KR101806169B1 - Method, apparatus, system and computer program for offering a shopping information - Google Patents

Abstract

The present invention relates to a method used by a server for providing shopping information. The method includes: step A of collecting product information from an arbitrary online market; step B of extracting product images from the product information clustering one or more products with a similarity value between product images exceeding a preset range into a group; and step C of extracting the text information of the products included in the product group, executing a natural language processing operation with the text information to generate reference product names of the products in the corresponding product group, and selecting a representative image of the product group.

Description

[0001] METHOD, APPARATUS, SYSTEM AND COMPUTER PROGRAM FOR OFFERING A SHOPPING INFORMATION [0002]

The present invention relates to a method and system for generating a query for shopping information and retrieving shopping information.

In the online shopping process, buyers generally subscribe to various shopping malls, and each time they access a different shopping mall, they are required to install the application, sign up, and log in. However, as more and more online commerce is activated, many applications are installed and the management of their shopping information scattered among many shopping malls is increasingly causing a rise in the users' shopping tiredness.

For example, let's take a look at how a user sees certain fashion items worn by famous actresses in an offline magazine and buys them online. The user should first find out the name of the item or the place of sale of the item. To this end, the user accesses a search site and enters a keyword such as a magazine name, an actress name, an event name, an item type, and a color, and finds out a product name or a manufacturer.

If you do not get the information you want, you can go to the website of the magazine, search for the article, describe the item by commenting on the article, and ask for the name of the product or the vendor.

At the end of the twists and turns, the user who finds the product name can search for the lowest price online using the product? M. To this end, the user can access the price comparison site, install a price comparison application, and obtain price information of the corresponding product through membership registration and login.

However, the price information provided by the price comparison site may not be accurate. If you apply various discount coupons of online shopping mall or consider shipping cost, the lowest price offered by price comparison site can be changed. Therefore, the user will visit several shopping malls to check actual sales information.

The user who accesses the shopping mall site can install the shopping mall application according to the demand of the shopping mall site to confirm the coupon applicable price and the shipping applicable price, and can put the item into the shopping cart through membership and login. A user who repeats this many times will be able to purchase the corresponding fashion item.

Such inconvenience of search, installation of a large number of applications, repetition of the login process, etc., are factors that increase the online shopping tiredness of users.

SUMMARY OF THE INVENTION The present invention has been made to solve the above problems. In particular, the present invention relates to a method and apparatus for clustering shopping information on the basis of images and texts, and generating a representative product name to provide shopping information search results.

Further, the present invention relates to a method and an apparatus capable of generating a query inquiring about shopping information without using the URL data of the web browser, but using only the image displayed on the device.

In a server according to an embodiment of the present invention, a method for providing shopping information includes the steps of: A) collecting product information from an arbitrary online market; extracting a product image from the product information; Clustering products having a set range or more into at least one product group; And a step C for extracting text information of the goods belonging to the goods group and processing the text information by natural language to generate a reference product name of the goods belonging to the goods group and selecting the representative image of the goods group .

Further, in a server according to an embodiment of the present invention, a method for processing shopping information is a method for processing merchandise having a degree of similarity of data extracted according to a preset reference from a product information collected from an arbitrary online market, Clustering < / RTI > Calculating group feature information for each product group and designing a classification model of the product groups using the calculated group feature information; And updating the classification model by collecting feedback on the classification of the product group or the component goods included in the product group.

Meanwhile, in a client according to an embodiment of the present invention, a method of generating a query for inquiring shopping information includes: detecting a color value in a preset logo candidate region of a displayed page; An A step of determining whether or not the shopping is for shopping and detecting the logo of the online shopping; A step B of setting a price candidate region based on a location where the logo is displayed and extracting price information from the price candidate region; A step C for determining whether to generate a query for shopping information using the price information; And a step D of extracting an image of a predetermined region based on a position where the price information is displayed on the page and transmitting a query inquiring about shopping information on a product included in the image.

