KR20230066922A - Method for partner matching using machine reading comprehension and computer readable recording medium thereof - Google Patents

Abstract

According to various embodiments of the present disclosure, provided is a computer-readable storage medium for storing one or more programs. When the programs are executed by at least one processor of a computing device having a communication circuit, the programs include instructions for leading the computing device to: collect a plurality of user information input from user devices of a plurality of users; define a plurality of user characteristic texts related to each of the users to use the user information for machine reading comprehension (MRC) on the basis of a text; receive a request message including a partner searching request input by a user from another user device; generate a question text by preprocessing the request message to be used for the MRC; determine a matching score by extracting an answer section for the preprocessed question text from each of the user characteristic texts using an MRC engine; and determine at least a part of the users corresponding to the user characteristic text in which the matching score is a threshold or more among the user characteristic texts as a subject of the partner searching request. In addition, various other embodiments are possible. Therefore, a partner with similar interests and tastes can be more precisely matched.

Description

Partner matching method using machine reading comprehension and computer readable recording medium

It relates to a partner matching method using machine reading comprehension and a computer readable recording medium.

As offline communities that have existed centered on schools, regions, and workplaces are rapidly dissolving, and the corona (COVID-19) incident overlaps and enters a hyper-individual society, the need for online community services based on loose connections is greater than ever. It has risen. However, currently, these online community services have a disadvantage in that it is difficult to find a dating partner or pursue diversity because they are biased toward a specific field of interest.

For example, Korean Patent Publication No. 10-2021-0109372 (published on September 6, 2021) discloses a prior invention called "golf partner matching method and device". However, the number of users using the prior invention for matching partners based on a unified interest of golf is limited in activities that can be enjoyed together with the number of users. Therefore, an interest-based online community service that helps more people enjoy leisure time with friends who have common tastes is needed.

Republic of Korea Patent Publication No. 10-2021-0109372 (published on September 6, 2021)

The present disclosure intends to provide a partner matching method and a computer readable recording medium that provide a service for matching partners with common interests and tastes as an interest-based community service.

In addition, the present disclosure allows a user to directly input a request message to find a partner in the form of a sentence, rather than simply searching for a partner with interests and tastes in accordance with the form suggested by the service, and through machine reading of the corresponding request. It is intended to provide a partner matching method using machine reading and a computer readable recording medium that determines a user suitable for a message as a partner.

The technical problem to be achieved in this document is not limited to the above-described technical problem, and other technical problems not mentioned will be clearly understood by those skilled in the art from the description below. .

According to various embodiments of the present disclosure, in a computer readable storage medium storing one or more programs, the one or more programs, when executed by at least one processor of a computing device having communication circuitry, for a plurality of users. Collecting a plurality of user information input from the user device; Defining a plurality of user characteristic texts related to each of the plurality of users so that the plurality of user information can be used for machine reading comprehension (MRC) based on text; Receive a request message including a request for finding a partner input by a user from another user device; generating question text by preprocessing the request message so that it can be used for the machine reading comprehension; extracting an answer section for the preprocessed question text from each of the plurality of user characteristic texts using a machine reading engine to determine a matching score; and instructions for causing the computing device to determine at least some of users corresponding to user characteristic text having a matching score equal to or greater than a threshold among the plurality of user characteristic texts as targets of the partner search request.

The one or more programs according to various embodiments of the present disclosure may cause the computing device to determine a user corresponding to a user characteristic having a matching score equal to or higher than a threshold among the plurality of user characteristic texts as a target of the partner search request. It may contain instructions that

The one or more programs according to various embodiments of the present disclosure cause the computing device to correct the matching score by performing sentiment analysis on a user characteristic text having a matching score greater than or equal to a threshold value among the plurality of user characteristic texts. It may contain instructions that

In the one or more programs according to various embodiments of the present disclosure, each of the plurality of user information further includes residence information of each of the plurality of users, and the one or more programs may include at least one of a computing device having a communication circuit. Instructions that, when executed by one processor, cause the computing device to be compared with the user's residence information of the other user device and to exclude the user from the partner search request if a certain criterion is not satisfied. can include more.

According to various embodiments of the present disclosure, each of the plurality of user information further includes personality information of each of the plurality of users, and the one or more programs are executed by at least one processor of a computing device having a communication circuit. When executed, it may further include instructions that cause the computing device to exclude the user from the target of the partner search request when the user's personality information of the other user device does not meet a predetermined criterion. .

According to various embodiments of the present disclosure, the one or more programs including instructions that cause the computing device to preprocess the request message to be used for machine reading comprehension to generate a query text, the one or more programs including keywords in the request message and instructions that cause the computing device to extract and generate the question text using the extracted keywords.

In various embodiments of the present disclosure, each of the plurality of user information includes information on at least one area of interest of each of the plurality of users, and the one or more programs convert the plurality of user information into text. and instructions that cause the computing device to define a plurality of user specific texts associated with each of the plurality of users so that they can be used for machine reading.

In various embodiments of the present disclosure, when the one or more programs are executed by at least one processor of a computing device having a communication circuit, the user for the user determined as the target of the partner search request to the other user device. It may include instructions that cause the computing device to transmit information.

