JP2021179979A - Method for extracting attribute of item for shopping search - Google Patents

Abstract

To provide a method for extracting the attribute of an item for a shopping search.SOLUTION: The method for extracting the attribute of an item for a shopping search includes the steps of receiving a user query; extracting the result of an item search for the user query from an item database; calculating the relevance score of the user query and a plurality of item attributes by a statistic foundation classification model; and adjusting an item search rank according to the item search result on the basis of the calculated relevance score.SELECTED DRAWING: Figure 1

Description

The present disclosure relates to a product attribute extraction method and system for a shopping search, specifically, a product attribute is extracted from a user query or a search query input for a shopping search, and a product search order is obtained. Regarding methods and systems that can be reflected in.

Recently, various services such as shopping, games, movies and music have been provided online in the Internet and mobile environments. In particular, online shopping is used by a large number of users because there are no restrictions on the time for purchasing products and it is possible to select and purchase products without directly visiting a product store.

Due to such convenience of online shopping, as the number of users who use online shopping increases, the number and types of products sold or handled online are also increasing geometrically. As a result, there is a problem that it is difficult for the user to search for a product that matches his / her intention among various products sold online.

On the other hand, if the user connects to the online shopping system and inputs a query (query) for searching for a desired product, the product search result including the product information related to the query and the connection information regarding the supplier can be obtained. A list can be provided. In such an online shopping system, since the user query is simply compared with the product information in the product database to extract the search results, it is difficult to derive the product search results that match the user's intention included in the query.

Korean Registered Patent No. 10-1017016

The present disclosure has been made to solve the above-mentioned problems, and provides a product attribute extraction method, a computer program, and a device (system) for a shopping search.

Further, the present disclosure provides a method and an apparatus (system) capable of more accurate product attribute extraction by using a statistical basis classification model for product attribute extraction for shopping search.

The present disclosure can be embodied in a variety of ways, including methods, devices (systems) or computer programs.

According to one embodiment of the present disclosure, the product attribute extraction method for a shopping search executed by a computer system includes a step of receiving a user query, a step of extracting product search results for the user query from a product database, and a statistical platform. The classification model includes a step of calculating the related points of the user query and a plurality of product attributes, and a step of adjusting the product search order based on the calculated related points based on the calculated related points.

According to another embodiment of the present disclosure, the product attribute extraction method for a shopping search executed on the user terminal is a step of receiving a user query, a plurality of products calculated by a statistical infrastructure classification model based on the user query. It includes a step of receiving the related score of the product attribute and a step of adjusting the product search order for the user query based on the calculated related score.

According to still another embodiment of the present disclosure, a computer program for executing the above-mentioned product attribute extraction method for shopping search on a computer is provided.

According to yet another embodiment of the present disclosure, the learning method of product attribute extraction for a shopping search performed on a computer system is a plurality of first user queries and a plurality of methods associated with the plurality of first user queries. A step of generating a plurality of first training data including the first product attribute of the above, and a related score of one or more product attributes corresponding to the second user query based on the plurality of first training data. Includes steps to train a statistically based classification model to calculate.

According to still another embodiment of the present disclosure, a computer program for executing the above-mentioned learning method of product attribute extraction for shopping search on a computer is provided.

According to yet another embodiment of the present disclosure, the product attribute extraction system for a shopping search is concatenated with a communication module, a memory, and a memory that receives a user query, and a computer-readable instruction contained in the memory. Includes at least one processor configured to run. At least one processor extracts the product search result for the user query from the product database, calculates the related score of the user query and a plurality of product attributes using the statistical infrastructure classification model, and based on the calculated related score, the said. Adjust the product search ranking based on the product search results.

According to various embodiments of the present disclosure, it is possible to output a search result reflecting the product attribute intended by the user from the user query to the user terminal. Further, according to various embodiments of the present disclosure, by displaying a product having a product attribute related to a user query at the upper part of the search result, it is possible to improve the click rate and the purchase rate of the users for the product. can.

According to the various embodiments of the present disclosure, the accuracy of the relevance of product attributes to user queries is improved by using a statistically based classification model to calculate the relevance scores of multiple product attributes to user queries. Can be made to.

According to the various embodiments of the present disclosure, the user query is divided into syllables and / or morpheme units, and the score related to the product attribute is calculated by the statistical basis classification model, so that the user is not in the typo or the training data. It is possible to improve the accuracy of calculating the related points of product attributes even for queries. In addition, the matching coverage of the training data of the statistical infrastructure classification model is maximized, and the accuracy of the user query and the calculation of the related points of a plurality of product attributes can be further improved.

According to the various embodiments of the present disclosure, the training data used for training the statistical basis classification model for calculating the number of relevant points of a plurality of product attributes for a user query may be the integrated search log data, shopping or product search log data, or the like. It can be generated from various search log data such as. In addition, since the learning data can be automatically generated based on the user query entered by the user in the search log data, the search result, and the user's action in response to the search result, the learning data of the statistical infrastructure classification model can be efficiently secured. ..

According to various embodiments of the present disclosure, by displaying a product attribute with a product attribute code, the same or similar product attributes are associated with different product types or user queries, so that the result of product attribute extraction is accurate. It is possible to prevent deterioration of sex. That is, by displaying the product attributes with the product attribute code and learning the statistical basis classification model, it is possible to improve the accuracy of calculating the related points even for the ambiguous product attributes and the range type product attributes.

The embodiments of the present disclosure will be described with reference to the following accompanying drawings. Here, similar reference numbers indicate similar elements, but are not limited thereto.
It is a figure which illustrates the screen which displays the search result which reflected the product attribute extracted from the user query by the product attribute extraction method for the shopping search which concerns on one Example of this disclosure. It is a schematic diagram which shows the system which connected the plurality of user terminals and the product attribute extraction server communicatively in order to provide the product attribute extraction service for the shopping search which concerns on one Embodiment of this disclosure. It is a block diagram which shows the internal structure of the user terminal and the product attribute extraction server which concerns on one Embodiment of this disclosure. It is a figure which shows the structure which calculates the relation point of the product attribute with respect to the user query using the statistical basis classification model which concerns on one Example of this disclosure. It is a figure which shows the structure which calculates the relation | association point using the classifier which includes the pretreatment device and the statistical basis classification model which concerns on one Example of this disclosure. It is a figure which shows the example which generates the learning data of the statistical basis classification model for product attribute extraction from the various search results which concerns on one Example of this disclosure. It is a figure which illustrates the method of generating the learning data by expressing the product attribute which concerns on one Example of this disclosure by the product attribute code. It is a flowchart which shows the product attribute extraction method for the shopping search performed by the product attribute extraction server which concerns on one Embodiment of this disclosure. It is a flowchart which shows the product search order adjustment method for the shopping search which concerns on one Example of this disclosure. It is a flowchart which shows the product attribute extraction method for the shopping search performed by the user terminal which concerns on one Embodiment of this disclosure. It is a flowchart which shows the learning method of the statistical basis classification model for product attribute extraction which concerns on one Example of this disclosure.

Hereinafter, the specific contents for the implementation of the present disclosure will be described in detail based on the attached drawings. However,
In the following description, if there is a risk of unnecessarily blurring the gist of the present disclosure, specific description of known functions and configurations will be omitted.

In the accompanying drawings, the same or corresponding components are given the same reference numerals. Further, in the following description of the embodiment, duplicate description of the same or corresponding components may be omitted. However, even if the description about the component is omitted, it should not be understood that such a component is not included in the embodiment.

The advantages and features of the disclosed embodiments and the methods for achieving them will be clarified with reference to the embodiments described below based on the accompanying drawings. However, the present disclosure is not limited to the examples disclosed below, and may be embodied in various forms different from each other. However, the present embodiment is provided only to ensure that the present disclosure is complete and that the present disclosure allows ordinary technicians to accurately recognize the category of the invention.

