by analyzing cosmetic review information collected through the Internet may be provided.

In addition, the step of digitizing text information includes morpheme analysis of the text information by the text review analysis unit, and matching the extracted word or phrase with the word or phrase stored in the user dictionary to affirm the matched word or phrase, Classifying to include negation; Assigning parts of speech to words extracted through morpheme analysis by the text review analysis unit; Assigning a preset weight to the parts of speech extracted by the text review analysis unit; A method of providing improved review information to a user by analyzing cosmetic review information collected through the Internet, including the step of digitizing the text information by adding a value obtained by multiplying the weight of each part-of-speech extracted by the text review analysis unit and the number of characters Can be provided.

In addition, in the step of digitizing the image information, the image review analysis unit assigns the highest weight to the color development category among the image information, assigns the second highest weight to the product container category, and assigns the lowest weight to the packaging. A method of providing improved review information to a user by analyzing cosmetic review information collected through the Internet may be provided.

In addition, in the step of digitizing the image information, the image information collected by the image review analysis unit is classified into one of a color development category, a product container category, and a packaging category, and the weight of each category is added to the inference rate of the classified category image. A method of providing improved review information to a user by analyzing cosmetic review information collected through the Internet, including the step of multiplying and quantifying the image information, may be provided.

In addition, the second collection of the reviews and the second analysis of text information include,

Analyzing cosmetic review information collected through the Internet, including the step of executing after a preset time has elapsed or a preset number of reviews has been uploaded to the web server after the priority exposure target review is rearranged on the website. A method of providing improved review information to a user may be provided.

In addition, in the first analysis step, when image information and text information exist together in one review collected first, a numerical value obtained from the image information by at least one of a text review analysis unit or an image review analysis unit, and the A method of providing improved review information to a user by analyzing cosmetic review information collected through the Internet including the step of summing numerical values obtained from text information may be provided.

In addition, the step of selecting a review subject to exposure first includes selecting a review belonging to a color development category of image information as a review subject to exposure priority over a review including only text information by a review comparison unit. A method of analyzing cosmetic review information and providing improved review information to a user may be provided.

According to another embodiment of the present invention, in a system for providing improved review information to a user by analyzing cosmetic review information collected through the Internet including a review analysis unit, the review analysis unit is posted on a specific web site. A text review collection unit that primarily collects text information of a plurality of reviews for a specific cosmetic product that is present; An image review collection unit that primarily collects image information of a plurality of reviews for a specific cosmetic product posted on a specific website; An image review analysis unit for dividing and quantifying image information into categories including at least color development, product container, and packaging in the first collected review; A text review analysis unit that divides and quantifies text information to include at least positives and negatives in the first collected reviews; A review comparison unit for selecting a review subject to be exposed with a higher value from among the digitized image information and the digitized text information; And a review rearrangement unit that rearranges the reviews for specific cosmetics posted on a specific website so that the reviews subject to priority exposure are shown to the user first or in a relatively larger proportion than other reviews. A system for analyzing review information and providing improved review information to a user may be provided.

In addition, the text review collection unit includes secondary collection of a plurality of reviews for the specific cosmetics posted on the specific website after relocating the reviews, and the text review analysis unit And dividing and digitizing the text information to include at least positive and negative from the text information, and the review comparison unit digitized result values of text information obtained from the first collected review and obtained from the second collected review. A system may be provided for providing improved review information to a user by analyzing cosmetic review information collected through the Internet, including comparing the numerical result value of text information and providing it to an administrator of a specific web site.

The method and system for providing improved review information to users by analyzing cosmetic review information collected through the Internet according to embodiments of the present invention improves the customer's service use experience by providing high-quality, high-quality reviews to customers. , Ultimately, it can contribute to improving sellers' sales.

1 is a diagram showing a system for providing improved review information according to an embodiment of the present invention.
FIG. 2 is a diagram schematically showing the configuration of the review analysis unit of FIG. 1.
3 is a flowchart illustrating a method of providing improved review information to a user using the review analysis unit of FIG. 2.
4 is a flowchart illustrating a process in which text information is quantified through morpheme analysis by the text review analysis unit of FIG. 2.
FIG. 5 is a diagram schematically illustrating reviews appearing on a web site before and after reviews are rearranged by the review analysis unit of FIG. 1.

