KR102387829B1 - Matching between client device and provider device method for providing a legal content - Google Patents

Abstract

The present invention provides a matching method between a client terminal and a provider terminal performed by a server that performs a process for matching between a client terminal requesting provision of legal content and the provider terminal. The matching method between the client terminal and the provider terminal of the present invention includes a first group object variable including a plurality of object variables that can be expressed as a multidimensional space vector in a service request message, and a plurality of object variables not belonging to the first group object variables. separating a first group residual variable including object variables of ; and calculating a matching score using the first group object variables and the first group residual variable. According to the present invention, in matching between a one-to-many or many-to-many client terminal and a provider terminal, it is possible to reduce the amount of computation for matching and improve the accuracy of matching.

Description

Matching between client device and provider device method for providing a legal content

The present invention relates to a matching method between a client terminal and a provider terminal for providing legal content based on artificial intelligence. In particular, the present invention includes a client terminal, a service provider terminal, and a server, and is provided by the server. A method of matching between a client terminal and a provider terminal based on artificial intelligence for providing legal content generated by a service provider to a client terminal through a platform.

In the prior art, necessary content was directly communicated through a direct network between a client terminal and a service provider terminal. This conventional network method is a one-to-one network, and there is a problem in that the client's dependence on the service provider for legal content requiring expertise and information asymmetry are high, and thus the efficiency of network management is reduced.

In order to communicate messages with other service providers through the client terminal, a separate network must be formed. In a network environment with high information asymmetry, it is difficult for a client terminal to proactively change a network entity. Also, in the related art, as a network capable of one-to-one communication between network entities, it is not suitable for one-to-many or many-to-many network communication, and it is difficult to maintain information security between different entities.

Korean Patent Laid-Open No. 10-2019-0177646 discloses a method of matching a customer and a service provider as a platform system capable of providing legal advice. However, in the case of the above patent, a method of matching with the counseling information analyzed by the counseling information analysis module using customer member information for customer information of age, age, gender, moving property / real estate holding status, family relationship, and occupation is disclosed. there is a bar This existing method has limitations that are difficult to apply to a platform that processes complex and diversified and large amounts of legal content requests.

In consideration of the limitations of the prior art, the AI-based matching system and method between a client terminal and a provider terminal of the present invention provides a method of providing legal content through network setting between one-to-many or many-to-many network entities. . In addition, the present invention reduces the processing time related to network matching between service provider terminals (computing devices connected and managed by a group of experts providing legal content) meeting the requirements required by the processor client terminals, and improves the accuracy of matching To provide a matching method between a client terminal and a provider terminal for improvement. In addition, although not described in this paragraph, the object of the present invention may be additionally considered within a range that can be easily inferred by detailed configuration.

In order to achieve the object of the present invention, the present invention provides a method for matching between a client terminal and a provider terminal for providing legal content, a process for matching between a client terminal requesting provision of legal content and a provider terminal It includes the following steps performed by the server.

In the matching method between a client terminal and a provider terminal of the present invention, a server receives a service request message from the client terminal by a communication interface, and receives a response message corresponding to a response to the service request message from the provider terminal to do;

First group object variables including a plurality of object variables that the server can express as a multidimensional space vector in the service request message, and a first group remainder including a plurality of object variables not belonging to the first group object variables isolating variables; and

and calculating a matching score for matching between network entities using the first group object variables or the first group residual variable.

Calculating the matching score in the present invention further uses second group object variables obtained from a response message or a second group residual variable including a plurality of object variables not belonging to the second group object variables. A matching score is calculated.

Here, each of the first and second group residual variables includes at least some text-type information. In the present invention, the server includes a first learning unit generating first and second fully connected feature values from an input image, and a second learning unit generating first and second images reconstructed by inputting the first group residual variable as an input. more wealth.

In the present invention, the service request message includes identification information for specifying the legal content to be requested, and object variables indicating attributes related to the service request. In addition, the object variables include at least two selected from the group consisting of cost, quality, speed, experience value of a service provider, and scale of a service provider related to the provision of the legal content.

The server calculates the matching score by using a second similarity value between the first image generated from the first group residual variable and the second image generated from the second group residual variable.

In addition, the server uses a first similarity value between the first group object variables and the second group object variables or a second similarity value between the first group residual variable and the second group residual variable to configure the network A matching score for matching between entities may be calculated.

