JP7014830B2 - Methods and systems for automatically extracting foreign synonyms using a transliteration model - Google Patents

Description

The following description relates to a technique for extracting synonyms, and more particularly to a technique for extracting the original word notation of a loan word.

When a user tries to collect information using the Internet, it is common to connect to a server that provides the information and obtain the relevant information by searching.

When you try to buy a product, you can use a search engine to get information about its price, characteristics, etc., or you can go to the website that sells the product.

In order to search for a product, a product registration procedure including a process of matching a product model and a category is required. For example, Patent Document 1 (registration date: April 03, 2007) discloses a technique for automatically matching a product model and a category and registering product information.

Korean Registered Patent No. 10-0705410

For keywords extracted from product information, it is possible to build a multigraph structure that shows the relationship between keywords, and in the case of loanwords, it is possible to automatically extract the original language notation as a synonym and add it to the multigraph. Provide possible methods and systems.

Provides a method and system that can extract the original word notation of the loan word from the word that appears in the product name or product-related query using the transliteration model for the loan word and process it as a synonym. do.

A method of extracting synonyms performed by a computer system, wherein the computer system includes at least one processor configured to execute a computer-readable instruction contained in memory, wherein the method of extracting synonyms is described above. A step of extracting a keyword list related to the target keyword for the target keyword by at least one processor, and a transliteration model of the target keyword from the keyword list by the at least one processor. Provided is a synonym extraction method including a step of extracting a synonym corresponding to the original language notation.

According to one aspect, the stage of extracting the keyword list is the product name list clicked after inputting the target keyword as a search query, other query lists related to each clicked product, and tag list related to each clicked product. At least one of them may be extracted.

According to another aspect, at the stage of extracting the synonyms, the synonyms may be extracted by using a transliteration model using the smallest unit of pronunciation.

According to another aspect, the step of extracting the synonym may include the step of separating the target keyword into the smallest unit of pronunciation by using the data format of linked bi-gram (LBG). ..

According to another aspect, the step of extracting the synonym is a step of extracting a word matching the target keyword from the words extracted from the keyword list using the transliteration model as the synonym. May include.

Further, according to another aspect, the step of extracting the synonym may further include a step of substituting the matching word for the target keyword and extracting a synonym candidate.

Further, according to another aspect, the step of extracting the synonym may further include the step of extracting a word of a specific language from the keyword list.

Further, according to another aspect, the step of extracting the synonym may further include the step of separating into a single word when the target keyword consists of a compound word.

According to another aspect, at the stage of extracting the synonyms, the product name list clicked after inputting the target keyword as a search query, other query lists related to each clicked product, and tags related to each clicked product. It may further include the step of verifying the synonym using at least one of the lists.

According to another aspect, at the stage of extracting the synonyms, the product name list clicked after inputting the target keyword as a search query, other query lists related to each clicked product, and tags related to each clicked product. It may further include determining the synonym based on whether or not a word matching the target keyword appears in at least one of the lists, or how often it appears.

According to another aspect, the synonym extraction method is a step of generating a multigraph showing a relationship between keywords on a product dictionary by the at least one processor, and the synonym is obtained by the at least one processor. It may further include a step of additional registration to the multigraph.

According to another aspect, the generation stage is a stage of converting the product information matching with the corresponding product into a keyword for each product included in the product dictionary, and between the information matching with the product from the product dictionary. It may include a step of extracting the relational data of the above as the relational data between keywords.

According to still another aspect, in the stage of additional registration, the relationship between the target keyword and the synonym is added on the multigraph, or the target keyword is searched among the keywords on the multigraph. Add the relationship between keywords that appear with the synonym to at least one of the product name list you clicked after entering, the other query list associated with each clicked product, and the tag list associated with each clicked product. good.

A program for causing a computer to execute the synonym extraction method is provided.

A computer system comprising at least one processor configured to execute a computer-readable instruction contained in memory, wherein the at least one processor is a keyword list associated with the subject keyword for the subject keyword. Provided is a computer system including a synonym extraction unit that extracts synonyms corresponding to the original word notation of the target keyword from the keyword list by using a transliteration model.

According to the embodiment of the present invention, it is possible to construct a multigraph structure showing the relationship between keywords for the keywords extracted from the product information, and in the case of foreign words, the original word notation is automatically extracted as a synonym. Can be added to the multigraph.