According to the present invention, a client device can determine the necessity of a query without accessing an online shopping mall server, and can provide information necessary for a query using an image displayed on the device. Furthermore, the shopping information search server can process the image and text information included in the query serially, thereby effectively providing more precise search results.

1 is a flowchart illustrating a process of creating a shopping information database in a shopping information search server according to an embodiment of the present invention;
2 is a view for explaining an example of generating a reference product name and a representative product image according to an embodiment of the present invention;
3 is a flowchart illustrating a process of providing shopping information on a query in a shopping information search server according to an embodiment of the present invention
4 is a flowchart for explaining a process of generating a query in a client device according to an embodiment of the present invention
5 is a diagram for explaining an example of query generation according to an embodiment of the present invention;
6 illustrates an example of a user interface for providing shopping information according to an embodiment of the present invention

It is to be understood that the present invention is not limited to the description of the embodiments described below, and that various modifications may be made without departing from the technical scope of the present invention. In the following description, well-known functions or constructions are not described in detail since they would obscure the invention in unnecessary detail.

In the drawings, the same components are denoted by the same reference numerals. And in the accompanying drawings, some of the elements may be exaggerated, omitted or schematically illustrated. It is intended to clearly illustrate the gist of the present invention by omitting unnecessary explanations not related to the gist of the present invention. Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings.

For example, you may want to consider shopping information for a specific brand digital player you've watched as you pass by.

In this case, the user will access the shopping information search service and input a search word. What the user desires is the latest and lowest price information of the item.

For example, if a user enters "A player" as a search term, a list of all player A's player models is provided. However, since the search range is too wide, the user can input the search word again as "A-player 256GB". However, since thousands of lists are searched by these search words, it is not easy for the user to obtain shopping information for the digital player before the model name is correctly inputted.

However, even if the user enters the correct model name, the list of products provided on the search result page is not one, but may be several thousand.

This is because each product name is described differently for each online shop. For example, the mall for Model AK240 is "iRiver AK-240SS [256G] Astel & nbsp AK240SS + MQS100 Song Coupon / Kangnam Sound Research Institute", B Mall is "iriver ┃ Astell & Kern AK240 Blue Note 75th Anniversary Package Astel Ancon AK240 Blue Note 75 weeks ".

In this case, under the text-based search paradigm, products of A-mall and products of B-mall can be classified into different products. Furthermore, if the C Mall is not a digital player but sells the player's LCD protective film, and if it is listed under the trade name "IRIVER ASTHEL ANAKON AK SERIES LIQUID PROTECTIVE GLASS", the C Mall protective film will also be included on the search list.

The more noise like this, the more difficult it is for the user to provide the desired search result such as the lowest price comparison, and the search quality is lowered. Therefore, conventionally, a separate manual work for eliminating text-based noise in product database creation has been required.

SUMMARY OF THE INVENTION The present invention has been made to solve the above problems and to provide a method of performing shopping information search with higher quality.

According to the embodiment of the present invention, the service server can clusters products on an image basis, and generates reference product names by processing text information such as product names for product groups whose image similarity is higher than the set range. When a query for inquiring product information is received from a client, the product information database is firstly filtered based on the product image included in the query, the product name text information of the second query is processed, It is possible to greatly reduce the noise and improve the search quality by searching for products in a similar manner.

The details of the shopping information search method according to the embodiment of the present invention will be described later with the description of the attached drawings.

Hereinafter, it is assumed that the client is a mobile device, but the present invention is not limited thereto. That is, the client device in the present invention should be understood as a concept including all kinds of electronic devices capable of requesting product search, such as a desktop, a smart phone, a tablet PC, and displaying search information.

Further, the term " page displayed on the device " in the present specification can be understood as a concept including a screen loaded on an electronic device and / or content inside the screen so that it can be immediately displayed on the screen according to a scroll of a user. For example, in the display of the mobile device, the entire execution screen of the application extended in the horizontal or vertical direction and displayed according to the scrolling of the user can be included in the concept of the page, and the screen during the camera roll is also included in the concept of the page .