In the present invention, the generating of the first and second images by the second learning unit includes obtaining first and second summary messages summarizing the first and second group residual variables, respectively, and from the first and second summary messages The method further includes generating the reconstructed first and second images.

In addition, the step of calculating the matching score by the server in the present invention comprises:

generating a set of requirement feature values for expressing the first group object variables as vectors on the multidimensional space coordinates and a set of response feature values for expressing the second group object variables as vectors on the multidimensional space coordinates; Visualizing and expressing a multidimensional spatial distribution image according to a set of requirement feature values and a set of response feature values on a screen of the client terminal; determining a correction variable for correction of a first group object variable in a multidimensional spatial distribution image visualized and expressed on a client terminal; and correcting the first group object variable in consideration of the correction variable. The server may calculate a matching score using a first similarity value between the first group object variables and the second group object variables.

Here, the requirement feature value set includes two or three different first group selection object variables selected from the first group object variables, and the response feature value set includes the second group object variable It may include two or three second group selection object variables that are different from each other.

The server visualizes and expresses on the screen of the client terminal, including expressing the first group selection object variables and the second group selection object variables as images projected on a multidimensional space. The server determines the correction variable by receiving a correction vector according to a user touch input on the screen for correcting the positions of the first group selection object variables that are visualized and displayed on the screen of the client terminal, The correction variable may be determined using the correction vector.

In the present invention, the second learning unit obtains a first summary message summarizing the first group residual variable, adds the first summary message and random noise to generate a first summed feature value, and the first summed feature value to reconstruct the first image, and determine whether the reconstructed first image is a real image or a fake image. The first learning unit applies a convolution filter to the reconstructed first image determined as a real image by the second learning unit, and uses the activated feature value obtained by activating the feature value according to the convolution filter to perform the first Fully connected feature values can be generated. In this case, the server calculates the matching score by using a second similarity value between the first fully connected feature value obtained from the reconstructed first image and a second fully connected feature value obtained from the reconstructed second image. can be calculated.

In a matching method between a client terminal and a provider terminal for achieving another object of the present invention, a communication interface receives a service request message from the client terminal, and sends a response message corresponding to a response to the service request message. receiving from a provider terminal; First group object variables including a plurality of object variables that the server can express as a multidimensional space vector in the service request message, and a first group remainder including a plurality of object variables not belonging to the first group object variables isolating variables; In the response message, second group object variables including a plurality of object variables expressible as multidimensional space vectors and second group residual variables including a plurality of object variables not belonging to the second group object variables are separated to do; and calculating a matching score for matching between network entities by using the relation between the first group object variables and the second group object variables, or between the first group residual variables and the second group residual variables. may include steps.

In a matching method between an interactive client terminal and a provider terminal for achieving another object of the present invention, a communication interface receives a service request message from the client terminal, and receives a response message corresponding to a response to the service request message receiving from the provider terminal;

First group object variables including a plurality of object variables that the server can express as a multidimensional space vector in the service request message, and a set of requirement feature values according to the second group object variables obtained from the response message; generating a set of response feature values; and

and calculating a matching score using the set of requirement feature values and the set of response feature values.

In the present invention, the requirement feature value set includes two or three different first group selection object variables selected from the first group object variables, and the response feature value set includes the second group object It includes two or three different second group selection object variables selected from the variables. According to the present invention, the requirement feature value set and response feature value set can be visualized and expressed on the screen of the client terminal, and the first group selection object variables and the second group selection object variables are displayed in a multidimensional space. It can be expressed as a projected image.

In the present invention, the requirement feature value set may include coordinates and attribute values of the requirement feature point for specifying the positions of the first group selection object variables in a multidimensional space. The attribute value further includes a position vector for indicating an attribute for the matchable region, and reverse image information for setting the matchable region. Of course, the same is true for the response feature value set.

Calculating the matching score in the present invention may be calculated using a distance between the first group selection objects and the second group selection objects.

In addition, for another object of the present invention, the present invention provides a computer-executable program including instructions for performing the above-described matching method between the client terminal and the provider terminal on the computer.

According to the artificial intelligence-based matching method between the client terminal and the provider terminal for providing legal content of the present invention, the speed and accuracy of matching decisions between network entities can be improved through artificial intelligence technology for interpreting service request messages and response messages. can be improved In addition, through the platform for matching between the client terminal and the provider terminal according to the present invention, it is easy to set up and manage a many-to-many, one-to-many network between the client terminal, which is a network entity, and the service provider terminal.