The one or more programs according to various embodiments of the present disclosure may check whether a user of another user device has achieved a challenge as a qualification requirement; and instructions for causing the computing device to authorize an operation of receiving text included in a request to find a partner from the other user device only when the user of the other user device achieves a challenge as a qualification. can

A partner matching method and a computer readable recording medium using machine reading according to various embodiments of the present disclosure are a conventional matching system and a computer readable recording medium that simply compare parameters through machine reading of user's direct input information and selective input information. By comparing, it is possible to more precisely match partners with interests and tastes.

Effects obtainable in the present disclosure are not limited to the above-mentioned effects, and other effects not mentioned will be clearly understood by those skilled in the art from the description below.

1 illustrates a system environment in which a user can interact with a user device as a local computing device to receive a service that matches other users as partners.
2 is a detailed configuration diagram of a partner matcher according to various embodiments of the present disclosure.
3A to 3D illustrate exemplary screens in which user information is input and a user profile generated based on the input user information is displayed on a user device, according to various embodiments of the present disclosure.
4A illustrates an exemplary screen displayed on a user device to receive a request to find a partner and a schematic pre-processing process, according to various embodiments of the present disclosure.
4B illustrates an example screen on which user information for other users determined as partners is displayed on a user device, in accordance with various embodiments of the present disclosure.
5 is a flow diagram illustrating a process of matching a partner using machine reading in accordance with various embodiments of the present disclosure.
6 is a flowchart illustrating a specific process performed when a request message including a request to find a partner is received according to various embodiments of the present disclosure.
7 illustrates a computer program product that may be used to operate a partner matching system using machine reading, in accordance with various embodiments of the present disclosure.

Hereinafter, with reference to the accompanying drawings, embodiments and embodiments of the present disclosure will be described in detail so that those skilled in the art can easily practice them. However, this disclosure may be embodied in many different forms and is not limited to the embodiments and examples set forth herein.

Specific structural or functional descriptions of the embodiments are disclosed for illustrative purposes only, and may be modified and implemented in various forms. Therefore, the embodiments are not limited to the specific disclosed form, and the scope of the present specification includes changes, equivalents, or substitutes included in the technical spirit.

Although terms such as first or second may be used to describe various components, such terms should only be construed for the purpose of distinguishing one component from another. For example, a first element may be termed a second element, and similarly, a second element may be termed a first element.

It should be understood that when an element is referred to as being “connected” to another element, it may be directly connected or connected to the other element, but other elements may exist in the middle.

Singular expressions include plural expressions unless the context clearly dictates otherwise. In this specification, terms such as "comprise" or "have" are intended to designate that the described feature, number, step, operation, component, part, or combination thereof exists, but one or more other features or numbers, It should be understood that the presence or addition of steps, operations, components, parts, or combinations thereof is not precluded.

Unless defined otherwise, all terms used herein, including technical or scientific terms, have the same meaning as commonly understood by one of ordinary skill in the art. Terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with the meaning in the context of the related art, and unless explicitly defined in this specification, it should not be interpreted in an ideal or excessively formal meaning. don't

1 illustrates a system 100 in which a user can interact with a user device 102 as a local computing device to receive services that match other users as partners. User device 102 may be any computing device, such as a desktop computer, notebook computer, tablet computer, smart phone, game console, television, or the like. User device 102 may communicate with one or more networked computing devices 106 (or servers) over network 104 . Network 104 may be any one or more types of data communication networks, such as LAN, WAN, Internet, telephone network, cable network, peer-to-peer network, mesh network, and the like. Network computing device 106 may be any type or combination of types of computing devices, such as network servers, web servers, file servers, supercomputers, desktop computers, and the like. The network computing device 106 may be coupled with a plurality of other user devices 108 . Another user device 108 may be substantially the same device as the user device 102, and various embodiments of the present disclosure will be described with reference to the user device 102 in the following description.

The partner matching system 110 may include at least one memory 112 and at least one processor 111 . Processor 111 may include any combination of CPU, graphical processing units (GPUs), single core processors, multi-core processors, application specific integrated circuits (ASICs), and the like. At least one processor 111 may be implemented in software and/or firmware in addition to hardware implementation. A software or firmware implementation of processor 111 may include computer- or machine-executable instructions written in any suitable programming language to perform the various functions described above. A software implementation of processor 111 may be stored in whole or in part in memory 112 .

The memory 112 may store programs of instructions that may be loaded and executed on the processor 111 and data generated during execution of these programs. Examples of programs and data stored on memory 112 include operating systems that control the operation of hardware and software resources available to user device 102, network computing device 106, other user devices 108, and hardware devices. It may include drivers for interacting with devices, communication protocols for sending and receiving data to and from the network 104 and other computing devices, and additional software applications. Depending on the configuration and type of user device 102 and/or network computing device 106 , memory 112 may be volatile (such as RAM) or non-volatile (such as ROM or flash memory).

Partner matching system 110 may also include additional computer-readable media, such as removable storage, non-removable storage, local storage, and/or remote storage. Memory 112 and any associated computer-readable media may provide storage of computer-readable instructions, data structures, program modules, and other data. Computer-readable media may include at least two types of computer-readable media: computer storage media and communication media.