The terms used herein will be briefly described and the embodiments of the disclosure will be specifically described. As the terms used herein, we have selected common terms that are as widely used today as possible, taking into account the functionality of this disclosure, but this is the intent or judicial precedent of engineers engaged in the relevant field. , Can change due to the emergence of new technologies. In addition, although there may be terms arbitrarily selected by the applicant in specific cases, their meanings will be described in detail in the description of the invention. Therefore, the terms used in this disclosure should be defined based on the meaning of the term and the general content of the present disclosure, rather than the simple name of the term.

In the present specification, the singular expression may include a plurality of expressions, and the plurality of expressions may include a singular expression, unless explicitly specified in the context.

When a component is referred to as "contains" a component throughout the specification, this does not exclude other components unless otherwise stated, and may further include other components. It means that you can do it.

Also, the term "module" or "department" used in the specification means software or hardware components, and the "module" or "department" performs a certain role. However, "module" or "part" is not limited to software or hardware. The "module" or "part" may be configured to be on an addressable storage medium or to regenerate one or more processors. Thus, as an example, a "module" or "part" is a component such as a software component, an object-oriented software component, a class component, a task component, as well as a process, function, attribute, processor, subroutine, etc. It can include at least one of a segment of program code, a driver, a firmware, a microcode, a circuit, data, a database, a data structure, a table, an array or a variable. A component and a "module" or "part" may be combined with a smaller number of components and "modules" or "parts" provided internally, or an additional component and a "module" or "part". Can be further separated into.

According to one embodiment of the present disclosure, a "module" or "part" may be embodied in a processor and memory. "Processor" should be broadly interpreted to include general purpose processors, central processing units (CPUs), microprocessors, digital signal processors (DSPs), controllers, microcontrols, state machines, and the like. In some environments, "processor" can also refer to application specific integrated circuits (ASICs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), and the like. A "processor" may be, for example, a combination of a DSP and a microprocessor, a combination of multiple microprocessors, a combination of one or more microprocessors coupled to a DSP core, or any other combination of such configurations. It can also refer to a combination of various processing devices. Also, "memory" should be broadly interpreted to include any electronic component capable of storing electronic information. "Memory" includes RAM (Random Access Memory), ROM (Read Only Memory), NVRAM (Non-Volatile Random Access Memory), PROM (Programmable Read-Only Memory), EPROM (Erasable Programmable Read-Only Memory), and EEPROM (Erasable Programmable Read-Only Memory). It can also refer to various types of processors-readable media such as Electrically Erasable Programmable Reda-Only Memory), flash memory, magnetic or optical data storage devices, registers, and the like. When a processor can read / read information from and record information in memory, the memory is said to be in electronic communication with the processor. The memory integrated in the processor is in electronic communication with the processor.

In the present disclosure, the "user query" can include, but is not limited to, text (eg, words, phrases, sentences, etc.) that can be entered and searched using a search engine. For example, a user query may contain text that the user enters through the input UI of a search program such as a web browser, search application, or shopping application, or text that the user can click to select from a query list provided by the search program. Can be included. In another example, the user query is not limited to text and can include text or information that can be extracted from multimedia content such as images, video, audio, voice and the like.

In the present disclosure, the "product attribute" can indicate various information about the product including the product name, product number, standard or size, form, hue, material, performance, price and the like. For example, a product attribute may be provided to a user as a result of a shopping search or a product search, along with connection information about the purchaser of the product (eg, link information about a product purchase web page).

In the present disclosure, a "statistical-based classification model" is a classifier or classifier trained to classify new unlabeled information based on its relevance to known labeled data. A classification model can be generally shown. For example, statistical infrastructure classification models include language model, SVM (support vector machine), maximum entropy model, CRF model (conditional random field model), FFNN (feed-forward neural network), and RNN. (Recurrent neural network), GRU (gated recurrent unit), LSTM (long short-term memory), transformer model, BERT model (bidirectional encoder representations from transformers model), etc. It can be one of the neural network models.

In the present disclosure, "embedding" can indicate a technique for converting a categorical or discrete variable into a continuous vector. Embedding can be used to reduce the dimension of that information and derive meaning by transforming categorical variables such as user queries into continuous vectors. Further, when the user query is expressed as a vector in the embedded vector space, the neighboring vectors and the like on the vector space can share similar or related characteristics. For example, the method of embedding a user query in a vector space can be realized by learning an artificial neural network.

FIG. 1 is a diagram illustrating a screen 100 displaying a search result 130 reflecting a product attribute extracted from a user query 110 by the product attribute extraction method for a shopping search according to an embodiment of the present disclosure. As shown in the figure, when the user inputs the user query 110 in the search window of the shopping site, the search result 130 reflecting the product attributes 134 and 138 extracted from the user query 110 may be displayed. The search result 130 includes an image indicating a product to be searched, a name 132, 136, a product attribute 134, 138, and the like. For example, when a user inputs "mini-notebook" as a search query, the product 132, 136 associated with "mini", which is an expression related to the size or standard of the "notebook" computer included in the query, is displayed. , The order of the search results may be adjusted to be displayed at the top of the search results 130. That is, in the search result 130, among various notebook (or laptop) computer products, the products 132 and 136 having screen sizes of "10 inches" and "11 inches" have priority over the upper part of the search results 130. Can be displayed in. Therefore, the user can easily and quickly confirm the information of the product attributes 134, 138 and the related products 132, 136 that can be extracted from the product search query entered by the user via the search result 130. In addition, the user can acquire more specific information about the selected product or execute online purchase by selecting or clicking any one of the products 132 and 136 as needed.

Further, on the screen 100, a user interface 120 capable of filtering or selecting product attributes may be displayed at the upper part of the search result 130. Therefore, the user selects one or more of the product attributes related to the user query 110, such as the product category, manufacturer, screen size, etc., from the user interface 120, and the product 132 displayed in the search result 130. The order or rank of 136 can be further adjusted. For example, when the user selects "Company A" from the manufacturer filtering items from the user interface 120, information such as a product produced by "Company A" may be displayed at the upper part in the search result 130.

In one embodiment, when the user query 110 is input, the relevance or the relation probability of a plurality of product attributes to the user query 110 can be calculated by the product attribute extraction method. For example, the product attributes of a notebook computer can include hue, case material, CPU performance, memory capacity, screen size and resolution. As another example, the product attributes of the home TV can include hue, display type, size, resolution, availability of Internet connection function, and the like. When the user query 110 is a "mini notebook", the product attribute extraction method is not limited to a plurality of product attributes of a notebook computer, but also a plurality of products of different types such as household TV, fashion miscellaneous goods, and household health goods. Relevance can also be calculated for product attributes. When the user query 110 is a "mini notebook", the notebook computer's product attributes are most relevant to the small screen size attribute (eg, "10 inches"), whereas the notebook is calculated to be the most relevant. The relevance of different product attributes of computers and different types of products to product attributes can be calculated low.

On the other hand, the product search result 130 for the user query 110 can be obtained from the product database. The product search by the user query 110 can be executed by extracting the product information highly related to the user query 110 from the product information stored in the product database. In the product search result 130 extracted in this way, the product information having a high relevance to the user query 110 may be displayed higher than the product information having a low relevance. Further, the product search order based on the product search result 130 can be adjusted based on the relevance (or related points) of the product attributes to the user query 110 calculated as described above. According to this, when the user query 110 is a "mini notebook", the product search result 130 may display the product 132, 136 information related to the notebook computer having a small screen size at the higher level.

Further, according to the product attribute extraction method for the user query 110, the user query 110 is divided into syllable and / or morpheme units and analyzed, so that the relation of a plurality of product attributes to the user query 110 is related. The score can be calculated. In one embodiment, in the product attribute extraction method, the user query 110 is divided into syllable units and analyzed, so that the number of points related to the plurality of product attributes with respect to the user query 110 can be calculated. Further, in the product attribute extraction method, the user query 110 is divided into morpheme units and analyzed, so that the points related to the plurality of product attributes with respect to the user query 110 can be calculated. The two related points extracted as described above are used as the final related points by summing or combining.