It should be noted, however, that the concept of goods in this specification is not limited to the type of goods. That is, in this specification, a product should be understood as a concept including not only a type of object but also an intangible service that can be sold.

1 is a flowchart illustrating a process of creating a shopping information database in a shopping information search server according to an embodiment of the present invention.

In step 110 of FIG. 1, the service server may collect product information.

For example, the service server includes a crawler, a parser, and an indexer to collect Web documents of an online store, and access text information such as a product image, a product name, and a price contained in the web document.

For example, a crawler may collect data from a Web site address in an online store, identify a Web site, and track links to deliver data related to product information to the service server. At this time, the parser analyzes the web document collected during the crawling process, extracts the product information such as the product image, commodity price, and product name included in the page, and the indexer can index the corresponding position and meaning.

Meanwhile, the service server can collect and index the product information from the website of the online store, but may also receive the product information from the server of the online store that is registered in advance.

In step 120, the service server may process the collected goods images. This is for filtering a product based on the similarity of images without depending on text information such as a product name. For this purpose, the service server extracts the feature region of the product image, calculates the feature descriptor, and indexes the feature information of the images for efficiency of search.

More specifically, the service server can perform an Interest Point Detection of characteristic regions of product images. The feature region refers to a main region for extracting a descriptor, that is, a feature descriptor, of an image for judging whether or not there is a similarity among images.

According to the embodiment of the present invention, such a feature region is a region including a corner, a corner or the like out of the contour including the image, a blob distinguished from the surrounding region, a region which is unchanged or crooked according to the deformation of the image, It can be a pole with dark or bright features.

Furthermore, the service server can extract the descriptor from the feature region (Descriptor Extraction). The feature descriptor is a vector value representing the characteristics of the image.

According to an embodiment of the present invention, the feature descriptor may be calculated using the position of the feature region for the image, or the brightness, color, sharpness, gradient, scale, or pattern information of the feature region. For example, the feature descriptor may be calculated by converting the brightness value, the brightness value, or the distribution value of the feature region into a vector.

According to an embodiment of the present invention, a feature descriptor for an image may include a global descriptor, a frequency descriptor, or a binary descriptor as well as a local descriptor based on the feature region, . ≪ / RTI >

More specifically, the feature descriptor may include a global descriptor (e.g., a global descriptor) for converting the brightness, color, sharpness, gradient, scale, pattern information, etc. of each of the regions divided by the entire image or an arbitrary reference, Global descriptor).

For example, the feature descriptor may include a frequency descriptor for converting the number of times the specific descriptors classified into the image are included in the image, the number of inclusions of the global feature such as the previously defined color table, etc. into a vector value, A binary descriptor may be included which extracts, in units of bits, whether or not the size of each element constituting the descriptor or the descriptor is larger or smaller than a specific value, and transforming it into an integer type.

Further, the service server may process the product image by applying the technique of machine learning, cluster the products into at least one product group, and extract the category for each product group.

For example, the service server can cluster products based on the similarity of the product images. (Step 130)

According to a representative embodiment of the present invention, the service server extracts groups of goods together with images of the same and / or similar images from each image, and stores the centroids information of feature information of the processed image, Value. ≪ / RTI > Furthermore, the service server can determine whether the corresponding product is included in the specific product group using the feature information of the processed image.

More specifically, the service server may calculate at least one piece of feature information of a product image and compute whether or not an arbitrary commodity is included in the corresponding commodity group by comparing with centroids information of the clustered feature information.

Further, the service server may create a new group including the product if the product is not included in any group.

According to a further embodiment of the present invention, the service server may design product groups into a hierarchical structure formed by a plurality of layers. Furthermore, a weight is assigned to the feature information of the product image according to the request of the layer, and the probability that the product is included in the specific product group using the feature information of the processed image can be calculated.