Computer storage media includes volatile and nonvolatile, removable and non-removable media implemented in any technology or method for storage of information such as computer readable instructions, data structures, program modules, or other data. Examples of computer storage media include RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, DVD or other optical storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, or other storage devices. , or other non-transportable media that can be used to store information that can be accessed by a computer, but is not limited thereto. On the other hand, communication media may embody computer readable instructions, data structures, program modules, or other data in a modulated data signal such as a carrier wave or other transport mechanism.

Partner matching system 110 may reside in whole or in part on one or both of user device 102 and network computing device 106 . Accordingly, partner matching system 110 may be a distributed system in which various physical and data components are located in one or more locations and work together to perform the role of partner matching system 110 . In some implementations, all features of partner matching system 110 may reside on user device 102 .

The partner matching system 110 may include a partner matcher 113 that calculates a matching score for each user for the user. In some embodiments, partner matcher 113 may be a storage unit that resides in whole or in part within memory 112 . In some embodiments, partner matcher 113 may be implemented as part of processor 111, such as part of an ASIC. Like partner matching system 110 itself, partner matcher 113 may reside in whole or in part on one or both of user device 102 and network computing device 106 . In various embodiments, the partner matcher 113 may be included in at least one computer program product. Computer program products may be traded between sellers and buyers as commodities. A computer program product is distributed in the form of a device-readable storage medium (eg compact disc read only memory (CD-ROM)), or through an application store (eg Play Store ^TM , or App Store ^TM ) or both. It can be distributed (eg, downloaded or uploaded) directly or online between user devices (eg, smartphones) of the user. In the case of online distribution, at least part of the computer program product may be temporarily stored or temporarily created in a device-readable storage medium such as a manufacturer's server, an application store server, or a relay server's memory.

In various embodiments, when a part or all of the partner matcher 113 is redundantly located on a plurality of computing devices, which computing device is used to implement the partner matcher 113, the related processing speed, The selection may be based on the rate of transmission of information over the network 104 and/or other factors.

2 is a detailed configuration diagram of a partner matcher according to various embodiments of the present disclosure. FIG. 2 may be the partner matcher 113 of the system 100 shown in FIG. 1 .

Referring to FIG. 2 , the partner matcher 113 according to various embodiments of the present disclosure includes a user information collection module 210, a user information database 220, a user characteristic text definition module 231, and preprocessing of a partner search request. Includes a text pre-processing module 230 including a module 232, a partner matching module 240, a machine reading comprehension engine 250, a sentiment analysis engine 770, a chat room creation module 270, and a challenge achievement module 280 can do.

In various embodiments, the user information collection module 210 may collect user information input by a user from another user device (eg, 108 of FIG. 1 ). For example, the user information may include, as selection input information, personality type (eg, MBTI), residence information, interests, and self-development activities. User information is direct input information, and may include a self-introduction letter and a self-development plan.

In various embodiments, the user information database 220 may be implemented as a single physical machine or may be implemented as multiple physical machines. The user information collection module 210 registers a user of the user device 102 (or another user device 108), which is a client device, based on the interworking with the user information database 220, or from the user device 102. Authentication of the user may be performed based on at least on a relationship between the received account information and the account information stored in the user information database 220 that stores the user information. The partner matcher 113 (or the user information collection module 210) creates a user profile by customizing information about the user of the user device 102 based on the linkage with the user information database 220. and the created user profile can be used to provide partner matching services.

3A to 3D illustrate exemplary screens in which user information is input and a user profile generated based on the input user information is displayed on a user device, according to various embodiments of the present disclosure. The screens 310, 320, 330, and 340 of FIGS. 3A to 3D are web pages, text documents, word processing documents, spreadsheet documents, or other arbitrary documents including documents written in markup languages such as HTML or XML. It may be a graphic user interface displayed by a type or application, but is not limited thereto.

Referring to FIG. 3A , the first screen 310 may be a screen that is displayed on the user device 102 and receives direct input information for the user. In various embodiments, the first screen 310 is a UI (user interface) for receiving direct input information, and includes a user image input object 311, a user name input object 312, and a self-introduction input object 313. , and an SNS input object 314.

Direct input information may be used to determine whether or not it is suitable as a target for finding a partner requested by another user. For example, in the self-introduction input object 313, the user may enter a sentence (or text) introducing himself using the keyboard on the user device 102. The sentence entered in the self-introduction input object 313 is machine It can be used for reading comprehension and can be used to find a partner. For example, information about a user image, user name (or nickname), and SNS may be directly communicated to other users to help other users find other partners.

Referring to FIG. 3B , the second screen 320 may be a screen that is displayed on the user device 102 and receives selection input information for the user. In various embodiments, the second screen 320 may display a user interface (UI) for receiving an input of a field of interest included in the selection input information. Areas of interest displayed on the second screen 320 may include, for example, career, culture and arts, writing, lifestyle, food & drink, outdoor, beauty & health, and detailed fields of each area of interest may also be included. can

Also, referring to FIG. 3C , the third screen 330 may be another screen that is displayed on the user device 102 and receives selection input information for the user. In various embodiments, the third screen 330 is a user interface (UI) that receives input and may include a personality type input object 331 and a residence setting object 332 .