Hereinafter, specific embodiments of the present invention will be described in detail with reference to the drawings.

In addition, in describing the present invention, when it is determined that a detailed description of a related known configuration or function may obscure the subject matter of the present invention, a detailed description thereof will be omitted.

1 is a diagram showing a system for providing improved review information according to an embodiment of the present invention.

Referring to FIG. 1, a system 1 for providing improved review information according to an embodiment of the present invention may include a user terminal 10, a web server 20, and a review analysis unit 30. Here, the web server 20 and the review analysis unit 30 may be configured as one physical device, and are provided as independent devices to enable direct data communication or enable data communication through the communication network 40. It can also be configured. In the present embodiment, the web server 20 and the review analysis unit 30 will be described as an example that is operated as separate and independent servers. In addition, when the review analysis unit 30 is operated as a separate and independent server, the review analysis unit 30 may be understood as a device that provides a review analysis result to a seller.

The user terminal 10, the web server 20, and the review analysis unit 30 may be connected to each other to transmit and receive data through the communication network 40.

In this embodiment, the system 1 for providing improved review information can provide a high-quality review related to a product to a user who wants to purchase a product, and visualize and provide the review analysis result to the seller. The functionality can be implemented by the review analysis unit 30.

For example, when a plurality of users upload a review to the website of the web server 20 that sells cosmetics through the user terminal 10, the review analysis unit 30 performs the first review uploaded to the website. By collecting and analyzing high-quality reviews, you can rearrange them and show them to users. Thereafter, when a preset condition is satisfied (for example, after a preset time has elapsed or a preset number of reviews is uploaded to the web server 20), the review analysis unit 30 is uploaded to the website. By secondary collection and analysis of reviews, the results of comparison between the primary and secondary reviews can be visualized and provided to the seller.

The user terminal 10 is a device provided to allow a user to input a review on cosmetics, and may be a fixed terminal implemented as a computing device or a mobile terminal. For example, the user terminal 10 may be a computer, a notebook computer, a tablet PC, a smart phone, a mobile phone, a wearable device, or the like. However, it is not limited to this example, and the user terminal 10 may be any terminal having a function of allowing the user to input information necessary for writing a review and displaying the review to the user.

The web server 20 may provide a web site for selling cosmetics, and the web server 20 may store product information for cosmetics to be sold and provide product information to a user. In addition, when a user inputs a review on cosmetics through the user terminal 10, the review on cosmetics may be stored in a database (not shown) separately provided in the web server 20, and a review analysis unit to be described later. It can also be stored in the database of (30).

2 is a diagram schematically showing the configuration of the review analysis unit 30 of FIG. 1. Referring to FIG. 2, the review analysis unit 30 may include an input/output interface 100, a memory 200, a processor 300, and a communication module 400.

The processor 300 may be configured to process instructions of a computer program by performing basic arithmetic, logic, and input/output operations. Instructions may be provided to the processor 300 from the memory 200 or the communication module 400. In addition, the command may be provided to the processor 300 through a bus, which is a communication channel between each of the components constituting the review analysis unit 30.

The processor 300 rearranges high-quality reviews and performs various functions such as input/output of data, data processing, data management, communication using the communication network 40, etc. necessary to visualize and provide the results of the comparison of the reviews to the seller. This can be done, and specific components of the processor 300 for this will be described later.

Components of the processor 300 may be functional modules implemented with program codes stored in the memory 200.

The memory 200 is a computer-readable recording medium and may include a permanent mass storage device such as a random access memory (RAM), a read only memory (ROM), and a disk drive. The processor 300 may load a program code stored in the memory 200 to implement a system that provides improved review information. The program code may be loaded from a recording medium (eg, DVD, memory card, etc.) readable by a separate computer, or transmitted from another device through the communication module 500 and stored in the memory 200.