In the product attribute extraction method for the user query 110, the related points of the product attributes can be calculated by the statistical base classification model. The statistically based classification model can be learned to calculate the probabilities that the user query 110 is associated with a plurality of product attributes. In one embodiment, the statistically based classification model is trained to take the user query 110 divided into syllable and / or morpheme units as input and calculate the probabilities (or related points) associated with all possible product attributes. can. For example, when "mini notebook" is input as the user query 110, the probability value corresponding to the product attributes 134 and 138 of "10 inches" or "11 inches" with a small screen size is "17 inches" with a large screen size. It can be calculated larger than the probability value corresponding to the product attribute of.

By the method described above, the search order of the products 132 and 136 included in the search result 130 can be adjusted based on the related points of the product attributes calculated by the statistical basis classification model. The product search order (or product search order score) can be determined based on the vocabulary similarity of the product name or product attribute to the user query 110, the user popularity of the product, and the like. At this time, the related points of the product attributes are applied as a sum or a weighted value to the search ranking points of each product 132 and 136, and the search ranking points of the final product 132 and 136 can be determined. In one embodiment, the search order of the products 132 and 136 can be adjusted based on the highest related score among the related scores of the product attributes calculated for the user query 110. As an alternative, the search order of the products 132 and 136 can be adjusted based on the high-order n (n is an integer larger than 1) related points among the related points of the product attributes calculated for the user query 110.

In another embodiment, the product attribute extraction method for the user query 110 is used to generate or update the product attributes of the product information from the product database. For example, in the product attribute extraction method, the related points of all possible product attributes are calculated based on the product name of the product information stored in the product database, and the product attribute related to the highest related score is the product of the product information. Can be saved or updated as an attribute.

In the example shown in FIG. 1, the product search result 130 or the product search order determined by the product attribute extraction method for the shopping search of the present disclosure is provided via a website or application for product search or shopping. However, it is not limited to this. The method according to the various embodiments of the present disclosure can also be provided via a search service of an integrated search site or a search application executed by a user terminal.

FIG. 2 shows a system 200 in which a plurality of user terminals 210_1, 210_2, 210_3 and a product attribute extraction server 230 are communicably linked in order to provide a product attribute extraction service for a shopping search according to an embodiment of the present disclosure. It is a schematic diagram which shows. As shown in the figure, the system 200 includes a product attribute extraction server 230 that provides a product attribute extraction service for shopping search, and a plurality of user terminals 210_1 and 210_2 that are connected to the product attribute extraction server 230 via a network 220. , 210_3 and the search log database 240. According to one embodiment, the product attribute extraction server 230 may be a computer-executable program (eg, a downloadable application) for providing a product attribute extraction service, or one or more capable of storing, providing, and executing data. It can include a server device and / or a database, and one or more distributed computer devices and / or a distributed database of a cloud computing service infrastructure. The product attribute extraction service provided by the product attribute extraction server 230 may be provided to the user via a search application or a web browser installed in each of the plurality of user terminals 210_1, 210_2, 210_3.

The plurality of user terminals 210_1, 210_2, 210_3 can communicate with the product attribute extraction server 230 via the network 220. The network 220 can be configured to enable communication between a plurality of user terminals 210 and the product attribute extraction server 230. The network 220 includes a wired network 220 such as Ethernet, PLC (Power Line Communication), a telephone line communication device, and RS-serial communication, a mobile communication network, and a WLAN (Wireless LAN), depending on the installation environment. , Wi-Fi, Bluetooth, ZigBee and the like, can consist of a wireless network 220 or a combination thereof. The communication method is not limited, and not only the communication method utilizing the communication network (for example, mobile communication network, wired Internet, wireless Internet, broadcasting network, satellite network, etc.) that can include the network 220, but also the user terminals 210_1 and 210_2. , 210_3 short-range wireless communication is also included. For example, the network 220 includes a PAN (personal area network), a LAN (local area network), a CAN (campus area network), a MAN (metropolitan area network), a WAN (wide area network), a BBN (broadband network), the Internet, and the like. It can include any one or more of the networks. Also, the network 220 can include any one or more of network topologies including bus networks, star networks, ring networks, mesh networks, star-bus networks, tree or hierarchical networks, etc. It is not limited to this.

In FIG. 2, a mobile phone, a smartphone 210_1, a tablet computer 210_2, and a laptop or desktop computer 210_3 are shown as examples of user terminals, but the user terminals 210_1, 210_2, 210_3 are wired and / or wireless. It can be any computer device capable of communication and capable of installing and running a search application, mobile browser application or web browser. For example, the user terminal 210 includes a smartphone, a mobile phone, a navigation terminal, a desktop computer, a laptop computer, a digital broadcasting terminal, a PDA (Personal Digital Assistants), a PMP (Portable Multimedia Player), a tablet computer, and a game console (game). A console), a wearable device, an IoT (internet of things) device, a VR (virtual reality) device, an AR (augmented reality) device, and the like can be included. Further, FIG. 2 shows that the three user terminals 210_1, 210_2, and 210_3 communicate with the product attribute extraction server 230 via the network 220, but the present invention is not limited to this, and different numbers of user terminals connect the network 220. It can also be configured to communicate with the product attribute extraction server 230 via.

According to one embodiment, the product attribute extraction server 230 can receive one or more user queries input from a plurality of user terminals 210_1, 210_2, 210_3. According to one embodiment, the product attribute extraction server 230 can receive a user query entered into a search engine of a search site displayed by a web browser running on a plurality of desktop or laptop computers 210_3. According to another embodiment, the product attribute extraction server 230 is a user query entered into a search engine of a search site displayed by a mobile browser application running on a plurality of mobile phones, smartphones and tablet computers 210_1 and 210_2. Can be received. According to another embodiment, the product attribute extraction server 230 can receive the user query input to the search engine of the online shopping site that sells the product.

Further, the product attribute extraction server 230 can calculate the related points of a plurality of product attributes for the received user query. In one embodiment, the product attribute extraction server 230 can include a statistically based classification model trained to calculate the relevant score of the product attribute. The product attribute extraction server 230 divides the received user query into units of syllables and / or morphemes, and uses the divided user queries in units of syllables and / or morphemes as input of a statistical basis classification model for calculating related points. can. Further, the product attribute extraction server 230 can receive learning data for learning the statistical infrastructure classification model from the search log database 240 linked via the network 220. For example, the learning data received from the search log database 240 includes a pair of product information (for example, attribute information of a specific product) selected by the user among the user query and the product search result for the user query. Can be done.

The product attribute extraction server 230 can extract product attributes related to one or more higher-level related points of a plurality of product attributes, and can search product information including the product attributes from a product database (not shown). The ranking of the product information searched in this way can be readjusted based on the related points of the product attributes. In addition, the product attribute extraction server 230 can adjust the product search order based on the product attribute weighted value that determines the importance of the product attribute related to the higher related points and the related points. Here, the product attribute weighted value can be changed by the user's response to the product search result (for example, the number of times the user has selected the higher-ranked product search result).

FIG. 3 is a block diagram showing an internal configuration of a user terminal 210 and a product attribute extraction server 230 according to an embodiment of the present disclosure. The user terminal 210 can indicate any computer device capable of executing a web browser or an application and capable of having / wireless communication. For example, the user terminal 210 is a mobile phone, a smartphone 210_1, or a tablet shown in FIG. It can include a computer 210_2, a laptop or a desktop computer 210_3, and the like. As shown in FIG. 3, the user terminal 210 can include a memory 312, a processor 314, a communication module 316, and an input / output interface 318. Similarly, the product attribute extraction server 230 can include a memory 332, a processor 334, a communication module 336, and an input / output interface 338. As shown in FIG. 3, the user terminal 210 and the product attribute extraction server 230 can be configured so that information and / or data can be communicated via the network 220 by using the respective communication modules 316 and 336. Further, the input / output device 320 is configured to input information and / or data to the user terminal 210 and output information and / or data generated from the user terminal 210 via the input / output interface 318. can.