The selection input information may be used to determine whether or not it is suitable as a target for finding a partner requested by another user. For example, the similarity rate of the field of interest may be determined by comparing the information on the field of interest selected based on the UI displayed on the second screen 320 with the field of interest selected by other users. For another example, information on a field of interest selected based on the UI displayed on the second screen 320 may be used to define (or create) a sentence (or text) related to the user's characteristics.

For another example, the personality type selected by the user from the personality type input object 331 displayed on the third screen 330 will be used to determine the matching success rate through comparison with the personality types of other users based on a predetermined criterion. can Specifically, if a user selects 'INFJ' as the personality type, the matching success rate can be determined to be high when the personality type of other users is 'ENFP', and it is different from the user whose personality type is 'ESFP'. It may be determined that the matching success rate is low.

In addition, the user's residence information (or location information) selected by the user in the residence information object 331 displayed on the third screen 330 is compared with the residence information of other users according to the distance setting criterion set by the user. It can be used to determine the matching success rate through For example, a user located within a range of at least 500 m and less than 2 km according to the distance setting criterion may be determined to have a high matching success rate, and other ranges may be determined to have a low matching success rate.

In various embodiments, the matching success rate determined according to the user's selection input information is an absolute criterion for finding a partner, and a decision result based on other information (eg, direct input information) (eg, direct input information) when the predetermined criterion is not met. Matching score) can be an absolute criterion for not being recommended as a partner. In another embodiment, the matching success rate may be a criterion for correcting a determination result based on other information.

Referring to FIG. 3D, a fourth screen 340 shows a user profile generated based on user information input through at least one of the screens 310, 320, and 330 of FIGS. 3A to 3C. For example, the fourth screen 340 includes a first area 341 displaying a user image and name, a second area 342 displaying self-introduction as direct input information, and displaying fields of interest as selection input information. A third area 343 may be included.

Referring back to FIG. 2 , the text pre-processing module 230 may pre-process the information collected by the partner matcher 113 so that it can be used in a language model. A language model is a model that assigns probabilities to word sequences in order to model the phenomenon of language. In other words, a language model is a model that finds the most natural word sequence. The language model may be generated based on a method using statistics or a method using an artificial neural network. The text pre-processing module 230 uses natural language processing (NLP) technology to analyze and process morphemes, syntax, etc., which are structural units of natural language, as a language normally used by humans, to characterize the user. Text can be preprocessed. In detail, the text pre-processing module 230 may translate the partner search request and user information into text in a form in which a language model or machine reading comprehension can be used through natural language processing.

In various embodiments, the partner finding request preprocessing module 230 preprocesses a sentence for a partner finding request input by another user device 108 (or user device 102) to obtain query text can create Specifically, the partner search request pre-processing module 230 may extract keywords from a sentence for a partner search request and define question text using the extracted keywords. In various embodiments, a 'keyword' may be a field of interest or a detailed field of interest as selection input information. In other words, the partner search request pre-processing module 230 may extract a keyword for a predetermined area of interest or detailed area of interest from a sentence for requesting a search for a partner input by another user device 108 .

4A illustrates an exemplary screen displayed on a user device to receive a request to find a partner and a schematic pre-processing process, according to various embodiments of the present disclosure.

Referring to FIG. 4 , a fifth screen 410 is a user interface (UI) that directly receives a request for finding a partner, and may include a request message input box 411 . For example, in the request message input box 411, the user may enter a sentence (or text) introducing himself using the keyboard on the user device 102. The sentence entered in the request message input box 411 is, Keywords may be extracted through a preprocessing process by the partner search request preprocessing module 230 . For example, in Figure 4, "Those who want to go to see the stars and sunrise of Mt. Seorak, join us! It's not easy going alone. Let's share hiking information together and have a good time. Let's become human tripods at the top!" Keywords 412 such as 'dawn hiking', 'night hiking', 'information sharing', and 'photographing' may be extracted from the partner search request message. The partner search request pre-processing module 230 may generate question text 413 to be used for machine reading comprehension through the extracted keyword 412 . For example, question text 413 may be generated based on the keyword 412, "Do you like to share information on mountain climbing at dawn or mountain climbing at night, climb together, and take pictures together at the summit?" .

In various embodiments, the user characteristic text definition module 232 may use information collected by the partner matcher 113 (or user information collection module 210), for example, direct input information and/or optional input. Information can be preprocessed so that it can be used in a language model. In various embodiments, the user characteristic text definition module 232 may generate user character text based on information input by the user device 102 . For example, the user characteristic text definition module 232 may generate user characteristic text by pre-processing a self-introduction sentence input by the user through the self-introduction input object 313 .

In various embodiments, the user characteristic text definition module 232 may generate user characteristic text by pre-processing the field of interest selected through a UI (eg, the second screen 320) that receives the field of interest as selection input information. can For example, if a user selects 'start-up', 'design', 'running', or 'mountain climbing' as an area of interest, this is used as the user characteristic text for the user "I am a startup, design, or design startup". I am interested in, and running and mountain climbing as hobbies" may be generated.