According to the embodiment of the present invention, the original word notation of the loan word can be extracted from the word appearing in the product name or the product-related query by using the transliteration model for the loan word and processed as a synonym.

It is a figure which showed the example of the network environment in one Embodiment of this invention. It is a block diagram for demonstrating the internal structure of the electronic device and the server in one Embodiment of this invention. It is a block diagram which showed the example of the component which the processor of a server can include in one Embodiment of this invention. It is a flowchart which showed the example of the method which a server can execute in one Embodiment of this invention. It is a figure which showed the structural example of the product dictionary for keyword management in one Embodiment of this invention. It is a figure for demonstrating an example of the process of converting the product information into a keyword in one Embodiment of this invention. It is a figure for demonstrating an example of the process of constructing the multigraph which shows the relationship with a keyword in one Embodiment of this invention. It is a figure which showed the example expressed by the vector numerical character of Hangul in one Embodiment of this invention. It is a figure for demonstrating an example of the separation method using the LBG data format in one Embodiment of this invention. It is a figure for demonstrating an example of the process of extracting the data set of MWS (minimal word by sound) unit in one Embodiment of this invention. It is a figure for demonstrating an example of the transliteration processing process in one Embodiment of this invention. It is a figure for demonstrating an example of the transliteration processing process in one Embodiment of this invention. It is a figure for demonstrating an example of the English synonym extraction process using the transliteration model in one Embodiment of this invention. It is a figure for demonstrating an example of the English synonym extraction process using the transliteration model in one Embodiment of this invention. It is a figure for demonstrating an example of the English synonym extraction process using the transliteration model in one Embodiment of this invention. It is a figure for demonstrating an example of the English synonym extraction process using the transliteration model in one Embodiment of this invention. It is a figure for demonstrating an example of the process of adding a synonym of the original word notation to a multigraph in one Embodiment of this invention.

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

An embodiment of the present invention relates to a technique for automatically extracting the original word notation of a loan word as a synonym.

In embodiments that include matters specifically disclosed herein, loanword notations can be automatically extracted and processed as synonyms, thereby automating keyword management, reducing work resources, and quality of service. It is possible to achieve considerable advantages in terms of improvement and service expansion.

FIG. 1 is a diagram showing an example of a network environment according to an embodiment of the present invention. The network environment of FIG. 1 shows an example including a plurality of electronic devices 110, 120, 130, 140, a plurality of servers 150, 160, and a network 170. Such FIG. 1 is merely an example for explaining the invention, and the number of electronic devices and the number of servers are not limited as in FIG.

The plurality of electronic devices 110, 120, 130, 140 may be fixed terminals or mobile terminals realized by a computer system. Examples of a plurality of electronic devices 110, 120, 130, 140 include smartphones, mobile phones, navigation systems, PCs (personal computers), notebook PCs, digital broadcasting terminals, PDAs (Personal Digital Assistants), and PMPs (Tablet Multimedia Players). ), Tablets, game consoles, wearable devices, IoT (internet of things) devices, VR (visual reality) devices, AR (agmented reality) devices, and the like. As an example, FIG. 1 shows a smartphone as an example of the electronic device 110, but in the embodiment of the present invention, the electronic device 110 substantially utilizes a wireless or wired communication method, and another via the network 170. It may mean one of a variety of physical computer systems capable of communicating with electrical devices 120, 130, 140 and / or servers 150, 160.

The communication method is not limited, and not only a communication method using a communication network that can be included in the network 170 (for example, a mobile communication network, a wired Internet, a wireless Internet, a broadcasting network, a satellite network, etc.), but also a device. Short-range wireless communication between them may be included. For example, the network 170 includes a PAN (personal area network), a LAN (local area network), a CAN (campus area network), a MAN (metropolitan area network), a WAN (wise Internet) network, etc. It may include any one or more of the networks. Further, network 170 may include, but is limited to, any one or more of network topologies, including bus networks, star networks, ring networks, mesh networks, star-bus networks, tree or hierarchical networks, and the like. Will not be done.