The memory 312, 332 can include any non-temporary computer-readable recording medium. According to one embodiment, the memory 312, 332 is permanent such as RAM (random access memory), ROM (read only memory), disk drive, SSD (solid state drive) and flash memory (flash memory). It can include a permanent mass storage device. As another example, a permanent large capacity storage device such as a ROM, SSD, flash memory, disk drive, etc. is a user terminal 210 or a product attribute extraction server as a separate permanent storage device separated from the memory 312, 332. Can be included in 230. Further, the memory 312, 332 stores an operating system and at least one program code (for example, a code for executing a product attribute extraction method installed and driven in the user terminal 210 or the product attribute extraction server 230). obtain.

Such software components can be loaded from a computer-readable recording medium separate from the memory 312, 332. Such a recording medium that can be read by another computer can include a recording medium that can be directly connected to such a user terminal 210 and a product attribute extraction server 230, and may include, for example, a floppy drive, a disk, a tape, or a DVD. / Can include computer-readable recording media such as CD-ROM drives and memory cards. As another example, software components and the like can be loaded into memory 312, 332 via communication modules 316, 336, which are not computer readable recording media. For example, at least one program is a computer program installed by a file distribution system that distributes an installation file of a developer or an application by a file provided via a network 220 (for example, a product attribute extraction service for shopping search). It can be loaded into memory 312, 332 based on the provided application).

Processors 314 and 334 can be configured to process instructions in a computer program by performing basic arithmetic, logic and input / output operations. Instructions may be provided to processors 314, 334 by memory 312, 332 or communication modules 316, 336. For example, the processor 314, 334 can be configured to execute an instruction received by a program code stored in a recording device such as memory 312, 332.

The communication modules 316 and 336 can provide a configuration and a function for the user terminal 210 and the product attribute extraction server 230 to communicate with each other via the network 220, and the user terminal 210 and / or the product attribute extraction server 230 can be used by others. It is possible to provide a configuration or a function for communicating with a user terminal of the user or another system (for example, a separate search engine system or the like). As an example, a request or data (for example, a search result output request considering product attributes for a user query) generated by a program code stored in a recording device such as a memory 312 by the processor 314 of the user terminal 210 is a communication module. By the control of 316, it can be transmitted to the product attribute extraction server 230 via the network 220. On the contrary, the control signal or instruction provided by the control of the processor 334 of the product attribute extraction server 230 may be received by the user terminal 210 via the communication module 336 and the network 220 and via the communication module 316 of the user terminal 210. For example, the user terminal 210 can receive a product search result for a user query, a related score of a plurality of product attributes, a product search result reflecting the calculated related score, and the like from the product attribute extraction server 230 via the communication module 316. ..

The input / output interface 318 may be a means for an interface with the input / output device 320. As an example, input devices may include devices such as cameras, keyboards, microphones, mice, etc., including image sensors, and output devices may include devices such as displays, speakers, haptic feedback devices, etc. .. As another example, the input / output interface 318 may be a means for an interface with a device such as a touch screen, which has a single integrated configuration and function for performing inputs and outputs. For example, when the processor 314 of the user terminal 210 processes the instruction of the computer program loaded in the memory 312, it is configured by using the information and / or the data provided by the product attribute extraction server 230 and the other user terminal 210. The service screen and the product search result can be displayed on the display via the input / output interface 318. In FIG. 3, it is shown that the input / output device 320 is not included in the user terminal 210, but the present invention is not limited to this, and the input / output device 320 can be integrally configured with the user terminal 210. Further, the input / output interface 338 of the product attribute extraction server 230 is connected to the product attribute extraction server 230, or is connected to a device (not shown) for input or output that can be included in the product attribute extraction server 230. It can be a means for an interface. In FIG. 3, the input / output interfaces 318 and 338 are shown as elements separately configured from the processors 314 and 334, but the input / output interfaces 318 and 338 are configured to be included in the processors 314 and 334. You can also do it.

The user terminal 210 and the product attribute extraction server 230 can include more components than those shown in FIG. However, it is not necessary to articulate most of the prior art components. According to one embodiment, the user terminal 210 can be embodied to include at least a part of the above-mentioned input / output device 320. Further, the user terminal 210 can further include other components such as a transceiver, a GPS (Global Positioning system) module, a camera, various sensors, a database, and the like. For example, when the user terminal 210 is a smartphone, it can include components generally possessed by the smartphone, for example, an accelerometer, a gyro sensor, a camera module, various physical buttons, a button using a touch panel, an input / output port, and the like. Various components such as a vibrator for vibration can be further embodied in the user terminal 210.

According to one embodiment, the processor 314 of the user terminal 210 can be configured to operate a search application, a mobile browser application or a web browser capable of connecting to a product search or shopping site. At this time, the program code associated with the application or the web browser can be loaded into the memory 312 of the user terminal 210.

The processor 314 of the user terminal 210 receives the information and / or data provided from the input / output device 320 via the input / output interface 318, and the information and / or data from the product attribute extraction server 230 via the communication module 316. Can be received, and the received information and / or data can be processed and stored in the memory 312. Further, such information and / or data can be provided to the product attribute extraction server 230 via the communication module 316.

According to one embodiment, the processor 314 can be input by an input / output device 320 such as a touch screen or keyboard connected to the input / output interface 318, and can receive selected texts, images, actions, and the like. The received text, image and / or operation can be stored in the memory 312 and provided to the product attribute extraction server 230 via the communication module 316 and the network 220. For example, the processor 314 can receive one or more user queries and / or product attributes corresponding to the user queries by an input / output device such as a touch screen or a keyboard, and the received user queries and / or product attributes. The input to the product attribute extraction server 230 can be provided to the product attribute extraction server 230 via the communication module 316 and the network 220.

According to one embodiment, the processor 314 can provide the user query received by the input device or the search result request for the user query to the product attribute extraction server 230 via the network 220 and the communication module 316. The product attribute extraction server 230 that has received the user query can calculate the related points of a plurality of product attributes from the user query. In addition, the product attribute extraction server 230 can provide information on the product search result or the product search order that reflects the calculated related points of the product attribute.

According to another embodiment, the processor 314 can provide the user query received by the input device or the search result request for the user query to the product attribute extraction server 230 via the network 220 and the communication module 316. Further, the processor 314 can receive the product search result by the user query from the product attribute extraction server 230 or the product database via the network 220 and the communication module 316. The processor 334 can calculate the related points of a plurality of product attributes for the user query in response to the user query or the search result request for the user query, and apply the calculated related points to the product search result to adjust the product search order. ..

According to another embodiment, the product attribute extraction server 230 can receive a user query and a product attribute selected by the user from the user terminal 210 via the network 220 and the communication module 316. As an alternative, the product attribute extraction server 230 can receive a user query and a data pair of product attributes selected by the user from the search log database (240 in FIG. 2) via the network 220. The product attribute extraction server 230 generates training data from the received user query and product attribute data pair, and uses the generated training data to calculate the related points of the user query and product attribute, which is a statistical infrastructure classification model. Can be learned.