In various embodiments, the user characteristic text definition module 232 may generate user characteristic text by pre-processing location information input through the residence setting object 332 as selection input information. For example, if the user's residence is selected as 'Mapo-gu, Seoul' and the distance setting criterion is 500m or more and 2km or less, using this as the user characteristic text for the user, "I live in Mapo-gu, Seoul, and my partners are Yongsan-gu and Seodaemun-gu. The text "I hope you live in ." may be generated.

As described above, in various embodiments, interests, residence information, and personality type information as optional input information are not used to create or define user characteristic text, but are combined with other users' interests, residence information, and personality type information. It can be used as an absolute criterion for partner matching through comparison of , or used to correct the judgment result (eg, matching score) based on other information.

In various embodiments, partner matching module 240 may use Machine Reading Comprehension (MRC) for user characteristic text for each of a plurality of users, which is defined by user characteristic text definition module 232. , Matching scores for the question text defined by the partner search request preprocessing module 231 may be determined. The matching score may represent the degree of relation between the question text and the user characteristic text and may be determined numerically. For example, it is up to the user who created the partner search request message to decide which criteria, request target, or area of interest to include in the question text and user property text, so user property text for other users is defined. It is checked whether it is related to the question text, and when the relevance is high, a high matching score may be determined. Here, high relevance may mean that an answer section to the question text can be extracted from the user characteristic text.

In various embodiments, the partner matching module 240 uses the machine reading comprehension engine 250 to determine a matching score between at least a portion of preprocessed user characteristic text and question text for a plurality of users, If the score is higher than a certain standard, the corresponding text may be judged to be highly relevant to the question.

For example, the partner matching module 240 assumes the case of "Do you like to share information about morning mountain climbing or night mountain climbing, climb together, and take pictures together at the summit?" as a defined question text. do. If, as a machine reading target for the above question text, a matching score of 0.8 can be determined for the user characteristic text "I like outdoor activities, and I like to greet a refreshing morning with running", and "I like mountain climbing. A matching score of 0.7 may be determined for the user characteristic text "I like having a leisurely time". In the case of the latter, although the field of interest in 'hiking' coincides, the degree of relevance to 'dawn mountain climbing' and 'night mountain climbing' can be evaluated low from the keyword 'relaxed time'.

For another example, as a machine reading target for the question text, a matching score may be determined as 0 for user characteristic text "I like watching movies or music, and I like going to learn something". Because no answer (or answer section) to the question text can be extracted from the user characteristic text, the matching score can be determined as 0.

In various embodiments, the partner matching module 240 may determine a matching score for a question text defined for a plurality of user characteristic texts and extract only user characteristic text when the matching score is equal to or greater than a threshold value. For example, assuming that there are a plurality of user characteristic texts targeting 100 people as analysis targets, the partner matching module 240 only uses user characteristic texts targeting 30 people whose matching scores are 0.7 or higher as a predetermined threshold. can be determined to be related to this defined question text.

In various embodiments, the partner matching module 240 may correct a matching score determined using machine reading by a matching success rate determined by other information. For example, a matching success rate determined by personality type or residence information as selection information input by a user may correct a matching score. For example, in a user characteristic text with a matching score of 0.6 determined using machine reading comprehension, the personality type input by the user of the corresponding user characteristic text is a personality type that is very compatible with the personality type of the user who entered the partner search request message. In this case, the matching success rate is determined as 1.2, and the matching score of 0.6 points is multiplied by 1.2 to be corrected to 0.72 points. Accordingly, even when the matching score determined using machine reading comprehension is 0.6. It can be corrected to 0.72 points by the matching success rate and determined as a target having a threshold value of 0.7 or higher. For another example, even when the matching score determined by machine reading is equal to or higher than the threshold value, it may be corrected by the matching success rate and determined as a score lower than the threshold value.

In various embodiments, the partner matching module 240 weights the user information collected by the user information collection module 210 and the matching score determined as a result of machine reading according to the suitability priority, so that the partner search request You can decide which other users are suitable for you. Parameters weighted according to suitability priority may further include personality type (e.g., MBTI), field of interest, residence information (or location information), and self-development activities, and user characteristic text and questions determined by self-introduction, etc. Matching scores determined as a result of machine reading with text may be included. For example, the suitability priority may be determined in the order of machine reading matching score > personality type > residence information > self-development activity > interest, and accordingly, the comparison priority of each parameter may be determined. A final matching score for determining a user to be determined as a partner may be determined according to a comparison result of parameters according to the above suitability priority order. The final matching score may mean that the matching score of the machine reading result is corrected according to the suitability priority, similarity, relevance, and the like of other parameters.

Meanwhile, the machine reading engine 250 may use a word embedding method of dense vectoring the meaning of words included in a sentence or text.

Word embedding is a method of expressing a word as a dense vector, and may mean converting a word into a dense representation. Dense representation is the opposite concept of sparse representation, and may mean unifying the dimension of vector representation of all words with a value set by the system. Also, word embedding can be implemented in a variety of ways, including LSA, Word2Vec, FastText, and Glove. The above word embedding can use a method of loading a pre-trained embedding vector with a vast amount of corpus (corpus) such as a web encyclopedia (eg Wikipedia) in order to vectorize words. . In addition, fine-tuning, which separately prepares training data suitable for user-characteristic text and defined questions, releases the upper layer of the existing model and trains it together with a new layer, and full-learning, which vectorizes by learning without using the existing model method can also be used. Such machine learning may use a pre-trained embedding model and a fine-tuned model included in the machine reading engine 250 .