Each of the servers 150, 160 is realized by one or more computer devices that communicate with a plurality of electronic devices 110, 120, 130, 140 via a network 170 to provide instructions, codes, files, contents, services, and the like. good. For example, the server 150 may be a system that provides the first service to a plurality of electronic devices 110, 120, 130, 140 connected via the network 170, and the server 160 may also be a plurality of systems connected via the network 170. It may be a system that provides a second service to electronic devices 110, 120, 130, 140. As a more specific example, the server 150 is a service (as an example, keyword management) aimed at by the application through an application which is a computer program installed and executed in a plurality of electronic devices 110, 120, 130, 140. (Services, etc.) may be provided to a plurality of electronic devices 110, 120, 130, 140 as a first service. As another example, the server 160 may provide a service for distributing the files for installing and executing the above-mentioned application to a plurality of electronic devices 110, 120, 130, 140 as a second service.

FIG. 2 is a block diagram for explaining an internal configuration of an electronic device and a server according to an embodiment of the present invention. FIG. 2 describes the internal configuration of the electronic device 110 and the internal configuration of the server 150 as examples for the electronic device. Further, the other electronic devices 120, 130, 140 and the server 160 may have the same or similar internal configuration as the above-mentioned electronic device 110 or the server 150.

The electronic device 110 and the server 150 may include memories 211, 221s, processors 212, 222, communication modules 213, 223, and input / output interfaces 214, 224. The memory 211 and 221 are non-temporary computer-readable recording media, and are a RAM (random access memory), a ROM (read only memory), a disk drive, an SSD (sold state drive), and a flash memory (flash memory). Permanent mass recording devices such as, etc. may be included. Here, even if a permanent large-capacity recording device such as a ROM, SSD, flash memory, or disk drive is included in the electronic device 110 or the server 150 as another permanent recording device that is separated from the memories 211 and 221. good. In addition, the memory 211 and 221 contain an operating system and at least one program code (for example, a browser installed and executed in the electronic device 110 or an application installed in the electronic device 110 to provide a specific service. Code for etc.) may be recorded. Such software components may be loaded from a computer-readable recording medium separate from the memories 211 and 221. Such other computer-readable recording media may include computer-readable recording media such as floppy (registered trademark) drives, discs, tapes, DVD / CD-ROM drives, and memory cards. In other embodiments, software components may be loaded into memory 211, 221 through communication modules 213, 223, which are not computer readable recording media. For example, at least one program is a computer program installed by a file provided by a file distribution system (eg, server 160 described above) that distributes a developer or application installation file via a network 170 (eg, described above). It may be loaded into the memory 211 or 221 based on the application.

Processors 212 and 222 may be configured to process instructions in a computer program by performing basic arithmetic, logic, and input / output operations. Instructions may be provided to processor 212, 222 by memory 211, 221 or communication modules 213, 223. For example, the processors 212 and 222 may be configured to execute instructions received according to a program code recorded in a recording device such as memories 211 and 221.

The communication modules 213 and 223 may provide a function for the electronic device 110 and the server 150 to communicate with each other via the network 170, and the electronic device 110 and / or the server 150 may provide another electronic device (as an example). , Electronic device 120) or another server (eg, server 160). As an example, a request generated by a processor 212 of an electronic device 110 according to a program code recorded in a recording device such as a memory 211 may be transmitted to a server 150 via a network 170 under the control of a communication module 213. On the contrary, control signals, instructions, contents, files and the like provided under the control of the processor 222 of the server 150 are received by the electronic device 110 through the communication module 213 of the electronic device 110 via the communication module 223 and the network 170. It's okay. For example, control signals, instructions, contents, files, etc. of the server 150 received through the communication module 213 may be transmitted to the processor 212 and the memory 211, and the contents, files, etc. may be further recorded by the electronic device 110. It may be recorded on a medium (permanent recording device described above).

The input / output interface 214 may be a means for an interface with the input / output device 215. For example, the input device may include a device such as a keyboard, mouse, microphone, camera, and the output device may include a device such as a display, speaker, haptic feedback device, and the like. As another example, the input / output interface 214 may be a means for an interface with a device that integrates functions for input and output, such as a touch screen. The input / output device 215 may be composed of an electronic device 110 and one device. Also, the input / output interface 224 of the server 150 may be a means for connecting to the server 150 or for interfacing with a device (not shown) for input or output that the server 150 can include. As a more specific example, a service screen or content configured by using data provided by the server 150 or the electronic device 120 when the processor 212 of the electronic device 110 processes an instruction of a computer program loaded in the memory 211. May be displayed on the display through the input / output interface 214.