FIG. 4 is a diagram showing a configuration for calculating the number of related points 434, 444, and 454 of the product attribute to the user query 410 using the statistical basis classification model 420 according to the embodiment of the present disclosure. The statistical basis classification model 420 can be generated, stored or executed by the product attribute extraction server 230 or the user terminal 210 shown in FIGS. 2 and 3. As shown in FIG. 4, the statistical infrastructure classification model 420 can output the related points 434, 444, and 454 for a plurality of product attributes 430, 440, and 450 by inputting the user query 410. Here, the plurality of product attributes 430, 440, and 450 may be represented by the corresponding plurality of product attribute codes 432, 442, and 452. At this time, the product attribute codes 432, 442, and 452 can be defined in order to correspond to the ambiguous product attribute or the range type product attribute. For example, among the product attributes, "10 inches" (430) indicating the screen size of the notebook and "10 inches" indicating the screen size of the home TV can indicate different product attributes from each other. Can be represented by different product codes. Further, the related points 434, 444, and 454 may be probability values in which the user query 410 is associated with each of the plurality of product attributes 430, 440, and 450. For example, when the user query 410 is a "mini notebook", the probability value associated with the product attribute of "10 inches" (430) having a small screen size is "0.7", and the screen size is intermediate ". The probability value associated with the product attribute of "12 inches" (440) is "0.17", and the probability value associated with the product attribute of "15 inches" (450) with a large screen size is "0.01". obtain.

The related points 434, 444, and 454 for each product attribute 430, 440, and 450 can be calculated by the statistical basis classification model 420. The statistical basis classification model 420 can be learned to calculate the probability values associated with a plurality of product attributes from the input user query 410. In this case, the product attribute extraction server or the user terminal can receive learning data for learning the statistical infrastructure classification model 420 from the search log database (for example, 240 in FIG. 2). For example, the learning data received from the search log database can include a pair of product information (for example, attribute information of a specific product) selected by the user among the user query 410 and the product search results for the user query 410. .. Further, the statistical basis classification model 420 may be a machine learning or artificial neural network model for determining the relationship of a plurality of product attributes to the user query 410.

In FIG. 4, it is shown that one statistical basis classification model 420 is used to calculate the related points 434, 444, and 454 of the three product attributes 430, 440, and 450 for the user query 410, but the present invention is limited to this. It's not something. The statistical base classification model 420 can be configured to calculate the relevant score of less than 3 or 4 or more product attributes for the user query 410 by various embodied methods. In addition, two or more statistically based classification models can be used by the method of preprocessing or analyzing the user query 410. In one embodiment, the user query 410 is divided into a plurality of data in smaller units (eg, data in units of morphemes and / or syllables), and the divided data is input to the corresponding statistical basis classification model. Can be done. In this case, the final related points can be calculated by adding or combining the related points calculated from a plurality of statistically based classification models.

FIG. 5 is a diagram showing a configuration for calculating related points 534, 544, and 554 using a classifier 520 including a pretreatment device 510 and a statistical base classification model 522 and 524 according to an embodiment of the present disclosure. The preprocessing device 510 and the classifier 520 can be generated, stored, or executed by the product attribute extraction server 230 or the user terminal 210 shown in FIGS. 2 and 3. As shown in FIG. 5, the product attribute extraction server or the user terminal is generated from the preprocessing device 510 and the preprocessing device 510 that analyze the user query 410 and generate the input data of the statistical infrastructure classification model 522 and 524. It can include a classifier 520 that calculates related points for a plurality of product attributes using input data.

The preprocessing device 510 can include a syllable unit dividing unit 512 that divides the user query 410 into syllable units, and a morpheme unit dividing unit 514 that divides the user query 410 into morpheme units. In one embodiment, the syllable unit division unit 512 and the morpheme unit division unit 514 can convert the text of the user query 410 into a syllable embedding vector and a morpheme embedding vector, respectively. Each of the syllable unit division unit 512 and the morpheme unit division unit 514 can be embodied by an artificial neural network trained to embed the user query 410 in a continuous vector space.

Further, the classifier 520 has a first statistical base classification model 522 that inputs a user query 410 divided into syllable units and a second statistical base that inputs a user query 410 divided into morpheme units. The classification model 524 and the connection unit 526 for calculating the final relation score 534, 544, 554 by adding or combining the relation points calculated from the respective statistical basis classification models 522 and 524 can be included.

In one embodiment, the product attribute extraction server or the user terminal inputs the first statistical infrastructure classification model 522 having the user query 410 divided into syllable units as input, and the user query 410 divided into morpheme units. The related points 534, 544, and 554 can be calculated by using the second statistical base classification model 524.

The statistically based classification model 522 and 524 included in the classifier 520 receive user queries 410 divided into syllable and morpheme units as inputs, and have related points 534, 544, and 554 for a plurality of product attributes 530, 540, and 550. Can be calculated. Each of the statistical infrastructure classification models 522 and 524 uses the learning data extracted from the search log database to calculate the related points 534, 544, and 554 of a plurality of product attributes 530, 540, and 550 for the user query 410. You can learn. The learning data extracted from the search log database can include a pair of product information (for example, attribute information of a specific product) selected by the user among the user query and the product search result for the user query.

The statistical base classification model 522, 524 does not use the texts (for example, sentences, phrases, words, etc.) constituting the user query 410 as they are, but divides them into units of syllables or morphemes and uses the user query 410. It can also be used when typographical errors are included in the data or when the user query 410 is not included in the training data and is unfamiliar. For example, if the user query 410 is a "mini-notebook", the "mini-notebook" is divided into syllable units, and "mi", "ni", "no", "to", and "book" are classified into statistical bases. If the related score calculated as the input of the model 522 and the user query 410 include typographical errors as in the "mini-notebook", "mi", "ni", "na", "to" divided into syllable units. , "Book" is the same as or similar to the related score calculated by inputting the statistical basis classification model 522. As another example, when the user query 410 is a "mini-notebook", "mini" and "notebook" obtained by dividing the "mini-notebook" into morpheme units are calculated as inputs of the statistical base classification model 524. When the related points and the order of the morphemes are changed like "Notebook mini" in the user query 410, "Notebook" and "Mini" divided into morpheme units are calculated as the input of the statistical base classification model 522. It is the same as or similar to the related points.

The related points calculated by the statistical basis classification model 522 and 524 can be added up or combined by the joining unit 526. As a result, it is possible to calculate the final related points 534, 544, and 554 with higher accuracy than when one statistically based classification model is used. With the above configuration, the matching coverage of the learning data used by the statistical infrastructure classification model 522 and 524 can be maximized, and the accuracy of the user query and the calculation of the related points of a plurality of product attributes can be improved.

FIG. 6 is a diagram showing an example of generating learning data of a statistical basis classification model for extracting product attributes from various search results according to an embodiment of the present disclosure. The training data used for training the statistically based classification model can be extracted from the search log database 650. The search log database 650 may include, for example, a user query (for example, a sentence, a phrase, a word, etc. containing the product name to be searched) entered by the user while using the shopping search service, or a shopping search result (for example,) by the user query. , Search results including product names and product attributes), and user responses to shopping search results (for example, user clickability or number of times for specific product names and product attributes) can be saved. The training data of the statistical basis classification model can include a user query extracted from the search log database 650 and a pair of product attributes associated with the user query.

As shown in the figure, for example, when a user inputs a user query 110 in a search window of a shopping site or an integrated search site, a search result 130 related to the input user query 110 may be output. The search result screen 100 has a search window in which a user query 110 is input, a user interface 120 for filtering product attributes, and product attributes 134, 138 related to the search products 132, 136 and the search products 132, 136. The search result 130 including the above can be included.

In one embodiment, when the user who inputs the user query 110 clicks the screen size attribute 610 included in the user interface 120, the pair of training data 640 of the user query 110 and the clicked screen size attribute 610 is searched. It can be saved in the log database 650. The processor of the product attribute extraction server can extract the training data including the user query 110 and the pair of product attributes from the search log database 650 in which the user query 110 and the click log information are stored. For example, if the user enters "mini-notebook" as the user query 110 and clicks "12 inches" in the product attributes via the user interface 120 to generate search log data, then "mini-notebook-" A pair of "12 inch" training data 640 can be generated.