In addition, the memory 200 may be provided with a database 210 capable of relocating high-quality reviews and storing data necessary to visualize and provide a result of comparing the reviews to sellers. That is, the memory 200 may be understood to include the database 210. The database 210 may be a software independently built database as well as any type of storage capable of storing data.

The communication module 400 may provide a function for the review analysis unit 30 and the user terminal 10 or the review analysis unit 30 and the web server 20 to communicate with each other through the communication network 40.

The input/output interface 100 is a component that receives data related to reviews from a user or an administrator, or outputs data provided from the processor 300, memory 200, and communication module 400 to a user or an administrator. It can be understood to include both the hardware interface of the side and the software interface of the logical side. For example, the input/output interface 100 may include a device such as a keyboard, a mouse, and a microphone as an input device, and may include a device such as a display or a speaker as an output device. As another example, the input/output interface 100 may include a device in which input and output are integrated into one, such as a touch screen, and an input tool such as a touch pen.

The processor 300 of the review analysis unit 30 includes a text review collection unit 310, an image review collection unit 320, a text review analysis unit 330, an image review analysis unit 340, and a review rearrangement unit 350. And a review comparison unit 360, a detailed description of which will be described later.

3 is a flowchart illustrating a method of providing improved review information to a user using the review analysis unit 30 of FIG. 2.

2 and 3, when a review on cosmetics is input through the user terminal 10, text information and image information of the review are divided by the text review collection unit 310 and the image review collection unit 320 It can be collected and stored in the primary (S10). For example, when a review in which an image and text are mixed is input from the user terminal 10, the text review collection unit 310 collects and stores only text information among the input reviews, and the image review collection unit 320 Among the input reviews, only image information can be collected and stored.

Thereafter, the text review analysis unit 330 analyzes the text information first collected by the text review collection unit 310, and the image review analysis unit 340 is the first collection by the image review collection unit 320 The image information can be analyzed (S20).

Specifically, the text information collected by the text review analysis unit 330 may include an emotion processing step, and the emotion processing step includes a process of determining the collected text information as positive and negative, and quantifying the collected text information. It may include at least one of the processes, and a detailed description thereof will be described later. Here, the emotion processing step may be understood as applying an emotion classification algorithm (eg, a hybrid method) after undergoing a pre-processing of removing unnecessary characters such as special characters from text information and processing spaces. The emotion classification algorithm can be understood to quantify text information by assigning weights for each part of speech through a synonym processing using a synonym dictionary and a morpheme analysis using a user dictionary.

Synonym processing using the synonym dictionary can be understood as a process of converting words or expressions included in text information into preset synonyms by matching words or expressions included in text information with words or expressions registered in the synonym dictionary. . Data on the synonym dictionary may be stored in the database 210, and words or expressions in the synonym dictionary may be updated at a preset period. In addition, words or expressions related to cosmetics may be registered in the synonym dictionary.

In addition, text information by the text review analysis unit 330 may be quantified by passing through a morpheme analysis process using a user dictionary and assigning weights for each part-of-speech. The morpheme analysis process may be understood as a process of separating text information for each sentence and then extracting words for each sentence.

In addition, preset expressions (for example, emotion phrases (word + phrase)) may be numerically converted into values indicating positive or negative polarity and intensity thereof, and stored in the user dictionary. For example, the emotion phrase indicating the polarity of the positive may be numerically converted into a positive number (eg, +1) and stored in the user dictionary as an expression such as'pretty, good, soft'. In addition, the emotional phrase indicating the polarity of the negative may be numerically converted into a negative number (for example, -1) and stored in the user dictionary as an expression such as “heavy, poor, expensive”. Here, data of the user dictionary may be stored in the database 210.

Using the data stored in the user dictionary, text information may be numerically converted by the text review analysis unit 330, and a more detailed description thereof will be described later. The image review analysis unit 340 infers the possibility of including the image information collected by the image review collection unit 320 into a color image, product container image, and packaging image, and selects the most likely image as a color development category and product container category. , The image information can be quantified by classifying it into one of the packaging categories, and assigning a weight for each category to the inference rate of the classified category image. For example, the weight of the color development category may be 10, the weight of the product container category may be 3, and the weight of the packaging category may be 1. However, these weighted scores are exemplary, and the weights may be set and changed according to factors that are considered important by cosmetic sellers. Here, the weights for each category may be preset in the database 210. A detailed description of a process of quantifying image information by the image review analysis unit 340 is as follows.