At this time, the service server may interpret the image belonging to the product groups and label the meaning of the product group. For example, the first group includes a woman's skirt, a skirt, a product of the first group is a woman's skirt, a skirt, a plaid, a group 1-1-1 is a woman's skirt, a skirt, a plaid, Product category can be calculated stochastically.

When such a machine learning technique is used, the service server may collect feedback on the search results and update the distance and weight values of the layers using the feedback information.

Meanwhile, according to another embodiment of the present invention, the service server clusters products into at least one group and stores product information (e.g., image, product name, category, price, option, manufacturer, , Capacity, number, etc.), it is possible to design a learning model capable of learning and classifying the difference between product groups and / or the similarity between products constituting a product group.

More specifically, the service server determines, based on the name of the product or the similarity of the product image, the constituent articles in each group of the initially grouped products, using the learning model, to determine the constituent articles that are not included in the group and filter them have.

Thereafter, the service server may update the parameter or the weight of the learning model by using feedback on the classification of the product groups and / or the suitability of the product configuration included in the product group.

Conventional price comparison sites clusters products based on product names. In particular, a method of sorting products with a high threshold value in the initial clustering step and then filtering the products through separate inspection is applied.

However, this method is very inefficient because human judgment is essential. Therefore, according to the present invention, initial clustering is performed using image feature information, appropriate classification of product group is performed using group feature information extracted separately for each product group, and a model for determining suitability of a component product included in a product group is designed We propose a method to filter the goods without human judgment.

On the other hand, in step 140, the service server may process the product name text on the product group having the highest similarity value.

This is for solving the problem that the product name is assigned differently for each shopping mall for the same product, and the keyword search is not easy. For this, according to the embodiment of the present invention, under the assumption that the product group having the highest image similarity value is the same product, the service server processes each product name belonging to the product group by natural language processing to generate a reference product name of the product, The product name of the products belonging to the group can be replaced or indexed by the generated reference product name.

More specifically, in step 150, the service server extracts product names of products belonging to the group, and divides the product name text information into data of the minimum semantic unit for morphological analysis. After that, the service server classifies the data rearranged regardless of the characteristics of the corresponding product as noise and filters the data.

In step 160, the service server can sort the data classified into the same meaning among the filtered data into one value. For example, if there is a slight difference in the Hangul notation of the foreign language, the data that can be classified into the same meaning, such as when the same alphabet is written in upper case and lower case, can be grouped into one value.

Thereafter, the service server can generate the reference product name by calculating the frequency with which the processed text data is used as the product name of the product group. At this time, the service server can designate or generate the representative image of the product group to configure the product database.

A more specific example of generating a reference product name is described below with reference to FIG. 2 attached hereto.

2 is a diagram for explaining an example of generating a reference product name and a representative product image according to an embodiment of the present invention.

Reference numeral 210 in FIG. 2 denotes a group of product images having a high similarity value. For example, even if different images are assigned to each shopping mall 210 with a model name of AK240, according to the embodiment of the present invention, the products can be clustered into one group based on image similarity.

In FIG. 2, reference numeral 220 is a list of products belonging to the same group as 210, that is, product names given from shopping malls of AK240 products. As shown at 220, since the shopping malls assign product names to each product individually, the search quality may be deteriorated by keyword search. In order to solve such a problem, according to an embodiment of the present invention, a plurality of product names for one product can be processed in a natural language, and a reference product name most suitable for describing a product can be generated.

In FIG. 2, reference numeral 230 denotes an example in which text such as 220 is divided into data of a minimum semantic unit, and data irrelevant to the characteristic of the product of the divided data is filtered by noise. As shown in 230, texts such as overseas purchase, overseas, output, mounting, high, sound source, correspondence, fluency, etc. will be filtered regardless of the characteristics of the goods.

In FIG. 2, 240 illustrates a sorting process of converting data classified into substantially the same meaning among the filtered data into one value. Since the iriver and the iriver in FIG. 2 are substantially the same data, the iriver can be replaced with the iriver. Because stainless steel, stainless steel and stainless steel are data of the same meaning, stainless steel can replace stainless steel.