As another embodiment, when the user who inputs the user query 110 clicks the product 132 and 136 output as the search result 130, the user query 110 and the product attributes 134 and 138 associated with the product 132 and 136 are displayed. Paired training data 620, 630 may be stored in the search log database 650. For example, if the user inputs "mini notebook" as the user query 110 and then clicks on products 132 and 136 in the search results 130, the product attribute "10 inches" (10 inches) related to the clicked product 132 ( 134) and the paired training data 620 of the user query 110 and the product attribute "11 inch" (138) associated with the clicked product 136 and the paired training data 630 of the user query 110 may be generated. As an additional or alternative, if the user who entered the user query 110 clicks on the product attribute 134, 138 associated with the product, the paired learning data 620, 630 of the clicked product attribute 134, 138 and the user query 110 is generated. Can be done.

FIG. 7 is a diagram illustrating a method of generating learning data 710, 720, 730 by expressing the product attributes 714, 724, 734 according to the embodiment of the present disclosure with the product attribute codes 716, 726, 736. As shown in the figure, the training data used for training the statistical basis classification model includes user queries 712, 722, 732, corresponding product attributes 714, 724, 734, and each product attribute 714. , 724, 734 can be included with the product attribute codes 716, 726, 736. In one embodiment, user queries 712, 722, 732 and product attributes 714, 724, 734 place user queries 712, 722, 732 in the search window of a shopping site or integrated search site, as described with reference to FIG. When the input user clicks the product or product attribute displayed in the search result, it can be extracted from the generated search log database. Further, each of the product attributes 714, 724, and 734 can be expressed by the corresponding product attribute codes 716, 726, and 736. In this way, by expressing the product attributes included in the training data with the product attribute code, when products of different types have the same or similar product attributes (that is, in the case of ambiguous product attributes and range type product attributes). However, the statistical basis classification model can calculate the accurate related points.

For example, if a user enters a "mini notebook" (712) in the search window of a shopping site and clicks on a product attribute of "10 inches" (714), the entered "mini notebook" (712) user. The training data 710 including the query, the clicked "10 inch" (714) product attribute, and the "10 inch" (714) product attribute code "M10834726" (716) can be generated. Also, when the user inputs "mini notebook" (722) in the search window of the shopping site and clicks the product attribute of "12 inches" (724), the entered "mini notebook" (722) user. The training data 720 including the query, the clicked "12 inch" (724) product attribute, and the "12 inch" (724) product attribute code "M10706689" (726) can be generated.

In another embodiment, when a user who inputs a user query clicks two or more product attributes in the search result, training data including all the input user query and the clicked product attribute can be generated. For example, when a user enters a "mini notebook" (732) in the search window of a shopping site and clicks on the product attributes of "13 inches and 12 inches" (734), respectively, the entered "mini notebook" ( The training data 730 including the user query of 732), the clicked product attributes of "13 inches and 12 inches" (734), and the corresponding product attribute codes "M10835015 and M10706689" (736) can be generated.

FIG. 8 is a flowchart showing a product attribute extraction method 800 for a shopping search performed by the product attribute extraction server according to the embodiment of the present disclosure. In one embodiment, the product attribute extraction method 800 for shopping search can be started by receiving a user query (S810). In one embodiment, the user query can be received from the search window of the shopping site or the integrated search site displayed on the user terminal and transferred to the processor of the product attribute extraction server.

Based on the received user query, the product search result for the user query is extracted from the product database (S820). In one embodiment, the processor of the product attribute extraction server can extract the product search result for the user query from the product database. At this time, the product search result can include product search ranking information determined based on information such as a user query, a vocabulary similarity of product names stored in a product database, and user popularity of the product. For example, if the user query is "mini-notebook", among the products with a product name that includes at least a part of "mini-notebook", the product with a high purchase rate by the user or the product with a high click rate by the user. However, it can be extracted as a product search result with a high ranking.

User queries can be divided into syllable and / or morpheme units (S830). In one embodiment, the processor of the product attribute extraction server can divide the user query into units of syllables and / or morphemes and input them into the statistical basis classification model. For example, a user query can be converted into a syllable-embedded vector and / or a morpheme-embedded vector and entered into a statistically-based classification model.

In addition, the statistical base classification model can be used to calculate the number of related points of a plurality of product attributes for a user query (S840). In one embodiment, the processor of the product attribute extraction server can calculate the probability that the user query is associated with the product attribute displayed by the product attribute code by using the statistical base classification model. In this way, by displaying the product attribute by the product attribute code, the statistical basis classification model can calculate an accurate related score corresponding to the ambiguous product attribute and the range type product attribute. When the user query is divided into units of syllables or morphemes, the statistical basis classification model can calculate the probability value that the user query divided into units of syllables or morphemes is associated with a plurality of product attributes.

In another embodiment, the processor of the product attribute extraction server can divide the user query into units of syllables and morphemes, respectively. The processor uses the first statistical-based classification model to calculate the first probability that a user query divided into syllable units is associated with multiple product attributes, and uses the second statistical-based classification model to calculate the first probability. A second probability that the user query divided into morpheme units is associated with the product attribute can be calculated. The processor can then add up or combine the first and second probabilities to calculate the relevant score for the final product attribute. As mentioned above, by dividing the user query into units of syllables and / or morphemes and calculating the relevant points for the product attributes by the statistical base classification model, even for user queries that are not in the typo or training data, the product It is possible to improve the accuracy of calculating the related points of the attribute.

Based on the calculated related points, the product search order based on the product search results for the user query can be adjusted (S850). In one embodiment, the processor of the product attribute extraction server can adjust the product search order by reflecting the calculated one or more related points in the product search result. As a result, products containing product attributes with a high related score can be displayed at the top of the search results. By displaying a product having a product attribute related to a user query at the top of the search result, it is possible to improve the click rate and purchase rate of the users for the product.

FIG. 9 is a flowchart showing a product search order adjusting method 900 for a shopping search according to an embodiment of the present disclosure. In one embodiment, the product search order adjustment method 900 can be started by calculating the related points of a plurality of product attributes for the user query (S910). The method of calculating the related points of the plurality of product attributes for the user query can be performed in the same manner as the method described with reference to FIG.

Next, among the calculated related points, the first product attribute related to the first related score, which is the highest related score, can be extracted (S920). In one embodiment, the processor of the product attribute extraction server or the user terminal can extract the first product attribute related to the first related score, which is the highest related score among the calculated related scores. Further, based on the extracted first product attribute, the first product including the first product attribute can be searched from the product database (S930).

As an alternative, of the calculated association scores, the first product attribute group (ie, one or more) associated with the first association score group (ie, one or more first association scores) whose association score is equal to or higher than the predetermined threshold value. , One or more first product attributes) can be extracted. After that, a product containing at least one product attribute included in the first product attribute group can be searched from the product database.

A product attribute weighted value that determines the importance of the first product attribute can be determined (S940). The product attribute weighted value can indicate the degree of influence of the extracted product attribute on the adjustment of the search order of the related products. In one embodiment, the product attribute weighted value may be a default value for each product attribute. Alternatively, the product attribute weighted value can dynamically fluctuate depending on the user's response to the product search results. Specifically, the product attribute extraction server or the processor of the user terminal outputs a search result screen displaying the first product adjusted based on the default product attribute weighted value and the related points, and the output product search. The number of clicks of the first product by the user can be received on the result screen. If the number of clicks on the first product is greater than or equal to the default threshold, the processor can change the product attribute weighted value.

The search order of the first product can be adjusted based on the first related points and the product attribute weighted value (S950). As a result of adjusting the product search order, the first product related to the extracted first product attribute is higher in the product search result than the other products not related to the extracted first product attribute. Can be displayed in. As a result, the search result reflecting the product attribute intended by the user can be output to the user terminal based on the user query.