When a user uploads a review including an image to a web page of the web server 20, image information is collected and stored by the image review collecting unit 320, and the image collected by the image review analyzing unit 340 The likelihood that the information will be included in the color image, product container image, and packaging image can be deduced. For example, when a review of image A including an image of color development is collected, the image review analysis unit 340 infers that it is a color image of 60%, a product container image of 30%, and a packaging image of 10%. Can be. Here, the sum of the inference rates of the color image, product container image, and packaging image is 100%.

After that, since the image review analysis unit 340 determines that the image A review has the highest probability of being a color image (inference rate 60), it may be classified into a color development category.

After that, the image A classified as a color development category by the image review analysis unit 340 may be numerically multiplied by a weight of the color development category and an inference rate. For example, by multiplying the weight 10 of the color development category and the inference rate of the color image 60, the image A may be numerically converted to 600.

As another example, if the B image review by the image review analysis unit 340 is 70% likelihood that the container image is a container image (inference rate 70), the container category weight 3 and the reasoning rate 70 are multiplied, so that the B image review will be numerically converted to 210. I can. Meanwhile, the image information classified in the color development category may be an image on which a cosmetic material is applied to the user's skin. For example, image information classified into a color development category may be a review including lip color, eye shadow color, and blusher color.

Image information classified into a product container category may be an image of a product container. For example, image information classified into a product container category may be a review including a lip product, an eye shadow product, and a blusher product.

Image information classified into a packaging category may be an image of a box packaging a product container. For example, image information classified into a packaging category may be a review including image information of a box, a delivery box, and a cosmetic box.

When users purchase cosmetics, they tend to value information related to the color of cosmetics, so a high weight may be set for reviews classified by color development categories.

In addition, when a user purchases a cosmetic product, the user tends to regard the information on the box packaging the cosmetic product as unnecessary information, and thus a low weight may be set for reviews classified by packaging category.

In addition, when a user purchases a cosmetic product, information on a product container tends to be regarded as the second most important information after color development, and an intermediate weight may be set for reviews classified into product container categories.

In addition, deep learning may be used to classify image information into images of color development, product containers, and packaging. For example, by classifying photos of eyes, lips, skin, product containers, product packaging, and delivery boxes, deep learning learning processing for color development, product containers, and product packaging, and lip learning learning content for image information Based on the image classifier, it can be classified into categories of color development, product containers, and packaging. Here, the deep learning learning may be image learning and reasoning using Python, Google tensorflow, and image/data processing tools.

On the other hand, when image information and text information exist together in one review, a value obtained from the image information through the text review analysis unit 330 or the image review analysis unit 340 and a value obtained from text information may be summed. .

When text information and image information are quantified by the text review analysis unit 330 and the image review analysis unit 340, the first exposure target review with a higher numerical value among the image information and text information digitized by the review comparison unit 360 May be selected (S30). For example, if the numerical value of text information or image information is 1000 or more by the review comparison unit 360, it may be classified as an excellent review, and first, may be selected as an exposure target review. In the case of image information, when the probability of inferring that it is a color image is 100%, it can be numerically converted to 1000, so it can be selected as a first exposure target review.

Here, the value 1000 is an example, and if the value of the text information is greater than or equal to a preset value by the review comparison unit 360, and the value of the image information is greater than or equal to the preset value, it is classified as an excellent review and first selected as an exposure target review. You may.

Thereafter, the reviews for specific cosmetics posted on a specific website may be rearranged by the review rearrangement unit 350 so that the first exposure target review is first shown to the user or is shown with a relatively larger weight compared to other reviews. (S40).

For example, a review (Review C) for image information including lip color is digitized as 600, a review on text information with polarity including negative (Review A) is digitized as 100, and includes positives. If the review (B review) of text information having the polarity of being 300 is numerically placed in the order of C review-B review-A review, or A review is placed relatively largest and B review is placed relatively smallest Can be. In this case, when the user accesses the web site that sells cosmetics of the web server 20 through the user terminal 10 and searches for a review, first, the review A, which is the subject review, is viewed first or the most prominent. I can.