In FIG. 2, the frequency generated from the online mall as a product name is generated as a histogram for the brand name data processed through filtering and word substitution. In the example of FIG. 2, the items were frequently used in the product names AK240, 256GB, Astell & Kern, and iriver in order. If you use it, you can confirm that the most suitable product name for the item is "AK240 256GB Astell & Kern iriver", and you can set it as a standard product name. (260 in Fig. 2)

3 is a flowchart illustrating a process of providing shopping information for a query in a shopping information search server according to an embodiment of the present invention.

In step 310 of FIG. 3, the service server may receive a query for shopping information from the user device.

For example, when a search request is input while a web browser is running, the user device transmits URL information of the displayed web page to a search request query or extracts text, image, and moving image data from a URL, It can be sent as a query. At this time, if the URL information can not be acquired, or if the text, image, or moving image data can not be extracted from the URL, a screen shot of the web page may be acquired and transmitted to the service server as a query. A method of generating a query when the URL can not be accessed will be described later.

According to another embodiment of the present invention, when a search request is received while an application other than a web browser is being executed, the user device acquires a screen shot and / or data that can be uploaded from the page, Lt; / RTI > For example, if a photo viewer application is running, the user device can send the photo file as a query. As another example, if a video playback application is running, the electronic device may send a screen shot at a particular point in time with the query.

In step 320, the service server extracts the product image from the query, compares the extracted query image with the product database image, and performs the primary search based on the image similarity.

For example, if the query contains the URL address of a particular online mall's product details page, the service server checks the index value for the query image contained in that URL, and if the indexed image You can see which product groups you belong to.

If the image is not an indexed image, the query image may be processed to calculate a degree of similarity with a product in the product database, and a group of products having an image similarity degree in a predetermined range or more may be selected.

Processing the query image is substantially the same as creating the product database. For example, the service server extracts feature information from a query image and reflects the weights requested by product groups distributed in a plurality of layers, processes the query image, and calculates at least one product group having a distance value within a certain range .

In step 320, the service server extracts text information such as a product name from the query, processes the text information in a natural language, and performs a secondary search by calculating a probability that matches the reference product name of the product group.

Processing the trade name data is substantially the same as generating the reference trade name. For example, the service server divides the product name of the query into data of the minimum semantic unit, rearranges the data, filters data irrelevant to the characteristics of the product among the rearranged data, The data classified in the same meaning can be processed to be converted according to the reference product name.

The service server can then identify the matching product by calculating the probability that the product name processing data of the query matches the reference product name of the database.

The service server may then provide shopping information for the matching goods at 340. The shopping information according to the embodiment of the present invention may include price information of the matching goods / services, the lowest price information, the lowest price online retailer, similar goods, coupons, and / or promotion information.

Further, information on other goods / services that can be recommended using the search history log of the user may also be included.

For example, if the user is browsing a web page containing a flower image, a flower delivery advertisement list, flower flower lesson promotion information, and / or rose festival discount coupon information may be displayed as shopping information according to the embodiment of the present invention .

As another example, if the user is browsing the SNS including the shoe image, according to the embodiment of the present invention, the shopping information may be written in different shoes of the same brand, same or similar color, shape and / Suggested shopping information for shoes can be displayed.

As another example, according to the embodiment of the present invention, when a user browses a picture including a toy image, the shopping information includes a toy name, a manufacturer, a manufacturing year, a lowest price, price fluctuation information, And / or how to use it.

Furthermore, the information may be displayed in text, image and / or action script form, each of which may be associated with an online merchant. Further, the information may be displayed as a separate page, or may be displayed as an icon including a shopping information link near the item on the page that the user is viewing.

Meanwhile, according to the preferred embodiment of the present invention, when a user browses a specific product at a specific online shopping mall, the lowest price of the corresponding product may be provided as shopping information.

A more specific example of providing shopping information will be described later with reference to FIG. 6 attached hereto.