Also, in other embodiments, the electronic device 110 and the server 150 may include more components than the components of FIG. However, most prior art components need not be clearly shown in the figure. For example, the electronic device 110 may be realized to include at least a part of the above-mentioned input / output device 215, such as a transceiver, a GPS (Global Positioning System) module, a camera, various sensors, a database, and the like. Other components may be further included. As a more specific example, when the electronic device 110 is a smartphone, an acceleration sensor or gyro sensor, a camera module, various physical buttons, a button using a touch panel, an input / output port, which are generally included in the smartphone, are used. Various components, such as accelerometers for vibration, may be realized to be further included in the electronic device 110.

Hereinafter, a method for extracting synonyms using a transliteration model and a specific embodiment of the system will be described.

FIG. 3 is a block diagram showing an example of components that can be included in the processor of the server in one embodiment of the present invention, and FIG. 4 is a block diagram of what the server executes in one embodiment of the present invention. It is a flowchart which showed the example of the possible method.

The server 150 according to the present embodiment serves as a platform for providing a service for managing keywords used on the Internet. In particular, the server 150 may construct a multi-graph structure showing the relationship between related keywords for each product included in the product dictionary, and in the case of a foreign word included in the product dictionary, the original word notation is automatically extracted. It may be added as a synonym to the multigraph.

As shown in FIG. 3, the processor 222 of the server 150 includes a graph generation unit 310, a synonym extraction unit 320, and a synonym addition unit 330 as components for executing the synonym extraction method shown in FIG. You may be prepared. Depending on the embodiment, the components of processor 222 may be selectively included or excluded from processor 222. Also, depending on the embodiment, the components of processor 222 may be separated or merged to represent the functionality of processor 222.

Such a processor 222 and the components of the processor 222 may control the server 150 to perform steps 410 to 430 included in the synonym extraction method of FIG. For example, the processor 222 and the components of the processor 222 may be implemented to execute an instruction by the code of the operating system included in the memory 221 and the code of at least one program.

Here, the components of the processor 222 may be representations of different functions of the processor 222 that are executed by the processor 222 according to the instructions provided by the program code recorded in the server 150. For example, the graph generator 310 may be used as a functional representation of the processor 222 that controls the server 150 according to the instructions described above so that the server 150 generates a multigraph indicating the relationship between the keywords.

The processor 222 may read the necessary instructions from the memory 221 in which the instructions related to the control of the server 150 are loaded. In this case, the read instructions may include instructions for controlling the processor 222 to execute steps 410 to 430 described below.

In step 410, the graph generation unit 310 may generate a multi-graph showing the relationship between the keywords extracted from the product information for each product included in the product dictionary. The product dictionary is a database system for managing keywords, and may be composed of keywords, dictionary types, and synonym sets. The product dictionary may include product information such as sales destinations, brands, categories, and attributes that are matched with the corresponding products for each product.

The above-mentioned product dictionaries include product management (product brand automatic matching, product category automatic matching, product quality check, etc.), shopping, search, keyword automatic completion, tag dictionary, advertisement, store, NLU (natural language understanding), AI (artificial language understanding). It can be applied to various services such as intelligence).

For example, as shown in FIG. 5, when it is assumed that the product dictionary 500 is divided into the sales destination 510, the brand 520, and the category 530, the sales destination-brand, brand-category, category-from such a product dictionary 500 The relationship data of the sales destination may be extracted.

As shown in FIG. 6, in the graph generation unit 310, the seller name "matched to the corresponding product with respect to the product A whose sales destination ID is S1, the brand ID is B1, and the category ID is C1". "ABC", the brand name "DEF", and the category name "GHI" may be converted into keywords. The graph generation unit 310 converts the product information matched to the product A into the keywords "ABC", "DEF", and "GHI" for the product A, and then generates a multi-graph showing the relationship between the converted keywords. You can do it. The graph generation unit 310 may generate a multigraph by extracting the relationship data between the information matched to the product from the product dictionary as the relationship data between the keywords converted into the corresponding information.