In this way, by arranging the reviews that are actually needed by the user or reviews that have a positive effect on the user's cosmetics sales so that they are clearly visible to the user, the user can obtain the necessary information when purchasing cosmetics, and the seller can increase the profits from the cosmetics sales. I can.

First, after the exposed review is rearranged on the web site of the web server 20, when a preset time elapses or a preset number of reviews is uploaded to the web server 20, the text review collection unit 310 and image review The text information and image information of the review may be classified by the collection unit 320 to be secondarily collected and stored (S50).

Thereafter, the text review analysis unit 330 analyzes the text information that is secondly collected by the text review collection unit 310, and the image review analysis unit 340 collects the second time by the image review collection unit 320 The image information is analyzed (S60). Specifically, the text information secondly collected by the text review analysis unit 330 is converted into a numerical value through an emotion processing step, and the image review analysis unit 340 uses the second collected image information to develop color, product container, and packaging categories. It can be divided into and quantified by classifying each category.

The above-described review collection of S10 and S50 and review analysis of S20 and S60 may be substantially the same process. However, it can be understood that there is a difference in time whether the review was written first or written after the collection/analysis of reviews. Here, a review written first can be defined as a primary review, and a review written after the primary reviews are rearranged can be defined as a secondary review. The procedures for S50 and S60 are the same as those for S10 and S20, respectively, and detailed descriptions thereof will be omitted.

Thereafter, the analysis result value of the first review and the analysis result value of the second review may be compared by the review comparison unit 360 (S70). For example, a numerical result value of text information of a first review and a numerical result value of text information of a second review may be compared.

When comparing the analysis result value of the first review and the analysis result value of the second review, the image information of the review is difficult to use as an index indicating the user's product use satisfaction, so the result value of the image information of the review is excluded, Only the results of the text information of the review can be compared.

Thereafter, the comparison result of the first review and the second review by the review comparison unit 360 may be visualized and displayed to a manager who sells cosmetics (S80).

For example, the figures for the first review and the figures for the second review can be compared and visualized to the manager. Here, visualization may mean that it appears through means such as a diagram or a graph.

In addition, product purchase rates before and after reviews are relocated may be visualized to the manager.

On the other hand, the process of determining the text information collected by the text review analysis unit 330 as positive and negative will be described in more detail as follows.

First, the text review analysis unit 330 may perform a process of first classifying the text information of the review into positive and negative based on data tagged as positive and negative in the user's dictionary (step 1).

Thereafter, a differential emotional value is assigned to a word or phrase included in the text information of the review using a database-generated user dictionary, and the values are summed and classified as positive if it is positive, and negative if it is negative (2). step).

After that, by using the classification values and the derived parameters in steps 1 and 2 as data, deep learning learning can be performed to finally determine positive and negative (step 3).

In addition, a process of digitizing text information collected by the text review analysis unit 330 described above will be described in detail with reference to FIG. 4.

FIG. 4 is a flow chart illustrating a process in which text information is digitized through morpheme analysis by the text review analysis unit 330 of FIG. 2.

4, a morpheme analysis is performed on text information by the text review analysis unit 330 (S210). Thereafter, parts of speech are assigned to words extracted through morpheme analysis (S220), and preset weights may be assigned to parts of speech necessary for emotional processing among these parts of speech. (S230). For example, a weight of 12 may be given to nouns and adjectives that express objects and emotions well, a weight of 7 may be given to verbs and adverbs, and a weight of 3 may be given to others.

Thereafter, the weight of each part-of-speech is multiplied by the number of characters of the morpheme, and the sum of the calculated values of each morpheme in the text information of the review is converted into a value of the corresponding text information (S240). For example, if there is a review of “Like”, the weight of the adjective is multiplied by 12 and the number of characters 3 (12X3), and the review of “Like” can be quantified as 36.