6 is an illustration of a user interface for displaying shopping information according to an embodiment of the present invention.

For example, when a user browses a model name S500ACF chair in a specific shopping mall, the user device can inquire shopping information about the item while transmitting the URL of the web page to the service server by query. To this end, a shopping information providing program according to an embodiment of the present invention may be executed in the user device.

The program may be in the form of an application or plug-in installed in the mobile device, an ActiveX, an advertising SDK (module, API), or the like, installed in the desktop, or an extension program installed in the web browsing tool.

The program may be downloaded and installed by a user device through a service server and / or application open market according to an embodiment of the present invention. Further, the program may be executed in the form of an advertisement SDK (API, module) loaded in another application.

In the example of FIG. 6, the service server receiving the query from the user device extracts the product image 610 from the URL of the query, performs the primary search based on the product image, extracts the product name 620 from the URL of the query, The matching product can be specified by performing the difference search.

The service server can then search the database for the lowest price of the matching goods based on the price information 630 of the query. At this time, if the price offered by the user is determined as the lowest price on the internet, the service server notifies the device of the fact, and the user device can display a notification such as 640 in FIG.

On the other hand, the user device can display the recommended shopping information according to the setting.

For example, shopping information on learning lighting may be provided on a page that searches for a chair (650 in FIG. 6), or shopping information may be provided for another model that is similar to but low in price. (660 of FIG. 6). At this time, the service server can provide the recommended product information in the form of a reference product name and a representative product image for the corresponding product calculated in advance.

Meanwhile, in the example shown in FIG. 6, it can be considered that the application according to the embodiment of the present invention can not access the web browser. In this case, you will not be able to obtain the URL directly, and you will need to determine the need for a query with just the images displayed on the device.

The necessity of the query will exist only when the user is browsing the shopping mall, that is, when the merchandise being viewed is different from the previous one.

For this, a method of capturing a page displayed on the device at a specific cycle and transmitting the entire captured image to a service server as a query may be considered. However, this is very inefficient as the data throughput on the server becomes too high.

In order to solve such a problem, according to an embodiment of the present invention, it is determined whether a page browsed by a user is a shopping mall using a unique color value of an online shopping mall logo, and a price It is possible to judge whether or not the item that the user is viewing is updated.

If it is determined that the query is necessary, the image can be extracted from an arbitrary region that can be specified based on the price position, and the query for the image can be transmitted to the service server.

In accordance with an embodiment of the present invention, a method of generating a query with only an image is described below with reference to FIGS. 4 and 5 attached hereto.

FIG. 4 is a flowchart illustrating a process of generating a query in a client device according to an embodiment of the present invention. FIG. 5 is a diagram illustrating an example of query generation according to an embodiment of the present invention.

In step 410 of FIG. 4, the device may capture the displayed page and detect the logo of the online mall in the captured image. More specifically, the device detects a color value in a predetermined logo candidate region of a displayed page, compares the color value with a color value of a logo of a specified online shopping mall to determine whether the displayed page is for online shopping, The location of the logo can be specified.

There are a lot of variants of the UX of online malls that are provided to devices. However, in general, the logo of the online mall is generally displayed on the upper left 400 X 300 area of the web page, and the color of the logos is specified as a unique value. The present invention uses this to specify the logo candidate region in advance, It is possible to determine whether the displayed page is for online shopping by extracting the color value from the area and comparing it with the color value of the actual online mall logo.

For example, 510, 530, and 540 in FIG. 5 indicate the upper left 400 X 300 region of the captured image, i.e., a logo candidate region. In the example of FIG. 5, the logo 520, 540, 560 of each shopping mall is located in the logo candidate region.

On the other hand, the logo of the first online shopping mall is (51, 150, 44), the logo of the second online shopping mall is (235, 0, 9), the logo of the third online shopping mall is 55). ≪ / RTI >

The present invention can determine whether the corresponding page is for online shopping by detecting the color value of the logo candidate region at the top of the captured image and comparing the color value with the logo color value of each shopping mall using this information, Can also be specified.