The sentiment analysis engine 260 may perform sentimental analysis on user characteristic data determined by the partner matching module 240 (or the machine reading comprehension engine 250 to have a matching score equal to or greater than a threshold value).

In various embodiments, the sentiment analysis engine 260 may be a model that classifies input data into three classes of positive, negative, or neutral, and according to other embodiments, strong positive, weak positive, strong negative, It could also be a model that is trained to classify the input data into more classes, such as weak negative, neutral, etc. The sentiment analysis engine 260 pre-trains a context-sensitive natural language processing model (e.g., Bidirectional Encoder Representations from Transformers (BERT) or KoBERT) with a large amount of corpus (or documents), and then converts the pre-trained It can be constructed by fine-tuning to perform sentiment analysis as described above using a model. As described above, such pre-learning may be constructed by additionally learning open natural language processing models such as BERT or data related to areas of interest.

The sentiment analysis engine 260 may function to supplement an error that may occur when a matching score is determined simply based on the degree of relevance between the question text and the user characteristic text. For example, if "Do you like to share information about mountain climbing or night mountain climbing, climb together, and take pictures together at the summit?" is defined as the question text, as a machine reading target, "I am outdoor For the user characteristic text "I like but I hate getting up early in the morning", a matching score of 0.8 can be determined because the keyword morning is included. However, in the above case, the emotion analysis engine 260 may determine that the emotion is strong negative by performing emotion analysis on the text "I hate waking up early in the morning" extracted as an answer section, and in this case A matching score of 0.8 points can be corrected to 0 points. In another example, when the emotion analysis is weakly negative, the matching score may be multiplied by 0.8, and in another example, the matching score may be corrected by 1 or more times when the emotion analysis result is strongly positive or weakly positive.

The chat room creation module 270 may create a chat room in which a user having a matching score equal to or higher than a threshold value and a user determined as a target of a partner search request are connected. A chat room may refer to a virtual space created to exchange text messages, voice messages, images, videos, and the like between users participating in the chat room in real time. For example, the chat room creation module 270 may create a chat room to which a user requesting to find a partner and a user with the highest matching score can access at the same time. For another example, the chat room creation module 270 may create a chat room to which at least one user selected from among users whose matching score is equal to or higher than a threshold value by the user requesting to find a partner can access the same type.

4B illustrates an example screen on which user information for other users determined as partners is displayed on a user device, in accordance with various embodiments of the present disclosure. Referring to FIG. 4B , the sixth screen 420 may display a name (or nickname) of at least one user whose matching score is equal to or greater than a threshold and part of user information (eg, personality type, self-introduction).

The challenge achievement module 280 may check whether the user who inputs the partner search request message is qualified. In the partner matching system 110, in order to prevent indiscriminate request of the partner finding service, qualification requirements for using the partner finding service may be required. Accordingly, the partner matching system 110 (or the partner matcher 113) may restrict only users who have achieved a certain challenge to input a partner search request message. As an example of achieving the challenge, various embodiments are possible, such as payment of a certain amount or more, achievement of a certain attendance rate, completion of an event, selection by lottery, and the like.

5 is a flow diagram illustrating a process 500 for matching a partner using machine reading in accordance with various embodiments of the present disclosure. For ease of understanding, the processes discussed in this disclosure are described as discrete operations represented as discrete blocks. However, it should not be construed that the operations described separately in this way must be performed in the order in which they are performed. The order of performance of the described processes is not to be construed as limiting, and any block(s) of the described process blocks may be combined in any order to implement this process or alternative processes. Additionally, one or more of the operations presented may be modified or omitted.

Processes are depicted as collections of blocks within a logical flow diagram, which represent a series of operations that may be implemented in hardware, software, or a combination of hardware and software. For discussion, the processes are described with reference to the configuration and user interface of the partner matching system using machine reading described with reference to FIGS. 1 to 4 . However, processes may be performed using other systems and/or user interfaces.

Referring to FIG. 5 , in operation 501, the partner matching system 100 (or the user information collection module 210) may collect a plurality of user information input from user devices of a plurality of users. User information may include selection input information and direct input information. For example, the selective input information may include personality type (eg, MBTI), residence information, and field of interest, and direct input information may include self-introduction.

In operation 503, the partner matching system 100 (or the user characteristic text definition module 232) uses the collected plurality of user information based on text to be used for machine reading comprehension (MRC). A plurality of user characteristic texts related to each of a plurality of users may be defined (or created). For example, the partner matching system 100 may pre-process a self-introduction sentence input by the user through a self-introduction input object displayed on the user device to generate user characteristic text. In various embodiments, interest, residence information, and personality type information as selection input information are absolute standards for partner matching through comparison with other users' interests, residence information, and personality type information, or matching success rate It can also be used to calibrate the matching score.