As another example, if there is a review of 『Really Like』, the adverb's weight 7 and the number of characters are multiplied for 『Really』, and the adjective's weight of 12 and the number of characters 3 times the value of 『Like』 is added. (7X2 + 12X3), the review can be quantified as 50.

As another example, if there is a review of 『Really really good』, one 『Really』 of the 2nd 『Really』 is recognized as an unnecessary value and is excluded or numerically converted to 0, and the weight of the adverb is 7 for the remaining 『Really』. And the number of characters are multiplied by 2, and the value obtained by multiplying the adjective's weight of 12 and the number of characters 3 for 『Like』 is added (0 + 7X2 + 12X3), and the corresponding review can be quantified as 50. In the case of digitizing through morpheme analysis as described above, the fidelity of the text can be grasped in a simple manner through a user dictionary or the like.

Hereinafter, the physical configuration of the above-described review analysis unit 30 will be described.

The text review collection unit 310 of the review analysis unit 30 may collect text information of a review and store it in the database 210.

The image review collection unit 320 may collect image information of a review and store it in the database 210.

The text review analysis unit 330 may analyze text information to quantify text information.

The image review analysis unit 340 may analyze the image information and convert the image information into a numerical value.

The review rearrangement unit 350 may first rearrange reviews for specific cosmetics posted on a specific website so that the exposed review is first shown to the user or shown with a relatively larger weight compared to other reviews.

The review comparison unit 360 may compare the analysis result value of the first review and the analysis result value of the second review, and may visualize and display the comparison result of the first review and the second review to a manager who sells cosmetics.

FIG. 5 is a diagram schematically illustrating reviews appearing on a web site before and after reviews are rearranged by the review analysis unit of FIG. 1.

Referring to FIG. 5, before the reviews are rearranged, reviews are posted on the website in the order of the most recent reviews. For example, a review on image information including lip color (C review) was first written in chronological order, a review on text information having polarity including positive (B review) was written second, and negative If a review (Review A) for text information having a polarity including A is recently written, it is shown to the user in the order of A review-B review-C review before the review is rearranged.

When reviews are rearranged by the review analysis unit 30 according to an embodiment of the present invention, the review (C review) for image information including lip color has the highest value and has a polarity including affirmative. Since reviews for text information (B review) have the second highest value, and reviews for text information with polarity including negative (A review) have the lowest value, the order of C review-B review-A review. Or, review C may be placed relatively largest and review A may be placed relatively smallest.

In the above, a method and a system for providing improved review information to a user by analyzing cosmetic review information collected through the Internet according to an embodiment of the present invention have been described as specific embodiments, but this is only an example, and the present invention is limited thereto. It is not, and should be construed as having the widest scope in accordance with the basic ideas disclosed herein. A person skilled in the art may combine and replace the disclosed embodiments to implement a pattern having a shape that is not indicated, but this also does not depart from the scope of the present invention. In addition, those skilled in the art can easily change or modify the disclosed embodiments based on the present specification, and it is clear that such changes or modifications also belong to the scope of the present invention.

1: System that provides improved review information
10: user terminal 20: web server
30: review analysis unit 40: communication network
100: input/output interface 200: memory
210: database 300: process
310: text review collection unit 320: image review collection unit
330: text review analysis unit 340: image review analysis unit
350: Review relocation section 360: Review comparison section
400: communication module

Claims (10)