Returning to the description of FIG. 4, in step 420, the device can extract price information from a preset price candidate area based on the identified logo position.

Generally, online malls use color values unique to the online mall for price display, and it is common to arrange a plurality of price displays adjacent to each other. Using this information, the present invention can extract, as price information, a number having a color value predetermined in advance for a plurality of number targets located adjacent to the price candidate region.

For example, 570 shown in FIG. 5 indicates a logo of a specific online shopping mall. According to an embodiment of the present invention, the price candidate region may be set as 580 based on the position of the logo. When the price candidate area is confirmed, a plurality of numbers are located adjacent to each other, and 590 having color values (239, 39, 5) specified in advance for the online mall can be extracted as price information.

Returning again to the description of FIG. 4, in step 430, the device may use pricing information to identify the need for a query, and in step 440, create a query that includes a product image.

More specifically, when the price information extracted from the captured image at that point of time is changed, the device determines that the item being read is updated and can generate a query. At this time, the query can extract and generate an image of a specific region based on the price position,

For example, reference numeral 595 shown in FIG. 5 is an area that can be specified based on the price 590, and corresponds to an area for exposing a product image in the market in general. Therefore, if the image of the corresponding region is extracted, it can be transmitted to the service server by querying the product image.

The embodiments of the present invention disclosed in the present specification and drawings are merely illustrative of specific embodiments of the present invention and are not intended to limit the scope of the present invention in order to facilitate understanding of the present invention. It is to be understood by those skilled in the art that other modifications based on the technical idea of the present invention are possible in addition to the embodiments disclosed herein.

Claims (15)