In operation 505, the partner matching system 100 may receive a request message including a request for finding a partner input by a user from another user device. In the next operation 507, the partner matching system 100 (or the partner finding request pre-processing module 231) may generate question text by pre-processing the request message so that it can be used for machine reading. For example, the partner matching system 100 may extract keywords from a request message and generate question text using them.

In operation 509, the partner matching system 100 (or the partner matching module 240) may determine a matching score for the preprocessed question text from each of a plurality of user characteristic texts using the machine reading comprehension engine 260. there is. For example, the partner matching system 100 extracts an answer section for the defined question text from each of the plurality of preprocessed user characteristic texts using a machine reading comprehension engine, and the relationship between the question text and the answer section. A matching score may be determined based on .

In operation 511, the partner matching system 100 (or the chat room creation module 270) selects at least some of the users corresponding to the user characteristic text whose matching score is equal to or greater than a threshold value among the plurality of user characteristic texts in response to the partner finding request. target can be determined. For example, the partner matching system 100 may transmit information about users whose matching scores are greater than or equal to a threshold value to the user device that has requested to find a partner. For another example, the partner matching system 100 may transmit information about users with the top 10 matching scores to the user device that has requested to find a partner.

In operation 513, the partner matching system 100 (or the chat room creation module 270) may create a chat room in which at least some of the users determined to be the target of the partner search request and the user requesting the user partner search are connected. For example, a chat room of a user who has requested to find a user partner and a user selected by an input from a user device of the user who has requested to find a user partner may be created. For example, when a request for creating a chat room with a specific user is received from a user device of a user who has requested a search for a user partner, the partner matching system 100 sends a message requesting permission to create a chat room to the user device of the specific user. In response to this, when a signal authorizing creation of a chat room is received from the user device of the specific user, the chat room can be created.

In various embodiments, the partner matching system 100 (or the partner matching module 240) obtains a matching score for the preprocessed question text from each of a plurality of user characteristic texts using the machine reading comprehension engine 260. In operation 509 of determining, the partner matching system 100 (or the emotion analysis engine 260) performs emotion analysis on a user characteristic text whose matching score is equal to or greater than a threshold value among a plurality of user characteristic texts to correct the matching score. can For example, the partner matching system 100 may set the matching score to 0 or correct it by a predetermined multiple according to the emotion analysis result of the text extracted as the answer section of the question text from the user characteristic text.

In various embodiments, the partner matching system 100 (or the partner matching module 240) obtains a matching score for the preprocessed question text from each of a plurality of user characteristic texts using the machine reading comprehension engine 260. In operation 509 of determining, the partner matching system 100 (or the partner matching module 240) may include an operation of correcting the determined matching score with a matching success rate. The matching success rate may be determined by selection input information (eg, field of interest, personality type, residence information, etc.) input as user information.

6 is a flowchart illustrating a specific process 505 performed when a request message including a request to find a partner is received according to various embodiments of the present disclosure. Referring to FIG. 6 , in operation 601, when the partner matching system 100 receives a request message including a request to find a partner input by a user from another user device, in operation 603, the user of another user device challenges the user. can be judged whether it has been achieved. As an example of achieving the challenge, various embodiments are possible, such as payment of a certain amount or more, achievement of a certain attendance rate, completion of an event, selection by lottery, and the like. If it is determined in operation 603 that another user has not achieved the challenge, the process 505 is immediately terminated. A request message including a request to find a partner may be received from

Through the process 505 described above, it is possible to prevent an indiscriminate request for a partner finding service according to various embodiments of the present disclosure by giving a condition of achieving a challenge.

7 illustrates a computer program product 700 that may be used to operate a partner matching system using machine reading, in accordance with various embodiments of the present disclosure. An exemplary embodiment of an exemplary computer program product is provided using a signal bearing medium 710. In some embodiments, signal bearing medium 710 of one or more computer program products 700 may include computer readable medium 730 and/or recordable medium 740 .

Instructions 720 contained in signal bearing medium 710 may be executed by a computing device such as processor 111 in partner matching system 110 and stored in memory 112 . Instruction 720, when executed, causes the computing device to collect a plurality of user characteristic text; pre-processing the collected sentences of the plurality of user characteristic texts so that they can be used for machine reading comprehension (MRC); defining a topic and a question related to the topic and usable for the machine reading comprehension; extracting an answer section for the defined question from each of the plurality of pre-processed user characteristic texts using a machine reading comprehension engine, and determining a matching score; and one or more instructions for performing an operation of determining, among the plurality of user characteristic texts, a user characteristic text having a matching score equal to or greater than a threshold value as the user characteristic text for the topic.

The above description of the present application is for illustrative purposes, and those skilled in the art will understand that it can be easily modified into other specific forms without changing the technical spirit or essential features of the present application. Therefore, it should be understood that the embodiments described above are illustrative in all respects and not restrictive. For example, each component described as a single type may be implemented in a distributed manner, and similarly, components described as distributed may also be implemented in a combined form.