First collecting a plurality of reviews for a specific cosmetic product posted on a specific website by at least one of a text review collection unit and an image review collection unit;
In the first collected reviews, image information is divided into categories including at least color development, product container, and packaging by the image review analysis unit and quantified, and text information is classified to include at least affirmations and negatives by the text review analysis unit. A first analysis step of digitizing;
Selecting a priority exposure target review having a high numerical value from among the image information digitized by a review comparison unit and the digitized text information; And
Relocating the reviews for the specific cosmetics posted on the specific website so that the first exposure target review is shown to the user first by the review rearrangement unit or is shown with a relatively larger weight compared to other reviews.
A method of providing improved review information to users by analyzing cosmetic review information collected through the Internet. The method of claim 1,
After the step of relocating the review,
Secondly collecting a plurality of reviews for the specific cosmetics posted on the specific website by at least one of a text review collection unit and an image review collection unit;
A second analysis step of storing text information included in the secondly collected review in a database, analyzing the data stored by the text review analysis unit, and dividing the text information to include at least positives and negatives and quantifying them; And
Comprising the step of comparing the digitized result value of the text information obtained in the first analysis step and the digitized result value of the text information obtained in the second analysis step by a review comparison unit and providing it to the administrator of the specific website
A method of providing improved review information to users by analyzing cosmetic review information collected through the Internet. The method according to claim 1 or 2,
The step of quantifying text information is:
Performing morpheme analysis on text information by a text review analysis unit, matching the extracted word or phrase with the word or phrase stored in the user dictionary, and classifying the matched word or phrase to include affirmative or negative;
Assigning parts of speech to words extracted through morpheme analysis by the text review analysis unit;
Assigning a preset weight to the parts of speech extracted by the text review analysis unit;
Comprising the step of digitizing the text information by adding a value obtained by multiplying the weight of each part of speech extracted by the text review analysis unit and the number of characters.
A method of providing improved review information to users by analyzing cosmetic review information collected through the Internet. The method of claim 1,
The step of quantifying the image information,
Including assigning the highest weight to the color development category among the image information by the image review analysis unit, assigning the second highest weight to the product container category, and giving the lowest weight to the packaging category.
A method of providing improved review information to users by analyzing cosmetic review information collected through the Internet. The method of claim 4,
The step of digitizing the image information,
The image information collected by the image review analysis unit is classified into one of a color development category, a product container category, and a packaging category,
Comprising the step of multiplying the inference rate of the classified category image by the weight for each category to quantify the image information
A method of providing improved review information to users by analyzing cosmetic review information collected through the Internet. The method of claim 2,
The second collection of the reviews and the second analysis of text information include,
Including the step of executing after the priority exposure target review is rearranged on the website, a preset time has elapsed or a preset number of reviews is uploaded to the web server,
A method of providing improved review information to users by analyzing cosmetic review information collected through the Internet. The method of claim 1,
In the first analysis step,
When image information and text information exist together in one of the first collected reviews, adding a value obtained from the image information and a value obtained from the text information by at least one of a text review analysis unit or an image review analysis unit Including
A method of providing improved review information to users by analyzing cosmetic review information collected through the Internet. The method of claim 1,
First of all, the step of selecting a review target for exposure is:
Further comprising the step of selecting, by the review comparison unit, a review belonging to a color development category of image information as a review subject to exposure prior to a review including only text information.
A method of providing improved review information to users by analyzing cosmetic review information collected through the Internet. In a system for providing improved review information to a user by analyzing cosmetic review information collected through the Internet including a review analysis unit,
The review analysis unit,
A text review collection unit that first collects text information of a plurality of reviews for a specific cosmetic product posted on a specific website;
An image review collection unit that primarily collects image information of a plurality of reviews for a specific cosmetic product posted on a specific website;
An image review analysis unit for dividing and quantifying image information into categories including at least color development, product container, and packaging in the first collected review;
A text review analysis unit for dividing and quantifying text information to include at least positives and negatives in the first collected reviews;
A review comparison unit for selecting a review subject to be exposed with a higher value from among the digitized image information and the digitized text information; And
Including a review rearrangement unit that rearranges the reviews for specific cosmetics posted on a specific website so that the first exposed review is shown to the user first or in a relatively larger proportion compared to other reviews,
A system that analyzes cosmetic review information collected through the Internet and provides improved review information to users. The method of claim 9,
The text review collection unit,
After relocating the reviews, including secondary collection of a plurality of reviews for the specific cosmetics posted on the specific website,
The text review analysis unit,
And dividing and digitizing to include at least positive and negative from text information in the second collected review,
The review comparison unit,
Comprising comparing the digitized result value of the text information obtained from the firstly collected review with the digitized result value of the text information obtained from the secondly collected review and providing it to the administrator of a specific website
A system that analyzes cosmetic review information collected through the Internet and provides improved review information to users.

Images