According to an embodiment of the present invention, in matching between a client terminal and a service provider terminal, artificial intelligence learning using a reconstructed and reconstructed image and adaptively calculating a matching score according to priorities, thus matching accuracy can improve

In addition, according to another embodiment of the present invention, since the status of the network is monitored and the matching score is calculated using the existing history information, trial and error caused by the matching error can be reduced, and the matching error can be prevented. Accordingly, it is possible to reduce the occurrence of an additional matching process that occurs additionally. This effect is important in a matching system between a client terminal and a provider terminal, in which the data accumulation rate is very fast over time.

In addition, according to another embodiment of the present invention, by using a network environment variable that continuously changes according to network conditions as a security key, it is possible to enhance the security of the distribution of legal content in which confidentiality is important.

1 is a conceptual diagram illustrating a matching system between a client terminal and a provider terminal for providing legal content according to an embodiment of the present invention.
2 is a detailed block diagram of a matching system between a client terminal and a provider terminal for providing legal content according to another embodiment of the present invention.
3 is a detailed block diagram illustrating a detailed operation performed by the processor 3200 of FIG. 2 .
4 is a flowchart illustrating a method of providing legal content for providing legal content according to an embodiment of the present invention.
5 is a reference diagram illustrating an example of a multidimensional spatial distribution image that may be displayed on a screen of a client terminal according to an embodiment of the present invention.
6 is a reference diagram illustrating an example of a multidimensional spatial distribution image that may be displayed on a screen of a client terminal according to another embodiment of the present invention.
7 is a detailed block diagram illustrating the learning unit 3400 of FIG. 2 .
8 is a reference diagram illustrating an example of a user screen provided through a matching system between a client terminal and a provider terminal according to the present invention.
9 is a reference diagram illustrating an example of legal content provided through a matching system between a client terminal and a provider terminal according to the present invention.
10 is a reference diagram illustrating an example of legal content related to a trademark search result provided through a matching system between a client terminal and a provider terminal according to the present invention.
11 is a flowchart illustrating an example of a method of generating a security message for enhancing security between network entities in a matching system between a client terminal and a provider terminal according to the present invention.
12 is a flowchart illustrating an example of a message generation method for message integrity in message transmission of a matching system between a client terminal and a provider terminal according to the present invention.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings. Advantages and features of the present invention, and a method for achieving them will become apparent with reference to the embodiments described below in detail in conjunction with the accompanying drawings. However, the present invention is not limited to the embodiments published below, but may be implemented in various different forms, and only these embodiments allow the disclosure of the present invention to be complete, and common knowledge in the art to which the present invention pertains It is provided to fully inform those who have the scope of the invention, and the present invention is only defined by the scope of the claims. Like reference numerals refer to like elements throughout.

Unless otherwise defined, all terms (including technical and scientific terms) used herein may be used with the meaning commonly understood by those of ordinary skill in the art to which the present invention belongs. In addition, terms defined in a commonly used dictionary are not to be interpreted ideally or excessively unless clearly defined in particular. In the present specification, terms such as “first” and “second” are for distinguishing one component from other components, and the scope of rights should not be limited by these terms. For example, a first component may be termed a second component, and similarly, a second component may also be termed a first component.

In the present specification, the identification number of each step is used for convenience of description, and the order of each step is not limited to the identification number, and each step may occur differently from the specified order unless the context clearly indicates a specific order. can That is, each step may occur in the same order as specified, may be performed substantially simultaneously, or may be performed in the reverse order. In this specification, expressions such as “have”, “may have”, “include” or “may include” indicate the existence of a corresponding feature (eg, a numerical value, function, operation, or component such as a part). and does not exclude the presence of additional features. In addition, the term '~ unit' as used herein means software or a hardware component such as a field-programmable gate array (FPGA) or ASIC, and '~ unit' performs certain roles. However, '-part' is not limited to software or hardware. '~' may be configured to reside on an addressable storage medium or may be configured to refresh one or more processors. Accordingly, as an example, '~' indicates components such as software components, object-oriented software components, class components, and task components, and processes, functions, properties, and procedures. , subroutines, segments of program code, drivers, firmware, microcode, circuitry, data structures and variables. The functions provided in the components and '~ units' may be combined into a smaller number of components and '~ units' or further separated into additional components and '~ units'.