In the above, the object to be claimed in the present disclosure has been specifically examined. The subject matter claimed in this disclosure is not limited in scope to the specific implementations described above. For example, in some implementations it may be in the form of hardware operatively used on a device or combination of devices, in other implementations it may be implemented in the form of software and/or firmware, and in still other implementations it may be in the form of a signal bearing medium; It may include one or more items, such as storage media. Here, the storage medium, such as a CD-ROM, a computer disk, a flash memory, etc., when executed by a computing device, such as a computing system, a computing platform, or other system, may cause the corresponding processor to execute according to the implementation described above. can be saved. Such computing devices may include one or more processing units or processors, a display, one or more input/output devices such as a keyboard and/or mouse, and one or more memories such as static random access memory, dynamic random access memory, flash memory and/or hard drives. can include

In the foregoing detailed description, various embodiments of devices and/or processes have been described using block diagrams, flow diagrams, and/or other examples. Such block diagrams, flow diagrams, and/or other examples may include one or more functions and/or operations, and each function and/or operation in a block diagram, flow diagram, and/or other example may be hardware, software, firmware, or or any combination thereof, individually or collectively. In various embodiments, some portions of the subject matter described in this disclosure may be implemented via application specific integrated circuits (ASICs), field programmable gate arrays (FPGAs), digital signal processors (DSPs), or other forms of integration. In contrast, some aspects of embodiments of the present disclosure are directed to one or more computer programs running on one or more computers (eg, one or more programs running on one or more computer systems), one or more programs running on one or more processors (eg, one or more programs running on one or more computer systems). for example, one or more programs running on one or more microprocessors), firmware, or substantially any combination thereof, which may be implemented in whole or in part equivalently on an integrated circuit, writing code for software and/or firmware. and/or the design of the circuit is within the skill of those skilled in the art in light of this disclosure. Further, those skilled in the art will understand that the mechanisms of the present disclosure may be distributed in various forms of program product, and the examples of the present disclosure apply regardless of the specific type of signal bearing medium used to actually perform the distribution. will understand

While certain exemplary techniques have been described and illustrated herein using a variety of methods and systems, those skilled in the art will appreciate the possibility of various other modifications or equivalent substitutions without departing from the claimed subject matter. Additionally, many modifications may be made to adapt a particular situation to the teachings of claimed subject matter without departing from the central concept described herein. Thus, it is intended that the claimed subject matter is not limited to the specific examples disclosed, but that such claimed subject matter may also include all implementations falling within the scope of the appended claims and their equivalents.

The scope of the present disclosure is indicated by the following claims rather than the detailed description above, and all changes or modifications derived from the meaning and scope of the claims and equivalent concepts thereof should be construed as being included in the scope of the present application. do.

Claims (7)

A computer readable storage medium storing one or more programs, the one or more programs, when executed by at least one processor of a computing device having communication circuitry, comprising:
collect a plurality of user information input from user devices of a plurality of users;
defining a plurality of user characteristic texts related to each of the plurality of users to be used for machine reading comprehension (MRC) based on the plurality of user information as text;
Receive a request message including a request for finding a partner input by a user from another user device;
preprocessing the request message to be used for the machine reading comprehension to generate question text;
using a machine reading engine, extracting an answer section for the preprocessed question text from each of the plurality of user characteristic texts to determine a matching score; and
Including instructions for causing the computing device to determine at least some of the users corresponding to the user characteristic text having the matching score equal to or greater than a threshold among the plurality of user characteristic texts as the target of the partner finding request.
A computer readable storage medium.
The one or more programs of claim 1 , further comprising instructions for causing the computing device to determine a user corresponding to a user characteristic having a matching score equal to or greater than a threshold among the plurality of user characteristic texts as a target of the request for finding a partner. heard,
Including instructions for causing the computing device to correct the matching score by performing sentiment analysis on user characteristic text having the matching score equal to or greater than a threshold value among the plurality of user characteristic texts.
A computer readable storage medium.
According to claim 1,
Each of the plurality of user information further includes personality type or residence information of each of the plurality of users,
The one or more programs, when executed by at least one processor of a computing device having a communication circuit, are compared with the user's personality type or residence information of the other user device, and if a predetermined criterion is not satisfied, the user Further comprising instructions for causing the computing device to be excluded from the target of a partner finding request.
A computer readable storage medium.
The method of claim 1 , wherein the one or more programs including instructions that cause the computing device to preprocess the request message for use in the machine reading comprehension to generate query text include:
Including instructions that cause the computing device to extract a keyword from the request message and to generate the question text using the extracted keyword.
A computer readable storage medium.
According to claim 1,
Each of the plurality of user information includes information on at least one field of interest of each of the plurality of users,
wherein the one or more programs cause the computing device to define a plurality of user specific texts associated with each of the plurality of users to be used for machine reading text of the plurality of user information;
A computer readable storage medium.
The method of claim 1, wherein the one or more programs, when executed by at least one processor of a computing device having communication circuitry,
Including instructions that cause the computing device to transmit user information about the user determined to be the target of the partner finding request to the another user device.
A computer readable storage medium.
The method of claim 1, wherein the one or more programs, when executed by at least one processor of a computing device having communication circuitry,
confirm whether a user of the another user device has achieved a challenge as a qualification; and
Including instructions that cause the computing device to authorize an operation of receiving text included in a partner search request from the other user device only when the user of the other user device achieves a challenge as a qualification requirement.
A computer readable storage medium.

Images