Referring to FIG. 7, the graph generation unit 310 has a connection relationship of a sales destination-brand between the keywords "ABC" and "DEF", and a connection relationship of a brand-category between the keywords "DEF" and "GHI". , The multigraph 700 generated by the connection relationship of sales destination-category may be constructed between the keywords "ABC" and "GHI".

In the multigraph 700, the relationship between keywords may be constructed not only in a single depth tree structure but also in a tree structure having a plurality of depths. For example, if there are many series within the same brand, i.e., sub-brands, a brand-series consolidation relationship may be added.

As described above, the graph generation unit 310 may accumulate relationship data between keywords to generate a database, which may be defined as a multigraph 700 structure showing relationships between keywords.

Again, in FIG. 4, at step 420, the synonym extraction unit 320 uses a transliteration model to describe the loanwords of the loanwords among the keywords included in the product dictionary from the product names and product-related queries. May be extracted as a synonym.

Some of the words used in shopping etc. may include loanwords assimilated from various foreign languages such as English, French, Italian and German.

In the case of such loanwords, there is a problem that it is difficult to identify the source (that is, country) of the word. For example, in the case of "Chanel", if it is pronounced in English, it is transliterated as "channel" or "channel", and if it is pronounced in French, it is transliterated as "chanel", "chanel", "chanel" or the like.

In order to use the keyword "Chanel", a transliteration range that can accommodate all the pronunciations of multiple countries is required.

There are learning models that support transliteration based on deep learning, but conventional learning models are not only heavy, but cannot support pronunciation in various countries at once. Since English learning data is required for English and French learning data is required for French, there is a limit that transliteration processing of each language must be performed.

In comparison with this, in the present embodiment, synonyms for foreign words are extracted by using a transliteration model in units of MWS (minimal word by sound). Unlike traditional learning models, MWS can minimize learning data and process multilingual pronunciations with a single transliteration process.

MWS is a concept different from syllables and means the smallest unit of pronunciation. Table 1 shows the word example "happy" divided into syllables and MWS.

Even the same word may be expressed in the smallest unit by various pronunciations for each country, language, and person. Table 2 shows the MWS of the word example “happy” in Hangul.

When looking at Hangul from the dimensional division of initial voice, middle voice, and final voice, it may be expressed by vector numerical characters. For example, the basic numerical information of Hangul is as shown in the table of FIG.

In order to easily separate words by MWS, an LBG (linked bi-graded) data format may be used.

As an example, as shown in FIG. 10, after collecting a learning data set 1001 composed of a pair of English words and Hangul pronunciation notation in a product dictionary, the MWS data set 1002 may be extracted by separate learning in MWS units. ..

The transliteration processing process will be described below.

11 and 12 show the transliteration processing process for the English word "naver"®.

Processor 222 may separate the English word "naver" into candidate groups in all cases separable by MWS (S1).

Processor 222 may extract the MWS present in the pre-constructed MWS data set 1002 from the candidate group separated in step S1 (S2).

Processor 222 may map the MWS extracted in step S2 to each candidate group separated in step S1 (S3).

Processor 222 may concatenate and merge the MWSs mapped in step S3 together (S4). At this time, if even one of the candidate groups does not exist in the MWS data set 1002, the corresponding candidate group may be excluded.

Processor 222 may transliterate the MWS merged in step S4 (S5). In other words, the processor 222 may obtain a transliteration of an English word by decoding a Hangul vector mapped to the merged MWS. In the result of the S5 process of FIG. 12, the number in parentheses may mean the number of the corresponding Korean words extracted from the MWS learning data set of FIG. 10, and when necessary, various synonyms should be extracted. You can use it as a basis for your work.

An example of transliteration processing results is shown below, but is not limited to this.

As an embodiment using English synonyms as an example, Korean words and English synonyms that could not go through the transliteration model were collected, and a data set common to the MWS data set extracted from the conventional product dictionary was extracted. Later, it may be merged with the final MWS dataset and applied.

The synonym extraction unit 320 may extract synonyms of the original word notation for foreign words included in the product dictionary by undergoing the above-mentioned transliteration processing.

In the following, a specific embodiment will be described by taking an English notation of a loan word written in Hangul, that is, an English synonym as an example.