FIG. 10 is a flowchart showing a product attribute extraction method 1000 for a shopping search performed by a user terminal according to an embodiment of the present disclosure. In one embodiment, the product attribute extraction method 1000 for the shopping search performed by the user terminal can be started by the processor of the user terminal receiving the user query (S1010). The user query can be received from the search window of the shopping site or the integrated search site displayed on the user terminal.

The processor of the user terminal can receive the related points of a plurality of product attributes calculated by the statistical basis classification model based on the user query (S1020). At this time, the statistical basis classification model can be learned in advance so as to calculate the related points of a plurality of product attributes for the user query. In addition, in order to calculate the relevant score with high accuracy, the statistical basis classification model receives a user query divided into units of syllables and / or morphemes and calculates the relevant score of the product attribute. It can include a base classification model.

The processor of the user terminal can adjust the product search order for the user query based on the calculated related points (S1030). Specifically, the processor can extract the first product attribute associated with the first related score, which is the highest related score among the calculated related scores. After that, the processor receives the first product including the first product attribute searched from the product database, and can adjust the product search order of the first product searched by the first related points. At this time, the search order of the first product can be adjusted based on the product attribute weighted value that determines the importance of the first product attribute and the first related score.

FIG. 11 is a flowchart showing a learning method 1100 of a statistical basis classification model for extracting product attributes according to an embodiment of the present disclosure. As shown in the figure, in the learning method 1100, the processor generates the first training data based on the plurality of first user queries and the plurality of first product attributes associated with the plurality of first user queries. It can be started by step (S1110). The processor can extract a plurality of first product attributes from a search log database associated with the plurality of first user queries. The search log database can include product attribute information selected or clicked by the user from the user query and the product search results extracted thereby. For example, the processor can extract product attributes associated with multiple user queries from at least one of the integrated search log database, shopping or product search log database. In addition, the processor can convert a plurality of first product attributes into a plurality of corresponding product attribute codes so that the statistical basis classification model can correspond to an ambiguous product attribute or a range type product attribute. With the above configuration, the training data is generated by the user's actions (for example, filtering of product attributes, selection or clicking of products or product attributes, etc.) with respect to the user query entered by the user and the product search results extracted by the user query. Since it is obtained from the search log database, it is possible to efficiently secure the training data of the statistical infrastructure classification model.

The processor can train the statistical basis classification model to calculate the relevant score of one or more product attributes corresponding to the second user query based on the plurality of first training data (S1120). In one embodiment, the processor can learn a statistically based classification model by dividing a plurality of first user queries into morpheme units. Specifically, the processor can learn a statistically-based classification model to calculate the relevant score of one or more product attributes corresponding to a user query based on a plurality of first user queries divided into morpheme units. .. As an addition or alternative, the processor can train a statistically based classification model by dividing a plurality of first user queries into syllable units. The processor can learn a statistically based classification model to calculate the relevant score of one or more product attributes corresponding to the user query based on the plurality of first user queries divided into syllable units.

In another embodiment, the processor can learn a statistically based classification model by dividing a plurality of first user queries into units of syllables and morphemes. Specifically, the processor can divide the plurality of first user queries into morpheme and syllable units. The processor is then trained by the first statistical infrastructure classification model to calculate the relevant score for one or more product attributes corresponding to the user query based on the plurality of first user queries divided into morpheme units. can. In addition, the processor is trained by the second statistical infrastructure classification model so as to calculate the relevant score of one or more product attributes corresponding to the user query based on the plurality of first user queries divided into syllable units. can. As described above, the processor can more accurately calculate the related points of the user query and the product attribute by using two or more statistical-based classification models by the method of partitioning the user query.

The product attribute extraction method and the product attribute extraction learning method for the shopping search described above can also be embodied as a computer-readable code on a computer-readable recording medium. The recording medium may be one that continuously stores a computer-executable program or temporarily stores it for execution or download. Further, the medium may be various recording means or storage means in the form of a single or a combination of a large number of hardware, but the medium is not limited to the medium directly connected to a certain computer system, and is distributed on the network. It can be a thing. Examples of media include magnetic media such as hard disks, floppy disks and magnetic tapes, optical recording media such as CD-ROMs and DVDs, and magnetic-optical media such as floptical disks. Medium), and includes ROM, RAM, flash memory, etc., and includes those configured to store program command words. Further, as an example of other media, a recording medium or a storage medium managed by an app store that distributes applications, a site that supplies or distributes various other software, a server, or the like can be mentioned.

The methods, actions or techniques of the present disclosure can be embodied by a variety of means. For example, such techniques can be embodied in hardware, firmware, software, or a combination thereof. A normal engineer should understand that the various exemplary logical blocks, modules, circuits and algorithm steps described in the disclosure of this application can be embodied in electronic hardware, computer software, or a combination of both. Is. To articulate such mutual substitution of hardware and software, a variety of exemplary components, blocks, modules, circuits and steps have been generally described above in terms of their functionality. Whether such a function is embodied as hardware or software depends on the design requirements added to the specific application and the entire system. The usual technician may also embody the functions described in various ways for each particular application, but such embodying shall not be construed as departing from the scope of the present disclosure.

In the realization of hardware, the processing units used to perform the technique are one or more ASICs, DSPs, digital signal processing devices (DSPDs), programmable logic devices (PLDs). , Field programmable gate arrays (FPGAs), processors, controllers, microprocessors, microprocessors, electronic devices, and other electronic units designed to perform the functions described in the present disclosure. , Computers, or combinations thereof.

Accordingly, the various exemplary logic blocks, modules and circuits described herein are general purpose processors, DSPs, ASICs, FPGAs and other programmable logic devices, discrete gates and transistor logic, discrete hardware components, or. , Can also be embodied or performed in any combination, such as those designed to perform the functions described herein. The general purpose processor can be a microprocessor, but as an alternative, the processor can be any conventional processor, controller, microprocessor, or state machine. The processor can also be embodied in a combination of computing devices, such as a DSP and microprocessor, multiple microprocessors, one or more microprocessors associated with a DSP core, or any other configuration combination. ..

In the realization of firmware and / or software, the techniques are RAM (random access memory), ROM (read-only memory), NVRAM (non-volatile random access memory), PROM (programmable read-only memory), EPROM (erasable programmable). It can be embodied by instructions stored on a computer-readable medium such as read-only memory), EPROM (electrically erasable PROM), flash memory, CD (compact disc), magnetic or optical data storage devices, and the like. Instructions can be executed by one or more processors, which can be made to perform the particular mode of function described in the present disclosure.

When embodied in software, the technique can be stored on a computer-readable medium as one or more instructions or codes, or transferred via a computer-readable medium. Computer-readable media include any medium that facilitates the transfer of computer programs from one location to another, including both computer storage media and communication media. The storage medium can be any available medium accessible by the computer. As a non-limiting example, such computer readable media may include RAM, ROM, EEPROM, CD-ROM or other optical disk storage, magnetic disk storage or other magnetic storage device, or desired program code. It can be used to transfer or store in the form of instructions or data structures and can include any other medium accessible by a computer. Also, any connection can be included as a computer readable medium.

For example, software transfers from websites, servers or other remote sources using coaxial cables, fiber optic cables, leads, digital subscriber lines (DSL), or wireless technologies such as infrared, wireless and microwave. Where so, coaxial cables, fiber optic cables, leads, digital subscriber lines, or wireless technologies such as infrared, wireless and microwave are included within the definition of medium. The discs and discs used in the present application include CDs, laser discs, optical discs, DVDs (digital versatile discs), floppy discs and Blu-ray discs, where the discs are usually magnetic. While the data is reproduced on an optical disc, the data is optically reproduced using a laser. The combination or the like must also be included within the range of a computer-readable medium or the like.