Hereinafter, a preferred embodiment of a matching system and method between a client terminal and a provider terminal for providing legal content according to the present invention will be described in detail with reference to the accompanying drawings.

1 is a conceptual diagram illustrating a matching system between a client terminal and a provider terminal according to an embodiment of the present invention. The matching system between the client terminal and the provider terminal of FIG. 1 includes the client terminal 110 , the service provider terminal 120 , the communication network 130 , the server 140 and the history of the matching system between the client terminal and the provider terminal and a storage unit 150 .

The matching system between the client terminal and the provider terminal of FIG. 1 performs a series of transactions on the network to match the service provider terminal 120 suitable for a service request from the client terminal 110 . Here, a series of transactions, for example, calculate a matching score, select a provider suitable for a service request through this, quality control for service results provided by the provider, and normalize the results provided by the provider Includes post management processes such as format management, cost handling process, and evaluation.

The client terminal 110 is connected to the server 140 of the matching system between the client terminal and the provider terminal through the communication network 130 such as the Internet. The client terminal 110 may be implemented in various forms, such as a mobile terminal capable of wired or wireless communication, a desktop computer, and a tablet. When the client terminal is a mobile terminal, its own location information may be obtained through a built-in GPS receiver, and various sensor values obtained in the environment in which the client terminal is located may be obtained through various environmental sensors.

In this embodiment, the server 140 of the matching system between the client terminal and the provider terminal mainly performs an operation for matching between the client terminal and the provider terminal. Here, the matching between the client terminal and the provider terminal means establishing and managing a network, ie, a communicable connection, between a single or a plurality of client terminals and a single or a plurality of service provider terminals. In the present embodiment, the service to be matched between the client terminal and the provider terminal includes, for example, a legal content provision service, particularly an intellectual property content provision service. Preferably, the matching between the client terminal and the provider terminal in this embodiment is suitable for use when the service providers have different nationalities and are located in remote locations from each other. In this embodiment, the matching between the client terminal and the provider terminal is not for direct communication between the client and the service providers, but means a connection on the platform 160 that provides matching between the client terminal and the provider terminal. That is, rather than direct communication, matching is eventually established by processing performed by the processor of the server 140 , and after the matching is established, communication between network entities is made via the server.

In this embodiment, the content includes legal content, contract content, research report content, and the like. In particular, the legal contents may be text, image, and audio information including legal contents, and may be an image file, an audio file, or a text file. In particular, due to the nature of legal content, it may be a mixed image in which an image and text are combined.

Conventionally, for example, a client (terminal) of station A requests a service from a service provider (terminal) of station A, and the service provider of station A requests a service from the service provider of station B. . When the legal content is intellectual property content, the service request information included in the client's request message includes the client's input for the request item required by the standardized template, and non-standardized input such as separate request items. , may include attachments such as video files or audio files.

Request items according to the standardized template here are, for example, "cost", "quality", "speed", "(service provider's) experience value", "(service provider's) scale", "intellectual right at the time of application Registration is necessary", etc. These request items are for defining the properties of the service related to the legal content cm that is the object of the service, and in this embodiment, they are called "object variables".

The client terminal of this embodiment may input service requirements through an app provided through a platform provided by a matching server between the client terminal and the provider terminal. On the screen (touch screen or monitor screen) of the client terminal, the above-described object variables are listed and displayed through the running app, and the user may be implemented to input object variable values for necessary object variables. In this embodiment, for example, the image file may be a logo image of a trademark or a drawing image of a patent. The voice file may be a file in which a request of the client is recorded by voice. A logo image or drawing image is material information that is substantially important in intellectual property content. This is because, in the case of intellectual property contents, determination of identity or similarity between contrasting contents is essential. Conventionally, separate automated processing of material information was not performed, and the determination of identity and similarity was performed through an inefficient method, so there was a problem in that processing takes a lot of time.

In the case of this embodiment, a physical connection through a platform between “the client terminal of station A and the server of the matching system” and “the server of the matching system between the client terminal and the provider terminal and the service provider terminal of multiple countries” It exists independently. The establishment of a network capable of communicating between the server, the client terminal, and the service provider terminals makes it possible to automatically and quickly respond to complex requests from clients. This is because, through this network structure, it is possible to reduce the amount of computation in the process of interpreting various requests from clients and responding to them.