と関連するクリック商品名リスト１３１１と、

と関連するクリック商品名リスト１３２１を例示的に示している。 The synonym extraction unit 320 sets a foreign word as a target keyword among the keywords included in the product dictionary, and the product name list clicked by the user who entered the target keyword as a search query for each target keyword and related to each clicked product. Tag list to be used, and other query lists related to each clicked product (for example, queries entered by other users who clicked on the same product, other queries entered before the same user clicked on the same product, etc.) At least one of the keyword lists may be extracted. In the present specification, the click may mean the selection input of all types by the user. FIG. 13 shows the target keywords.

Related to Click Product Name List 1311 and

The click product name list 1321 related to the above is exemplified.

The synonym extraction unit 320 extracts a word in a specific language, that is, an English word from at least one keyword list among the click product name list, the click product-related tag list, and the query list extracted for the target keyword. You can do it.

と関連するクリック商品名リスト１３１１から抽出された英文単語リスト１４１１と、

と関連するクリック商品名リスト１３２１から抽出された英文単語リスト１４２１を例示的に示している（なお、図１４において、ｕｎｉｑｌｏおよびエアリズム（ａｉｒｉｓｍ）は登録商標である（他の図面においても同様））。 FIG. 14 shows the target keywords.

The English word list 1411 extracted from the click product name list 1311 related to

An English word list 1421 extracted from the click product name list 1321 related to the above is exemplified (note that in FIG. 14, uniqlo and airism are registered trademarks (the same applies to other drawings)). ..

The synonym extraction unit 320 may execute the transliteration model and extract an English word that matches the target keyword from the previously extracted English word list.

と英文単語リスト１４１１とのマッチング結果１５１１と、対象キーワード

と英文単語リスト１４２１とのマッチング結果１５２１を例示的に示している。 FIG. 15 shows the target keywords.

Matching result 1511 with English word list 1411 and target keywords

The matching result 1521 with the English word list 1421 is shown as an example.

The synonym extraction unit 320 may extract English synonym candidates by substituting English words matched for the target keyword.

に対する英文同義語候補１６１１と、

に対する英文同義語候補１６２１を示している。 FIG. 16 shows the target keywords.

English synonym candidate 1611 for

Shows English synonym candidates 1621 for.

The synonym extraction unit 320 determines the final English synonym based on whether or not an English synonym candidate appears in the click product name list and the click product-related tag / query list related to the target keyword, or the frequency of appearance. You can do it. The synonym extraction unit 320 may determine as the final English synonym when the word of the English synonym candidate matches the pattern appearing in the click product name list and the click product-related tag / query list related to the target keyword. Alternatively, when there are a plurality of English synonym candidates, the word with the pattern most frequently appearing in the click product name list and the click product-related tag / query list related to the target keyword may be determined as the final English synonym. In other words, the synonym extraction unit 320 can verify the English synonyms that are the original language notation of the target keyword by using the click product name list and the click product-related tag / query list related to the target keyword. Become.

Again, in FIG. 4, at step 430, the synonym addition unit 330 may additionally register the original word notation of the loan word as a synonym on the multigraph. At this time, the relationship between the conventional keyword on the multigraph and the synonym may be set.

Referring to FIG. 17, the keyword "ABC" and "DEF" have a connection relationship of sales destination-brand, the keywords "DEF" and "GHI" have a connection relationship of brand-category, and the keyword "ABC". In the multigraph 700 generated by the connection relationship of "sales destination-category" with "GHI", if the seller name "ABC" corresponds to a foreign word, the original word notation of the foreign word "ABC" is displayed on the multigraph 700. "XXX" may be added as a synonym 1701 and a synonym relationship may be set between the keywords "ABC" and "XXX".

Further, among the keywords on the multigraph 700, the relationship between the click product name list related to the target keyword “ABC” and the keyword appearing together with the synonym 1701 in the click product-related tag / query list may be additionally set.

Therefore, in the present embodiment, it is possible to find synonyms of the original word notation for foreign words registered in the product dictionary and automatically register them.

By extracting the keywords in the original language as synonyms for the keywords registered in the product dictionary using the above method and additionally registering them on the multigraph, the multi-synonyms can be processed while solving the synonym processing between the keywords. The graph can be expanded automatically.