The software module can also reside in RAM memory, flash memory, ROM memory, EPROM memory, EEPROM memory, registers, hard disk, mobile disk, CD-ROM, or any known different form of storage medium. The exemplary storage medium can be coupled to the processor such that the processor reads information from the storage medium or writes information to the storage medium. Alternatively, the storage medium can be integrated into the processor. The processor and storage medium can also be present in the ASIC. The ASIC can also be present in the user terminal. Alternatively, the processor and storage medium can also exist as individual components at the user terminal.

Although the embodiments described above are described as taking advantage of the aspects of the subject currently disclosed in one or more stand-alone computer systems, the present disclosure is not limited to this, such as networks and distributed computing environments. It can be realized by any computing environment. Furthermore, the aspects of the subject matter in the present disclosure can be embodied in multiple processing chips and devices, and storage can be similarly affected across multiple devices. Such devices may also include PCs, network servers and portable devices.

Although this disclosure has been described herein by some embodiments, various modifications and modifications are made within the scope of this disclosure that are understandable to ordinary technicians in the art to which the invention of the present disclosure belongs. It is possible. It should also be understood that such modifications or alterations fall within the scope of the claims attached herein.

100 screen 110 user query 120 user interface 130 search result 132, 136 product 134, 138 product attribute 210 user terminal 220 network 230 product attribute extraction server 240 search log database 420 statistical infrastructure classification model 510 preprocessing device 520 classifier

Claims (20)

In the product attribute extraction method for shopping search executed by the computer system
Steps to receive user queries;
Step to extract product search results for the user query from the product database;
The step of calculating the related points of the user query and a plurality of product attributes by the statistical base classification model; and
A product attribute extraction method for a shopping search, which comprises a step of adjusting a product search order based on the product search result based on the calculated related points. The step of calculating the related points of the user query and the plurality of product attributes by the statistical basis classification model is
The product attribute extraction method for a shopping search according to claim 1, further comprising a step of calculating the probability that the user query is associated with the plurality of product attributes displayed by the product attribute code by the statistical basis classification model. .. The step of calculating the related points of the user query and the plurality of product attributes by the statistical basis classification model is
The step of dividing the user query into at least one of syllable and morpheme units; and
The first or second claim comprises the step of calculating the probability that the user query divided into at least one of the units of the syllable and the morpheme by the statistical basis classification model is associated with the plurality of product attributes. Product attribute extraction method for shopping search. The step of calculating the related points of the user query and the plurality of product attributes by the statistical basis classification model is
The step of dividing the user query into units of syllables and morphemes;
A step of calculating the first probability that the user query divided into the syllable units is associated with the plurality of product attributes by the first statistical basis classification model;
The step of calculating the second probability that the user query divided into the units of the morpheme is associated with the plurality of product attributes by the second statistical basis classification model;
The product attribute extraction method for a shopping search according to claim 1 or 2, which comprises a step of adding up the first probability and the second probability and calculating as a related score of the plurality of product attributes. The step of adjusting the product search order based on the product search result based on the calculated related points is
A step of extracting the first product attribute related to the first related score, which is the highest related score among the calculated related scores;
A step of searching for a first product containing the extracted first product attribute from the product database; and
The product attribute extraction method for a shopping search according to any one of claims 1 to 4, which comprises a step of adjusting the product search order of the first product searched by the first related points. The step of adjusting the product search order based on the product search result based on the calculated related points is
A step of extracting a first product attribute group related to a first related score group whose related score is equal to or higher than a predetermined threshold value from the calculated related scores;
A step of searching for a product containing at least one product attribute included in the extracted first product attribute group from the product database; and.
The product attribute extraction method for a shopping search according to any one of claims 1 to 4, comprising a step of adjusting the product search order of the product searched by the first related score group. The step of adjusting the product search order based on the product search result based on the calculated related points is
The shopping search according to claim 5, further comprising a step of adjusting the search order of the first product based on the product attribute weighted value for determining the importance of the first product attribute and the first related points. Product attribute extraction method for. A step of outputting a search result screen in which the first product is displayed by adjusting the search order;
The step of receiving the number of clicks of the first product on the output search result screen; and
The product attribute extraction method for a shopping search according to claim 7, further comprising a step of changing the product attribute weighted value when the number of clicks of the first product is equal to or greater than a predetermined threshold value. In the product attribute extraction method for shopping search executed on the user terminal
Steps to receive user queries;
A step of receiving the related points of a plurality of product attributes calculated by the statistical basis classification model based on the user query; and
A product attribute extraction method for a shopping search, which comprises a step of adjusting a product search order for the user query based on the calculated related points. The step of adjusting the product search order for the user query based on the calculated related points is
A step of extracting the first product attribute related to the first related score, which is the highest related score among the calculated related scores;
The step of receiving the first product containing the extracted first product attribute searched from the product database; and
The product attribute extraction method for a shopping search according to claim 9, which comprises a step of adjusting the product search order of the first product searched by the first related points. The step of adjusting the product search order for the user query based on the calculated related points is
The shopping search according to claim 10, further comprising a step of adjusting the search order of the first product based on the product attribute weighted value for determining the importance of the first product attribute and the first related points. Product attribute extraction method for. A computer program for executing a product attribute extraction method for a shopping search according to any one of claims 1 to 11 on a computer. In the learning method of product attribute extraction for shopping search performed on a computer system
A step of generating a plurality of first training data including a plurality of first user queries and a plurality of first product attributes associated with the plurality of first user queries; and.
A product for shopping search, including a step of learning a statistical infrastructure classification model to calculate the relevant score of one or more product attributes corresponding to a second user query based on the plurality of first learning data. How to learn attribute extraction. The step of generating the plurality of first training data is
13. The shopping search according to claim 13, comprising the step of extracting the plurality of first product attributes from at least one of the plurality of integrated search click information, shopping click information, or product database associated with the plurality of first user queries. How to learn product attribute extraction for. The step of generating the plurality of first training data is
The learning method of product attribute extraction for a shopping search according to claim 13 or 14, which comprises a step of converting the plurality of first product attributes into a plurality of first product attribute codes that can correspond to ambivalent product attributes. .. The step of learning the statistically based classification model is
The step of dividing the plurality of first user queries into morpheme units; and
A step of learning the statistical basis classification model to calculate the relevant score of one or more product attributes corresponding to the user query based on the plurality of first user queries divided into units of the morpheme. The learning method for extracting product attributes for a shopping search according to any one of claims 13 to 15. The step of learning the statistically based classification model is
The step of dividing the plurality of first user queries into syllable units; and
A step of learning the statistical basis classification model to calculate the relevant score of one or more product attributes corresponding to the user query based on the plurality of first user queries divided into the syllable units. The learning method for extracting product attributes for a shopping search according to any one of claims 13 to 15. The step of learning the statistically based classification model is
The step of dividing the plurality of first user queries into morpheme units;
The step of dividing the plurality of first user queries into syllable units;
A step of learning a first statistical basis classification model to calculate the relevant score of one or more product attributes corresponding to a user query based on the plurality of first user queries divided into units of the morpheme; as well as,
Based on the plurality of first user queries divided into the syllable units, a step of learning a second statistical basis classification model so as to calculate the relevant score of one or more product attributes corresponding to the user query. The learning method for extracting product attributes for a shopping search according to any one of claims 13 to 15, including. A computer program for executing a learning method of product attribute extraction for a shopping search according to any one of claims 13 to 18 on a computer. In the product attribute extraction system for shopping search
Communication module that receives user queries;
Memory; and
Containing at least one processor concatenated with said memory and configured to execute computer-readable instructions contained in said memory.
The at least one processor
The product search result for the user query is extracted from the product database, the related score of the user query and a plurality of product attributes is calculated using the statistical basis classification model, and the product search result is calculated based on the calculated related score. Product attribute extraction system for shopping search that adjusts the product search ranking by.

Images