The server 140 of the matching system between the client terminal and the provider terminal provides the content provided by the service provider terminal to the client terminal 110 through a matching platform 160 between the client terminal and the provider terminal. do. If necessary, the server 140 may convert legal content into a regular format, or may provide language-converted content after performing language conversion processing in another language. In addition, the server 140 matches the client 110 and the service provider 120 according to a predetermined algorithm for prioritization.

In addition, the server 140 stores history information including legal content from the service provider terminal in the history storage unit 150 . In this case, the history information related to the content stored in the history storage 150 is past history information, and may be a reference value that is preferentially considered in an operation for matching a client and a service provider. Considering that the server of the present embodiment manages legal content in which a continuous network is maintained, it is possible to reduce the amount of computation performed by the history information in the server.

The server 140 of the matching system between the client terminal and the provider terminal provides a matching platform between the client terminal and the provider terminal for performing a matching service between the client terminal and the provider terminal. In this embodiment, the platform is used as a concept of software that implements matching between the client terminal and the provider terminal. A platform may also be interpreted as including a hardware architecture or software framework (including an application program framework) on which such software runs. In addition, it will be apparent to those skilled in the art that the platform may include an architecture of a computer, an operating system (OS), a programming language, and related runtime libraries or GUIs.

The main process performed by the server 140 is to calculate a matching score between the client terminal 110 and the service provider terminal 120, and to arrange a network between the client terminal and the service provider terminal according to the matching score. . In particular, the server 140 calculates a matching score between the service request message of the client terminal and the response message from the service provider terminal.

The service provider terminal 120 does not communicate directly with the client terminal 110, but communicates with the client terminal 110 through the matching platform 160 between the client terminal and the provider terminal, provide necessary services. The service provider terminal is, for example, a computing device connected and managed by a law firm, a patent firm, or the like. The service provider terminal delivers content to be provided to the client terminal, for example, legal content in the form of an electronic document including legal opinions to the client terminal 110 through the server 140 .

According to another embodiment of the present invention, the client terminal 110 may request the response content for the same event from the plurality of service provider terminals 120 through the server 130 . Among a plurality of service providers subscribed to the matching platform 160 between the client terminal and the provider terminal, there may be a plurality of service providers satisfying the requirements requested by the client. In this case, the response message provided from the service provider terminal 120 will be very diverse. By utilizing these plurality of response messages, it is possible to provide a service with enhanced diversity to the client terminal side. In addition, when the language of the response content provided by the service provider as a result is different from the language used on the client terminal side, the server 140 translates the response content through a translation engine provided inside the system or an externally accessible translation engine. can be provided to the client terminal.

The server 140 managing the matching platform between the client terminal and the provider terminal stores information about the service provider in the history storage unit 150 . In addition, information on the service quality of the service provider may be further stored as history information. The service quality may be calculated based on, for example, compliance with procedural matters such as compliance with a delivery date, and an evaluation opinion received from a client terminal on the quality of content. The server 140 may manage information of service providers through an expert pool.

The client terminal 110 may access the matching platform through an application for the client terminal provided on the matching platform between the client terminal and the provider terminal, and may transmit a necessary service request message. The same is true for the service provider terminal 120 .

In this embodiment, the client terminal service provider terminal and the server are network entities constituting a network according to the present system, respectively. In the present embodiment, a matching score indicating the degree of matching between network entities will be determined by calculating the degree of matching between the entities.

However, in the present invention, the network entity to which the matching score is calculated does not need to be limited to the above-described client terminal, service provider terminal, and server. This is because, apart from the service provider terminal, a transaction for actually generating legal content can be performed through another sub-terminal connected through the service provider terminal and the network through the internal network.

In addition, the connection structure between network entities according to the present invention may be implemented such that the sub-terminal communicates with the client terminal through the server, and the sub-terminal can communicate with the service provider terminal as well as directly communicate with the server It can be implemented so that it does not. However, since service provider identification information may be given to the sub-terminal, history information including the sub-terminal profile may be managed separately. In this case, the complexity of the system will increase, but the burden of processing time according to the complexity of the system can be alleviated by introducing the learning unit proposed in the present invention.

2 is a block diagram illustrating a matching system between a client terminal and a provider terminal according to another embodiment of the present invention. The matching system between a client terminal and a provider terminal illustrated in FIG. 2 includes a client-side apparatus 1000 , a service provider terminal 2000 , and a server 3000 for matching between the client terminal and the provider terminal.