By automatically processing the original word notation of loanwords as synonyms using a transliteration model, it is possible to realize automation of keyword management work and dramatically reduce resources for related work.

In the above, the synonym processing process using the transliteration model is described as a subordinate process performed after the process of generating the multigraph, but the present invention is not limited to this, and depending on the embodiment, the multigraph may be used. Can also be carried out as a separate and independent process, regardless of.

As described above, according to the embodiment of the present invention, it is possible to construct a multigraph structure showing the relationship between the keywords for the keywords extracted from the product information, and in the case of a loan word, the original language is expressed as a synonym. Can be automatically extracted and added to the multigraph. Further, according to the embodiment of the present invention, it is possible to extract the original word notation of the loan word from the word appearing in the product name or the product-related query by using the transliteration model for the loan word and process it as a synonym. can.

The devices described above may be implemented by hardware components, software components, and / or combinations of hardware components and software components. For example, the apparatus and components described in the embodiments include a processor, a controller, an ALU (arithmetic logic unit), a digital signal processor, a microcomputer, an FPGA (field programgable gate array), a PLU (programmable log unit), a microprocessor, and the like. Alternatively, it may be implemented using one or more general purpose computers or special purpose computers, such as various devices capable of executing and responding to instructions. The processing device may execute an operating system (OS) and one or more software applications running on the OS. The processing device may also respond to the execution of the software, access the data, and record, manipulate, process, and generate the data. For convenience of understanding, one processing device may be described as being used, but one of ordinary skill in the art may include a plurality of processing elements and / or a plurality of types of processing elements. You can understand. For example, the processing device may include multiple processors or one processor and one controller. Also, other processing configurations such as parallel processors are possible.

The software may include computer programs, codes, instructions, or a combination of one or more of these, configuring the processing equipment to operate at will, or instructing the processing equipment independently or collectively. You may do it. The software and / or data is embodied in any type of machine, component, physical device, computer recording medium or device to be interpreted based on the processing device or to provide instructions or data to the processing device. May be good. The software is distributed on a computer system connected by a network and may be recorded or executed in a distributed state. The software and data may be recorded on one or more computer-readable recording media.

The method according to the embodiment may be realized in the form of program instructions that can be executed by various computer means and recorded on a computer-readable medium. Here, the medium may be a continuous recording of a computer-executable program or a temporary recording for execution or download. Further, the medium may be various recording means or storage means in the form of a combination of a single piece of hardware or a plurality of pieces of hardware, and is not limited to a medium directly connected to a certain computer system, but is distributed over a network. It may exist. Examples of media include hard disks, floppy (registered trademark) disks, magnetic media such as magnetic tapes, optical media such as CD-ROMs and DVDs, optical magnetic media such as floptic discs, and It may include a ROM, a RAM, a flash memory, and the like, and may be configured to record program instructions. Other examples of media include recording media or storage media managed by application stores that distribute applications, sites that supply or distribute various other software, servers, and the like.

As described above, the embodiments have been described based on the limited embodiments and drawings, but those skilled in the art will be able to make various modifications and modifications from the above description. For example, the techniques described may be performed in a different order than the methods described, and / or components such as the systems, structures, devices, circuits described may be in a different form than the methods described. Appropriate results can be achieved even if they are combined or combined, or replaced or replaced by other components or equivalents.

Therefore, even if the embodiments are different, they belong to the attached claims as long as they are equivalent to the claims.

222: Processor 310: Graph generation unit 320: Synonym extraction unit 330: Synonym addition unit

Claims (19)