A method for providing shopping information at a server,
Step A for collecting product information from an arbitrary online market,
A step B for extracting a product image from the product information, and clustering the products having the similarity degree among the product images in a predetermined range or more to at least one product group; And
Extracting text information of the products belonging to the product group, processing the text information in a natural language to generate a reference product name of products belonging to the product group, and selecting a representative image of the product group,
In the step B,
Detecting a product object area in the product image and extracting a feature information vector describing the object area; And
Comparing a center value of feature information of any image cluster generated in advance with a feature information vector of the product image and determining whether the product image is included in the image cluster,
In the step C,
Extracting product name information of products belonging to the product group;
Dividing the product name information into the minimum meaning unit data, and filtering data not related to the characteristic of the product of the divided data; And
And replacing the data classified in the substantially same meaning in the filtered data with one text and processing the same. The method according to claim 1, wherein after the replacing and processing step,
And calculating the reference product name using the frequency at which the processed data is generated as the product name information of the product group. 3. The method of claim 2,
Receiving a query inquiring shopping information;
Extracting an image of the query product from the query, and selecting at least one product group having the image similarity of the query product image and a predetermined range or more;
Extracting textual information of the query item from the query, processing the textual information in a natural language, and calculating a probability of matching the reference product name of the selected product group; And
And providing the shopping information including the reference product name and the representative image of the product group to which the query product belongs. 2. The method according to claim 1,
Comparing a center value of feature information of any image cluster generated in advance with a feature information vector of the product image and determining whether the product image is included in the image cluster,
And creating a new image cluster that includes the product image if the product image is not included in any image clusters. 4. The method of claim 3, wherein in step B,
Designing the product groups into a hierarchical structure formed by a plurality of layers;
Calculating at least one feature information of the product image;
Assigning a weight to arbitrary feature information requested by a corresponding layer of the feature information;
Calculating a probability that the commodity is included in a specific commodity group using the weighted feature information, interpreting the category of the commodities belonging to the commodity groups, and labeling the meaning of the commodity group;
Collecting feedback on the provided shopping information; And
And updating the distance and the weight value of the layers using the feedback. 6. The method of claim 5,
Collecting address information of a product sale page of the online market; And
And obtaining at least one of the product image and the text information using the address information. A method for processing shopping information in a server,
Collecting product information from an arbitrary online market;
Extracting a product image from the product information, detecting a product object area in the product image, and extracting a feature information vector describing the object area;
Comparing a center value of feature information of a predetermined image cluster with a feature vector of the product image and determining whether the product image is included in the image cluster; And
And generating a learning model for learning similarities among the products constituting the product group using the text product information of the products belonging to one image cluster. CLAIMS 1. A method of generating, in a client, a query inquiring shopping information without accessing a web browser application,
A step of detecting a color value in a preset logo candidate region of a displayed page, determining whether the displayed page is for online shopping using the color value, and detecting a logo of the online shopping;
A step B of setting a price candidate region based on a location where the logo is displayed and extracting price information from the price candidate region;
A step C of checking whether the price information has been changed at a preset cycle, preparing a query to inquire shopping information about the goods when it is determined that the goods displayed on the page are updated when the price information is changed; And
And a step D for extracting an image of a region specified in advance on the basis of the position where the price information is displayed on the page and for querying shopping information about the product included in the image, Way. delete 9. The method of claim 8,
And extracting, as price information, a number having a color value specified in advance for each of the plurality of number targets located adjacent to the price candidate region and for each of the online shopping. 11. The method of claim 10,
Detecting a color value of an online shopping website logo in a 400X300 area at the upper left of the displayed page. A server for providing shopping information,
A communication unit for receiving product information from an arbitrary online market; And
Extracting a product image from the product information, clustering products having a degree of similarity between the product images in a predetermined range or more into at least one product group, extracting text information of products belonging to the product group, And a controller for processing the natural language to generate a reference product name of the goods belonging to the goods group and selecting a representative image of the goods group,
The control unit detects a product object region in the product image, extracts a feature information vector describing the object region, extracts a center value of feature information of an arbitrary image cluster generated in advance, and a feature information vector of the product image And determines whether the product image is included in the image cluster,
Wherein the control unit extracts product name information of products belonging to the goods group, divides the product name information into the minimum meaning unit data, filters data irrelevant to the property of the product of the divided data, And replacing the data classified into the same meaning with a single text. A server for processing shopping information,
A communication unit for receiving product information from an arbitrary online market; And
Extracting a product image from the product information, detecting a product object area in the product image, extracting a feature information vector describing the object area, and calculating a center value of feature information of an arbitrary image cluster, A similarity between products constituting a product group is learned using the text product information of products belonging to one image cluster by comparing the feature information vectors of the images, determining whether the product image is included in the image cluster, And generating a learning model for the shopping information processing server. A client that generates a query for shopping information without accessing a web browser application,
The method comprising: detecting a color value in a preset logo candidate region of a displayed page, determining whether the displayed page is for online shopping using the color value, detecting a logo of the online shopping, Determining whether the price information is changed at a predetermined cycle, determining that the goods displayed on the page have been updated when the price information is changed, setting the price candidate region as the updated price information, A query for inquiring shopping information about a product, extracting an image of a predetermined region based on a position where the price information is displayed on the page, and inquiring shopping information about the product included in the image ; And
And a communication unit for transmitting the query to the service server. A computer program stored in a computer-readable recording medium for performing a function of providing shopping information in a server,
A function of collecting product information from an arbitrary online market;
A function of extracting a product image from the product information and clustering the products having the similarity degree between the product images in a predetermined range or more into at least one product group; And
Extracting text information of products belonging to the product group, processing the text information by natural language to generate a reference product name of products belonging to the product group, and selecting a representative image of the product group,
The clustering function includes:
Detecting a product object area in the product image and extracting a feature information vector describing the object area; And
And comparing the center value of feature information of any image cluster generated in advance with the feature information vector of the product image and determining whether the product image is included in the image cluster,
The function of selecting,
A function of extracting product name information of products belonging to the product group;
A function of dividing the product name information into the minimum meaning unit data and filtering data not related to the characteristic of the product of the divided data; And
And a function of replacing and processing data classified in substantially the same meaning in the filtered data into one text.

Images