It is a synonym extraction method performed by a computer system.
The computer system comprises at least one processor configured to execute a computer-readable instruction contained in memory.
The synonym extraction method is
The stage of generating a multigraph showing the relationship between keywords on the product dictionary by the at least one processor.
A step of extracting a keyword list related to the target keyword for a target keyword corresponding to a loan word among the keywords included in the product dictionary by the at least one processor .
The stage of extracting the original word notation keyword of the target keyword from the keyword list as a synonym of the target keyword by using the transliteration model by the at least one processor , and
A method for extracting synonyms, which comprises a step of additionally registering the synonyms in the multigraph by the at least one processor . The stage of extracting the keyword list is
Extract at least one of the product name list clicked after inputting the target keyword as a search query, other query lists related to each clicked product, and tag list related to each clicked product.
The synonym extraction method according to claim 1. The stage of extracting the synonyms is
Extract the synonyms using a transliteration model that uses the smallest unit of pronunciation.
The synonym extraction method according to claim 1 or 2. The stage of extracting the synonyms is
The synonym extraction method according to claim 3, further comprising a step of separating the target keyword into the smallest unit of pronunciation by using a data format of LBG (linked bi-graded). The stage of extracting the synonyms is
The synonym extraction method according to claim 1 or 2, which comprises a step of extracting a word matched with the target keyword from the words extracted from the keyword list using the transliteration model as the synonym. The stage of extracting the synonyms is
The synonym extraction method according to claim 5, further comprising a step of extracting a synonym candidate by substituting the matched word for the target keyword. The stage of extracting the synonyms is
Extract words of a specific language from the keyword list
The synonym extraction method according to claim 5, which comprises the above. The stage of extracting the synonyms is
Prior to the step of extracting the word matched with the target keyword from the words extracted from the keyword list as the synonym, if the target keyword consists of compound words, the step of separating into a single word is further included. , The synonym extraction method according to claim 5. The stage of extracting the synonyms is
When it matches the pattern that appears in at least one of the product name list clicked after entering the target keyword as a search query, other query lists related to each clicked product, and tag list related to each clicked product. It is determined as the synonym, or when there are a plurality of synonym candidates, it appears most frequently in at least one of the clicked product name list, the other query list, and the tag list related to the target keyword. The synonym extraction method according to claim 5, further comprising a step of determining a word having a pattern to be used as a final English synonym . The stage of extracting the synonyms is
A word that matches the target keyword in at least one of the product name list clicked after entering the target keyword as a search query, other query lists related to each click product, and the tag list associated with each click product. The synonym extraction method according to claim 5, further comprising a step of determining the synonym based on whether or not the item appears or how often the item appears. The generation stage is
For each product included in the product dictionary, the stage of converting the product information matched to the corresponding product into a keyword, and extracting the relationship data between the information matched to the product from the product dictionary as the relationship data between the keywords. The synonym extraction method according to claim 1 , which comprises the step of performing. The stage of additional registration is
Add the relationship between the target keyword and the synonym on the multigraph .
The synonym extraction method according to claim 10 . A program for causing a computer to execute the synonym extraction method according to any one of claims 1 to 12 . It ’s a computer system,
Contains at least one processor configured to execute computer-readable instructions contained in memory.
The at least one processor
Graph generator that generates a multi-graph showing the relationship between keywords in the product dictionary,
After extracting a keyword list related to the target keyword for the target keyword corresponding to a loan word among the keywords included in the product dictionary, the original word notation keyword of the target keyword is used from the keyword list using a transliteration model. As a synonym of the target keyword, a synonym extraction unit, and
Synonym addition unit that additionally registers the synonym in the multigraph
A computer system. The synonym extraction unit
Extract at least one of the product name list clicked after inputting the target keyword as a search query, other query lists related to each clicked product, and tag list related to each clicked product.
The computer system according to claim 14 . The synonym extraction unit
Extract the synonyms using a transliteration model that uses the smallest unit of pronunciation.
The computer system according to claim 14 . The synonym extraction unit
Using the LBG (linked bi-graded) data format, the target keyword is separated into the smallest units of pronunciation.
The computer system according to claim 16 . The synonym extraction unit
Among the words extracted from the keyword list using the transliteration model, the words that match the target keyword are extracted as synonyms.
The computer system according to claim 14 . The synonym extraction unit
A word that matches the target keyword in at least one of the product name list clicked after entering the target keyword as a search query, other query lists related to each click product, and the tag list associated with each click product. Determines the synonym based on whether or not it appears, or how often it appears, or matches a pattern that appears in at least one of the clicked product name list, the other query list, or the tag list. In this case, it is determined as the synonym, or when there are a plurality of synonym candidates, at least one of the clicked product name list, the other query list, and the tag list related to the target keyword. Determine the word with the most frequently occurring pattern as the final English synonym ,
The computer system according to claim 18 .

Images