The client-side device 1000 includes a client terminal 1100 , an environment sensor 1200 , an I/O interface 1300 , and an image display unit 1400 .

The client terminal 1100 is set to communicate with the first interface 3110 of the communication interface 3100 of the server 3000 . For example, in order to obtain legal content, the client terminal 1100 generates a service request message and transmits the generated service request message to the first interface 3110 of the server 3000 .

Here, the service request message includes identification information for identifying a service subject or an object of the service, and a plurality of object variables for describing the requested legal content. Here, the legal content is, for example, content including legal opinions and information prepared in the form of an electronic file, such as a legal report, an appraisal, a prior research report, and an opinion. From a physical point of view, legal content is an electronic file containing text information, video information, audio information, or mixed images (when information corresponding to text is included in the video), and there are special restrictions on the type of legal content. there is no When the legal content is content related to intellectual property rights, an example of a service request message is illustrated in FIG. 8 . 8A shows an example of a screen for inputting identification information. For example, the identification information may also include optional information such as the client's name, contact information, the type of intellectual property rights (patent, trademark, design, etc.), the type of currency for payment, and whether to share location information. .

When the client terminal is a mobile terminal, the environment sensor 1200 obtains a GPS sensor for obtaining information of the mobile terminal, a three-axis sensor for obtaining information about movement in three dimensions, a pressure sensor, a temperature sensor, or biometric information There are sensors, etc. When the client terminal is a mobile terminal such as a smart phone, such an environment sensor may be built-in and implemented in the mobile terminal.

The I/O interface 1300 receives input information from the client and performs signal processing on output information to be provided to the client side. The I/O interface 1300 may include, for example, a camera for image acquisition, a microphone for voice acquisition, and a keyboard.

The image display unit 1400 displays, on the screen of the client terminal, an image of content and results related to matching between the client terminal and the provider terminal transmitted through the first interface terminal. In addition, the image display unit may be implemented in the form of a display panel further including a touch screen capable of receiving a touch input from a user.

The service provider terminal 2000 generates a "response message" as a response to the "service request message" from the service provider terminal and delivers it to the server. In addition, the service provider terminal 2000 generates final legal content as a result of the service and delivers it to the service provider terminal through the server.

The server 3000 performs a process for matching between the client terminal and the service provider terminal between the client terminal and the provider terminal. The server 3000 includes a communication interface 3100 , a processor 3200 , a memory 3300 , a learning unit 3400 , an image processing unit 3500 , and an output unit 3600 .

The server 3000 is implemented including hardware components such as the processor and memory and software executed on the hardware. The server performs a transaction related to matching between the client terminal and the provider terminal through a platform provided in the form of software on the hardware configuration. In this embodiment, the client terminal 1100 and the service provider terminal 2000 are indirectly connected through such a platform.

The communication interface 3100 includes a first interface 3110 and a second interface 3120 . The communication interface 3100 shares necessary information (text information, image information, voice information, etc.) through communication with a client terminal that is an external device and a service provider terminal. The first interface 3110 is an interface for communication between the client terminal 1100 and the server, and the second interface 3120 is an interface for communication between the service provider terminal 2000 and the server.

The processor 3200 performs an operation for a matching process between the client terminal and the provider terminal, and controls related devices. 3 is a block diagram illustrating the main functions of the processor 3200 as components. The processor of FIG. 3 includes an object variable separation unit 3210, a requirement feature value set generator 3220, and a response feature value set generator. 3230 , a correction variable determining unit 3240 , a spatial projection unit 3250 , and a matching score calculating unit 3260 . A detailed example of the matching process between the client terminal and the provider terminal performed by the detailed components is shown in FIG. 4 . It will be described in more detail with reference to FIG. 4 .

4 is a flowchart illustrating a matching method between a client terminal and a provider terminal according to an embodiment of the present invention. The method shown in FIG. 4 includes the following steps performed in time series in a server 3000 that performs a matching service between a client terminal and a provider terminal.

In step S100 , the processor 3200 receives a service request message for requesting a service related to provision of legal content from the client terminal 1100 through the communication interface 3100 . In this embodiment, the service request message is a message received to request a service related to service provider matching. An example of a detailed request message for such a service request message is shown in FIG. 8 .

The format of the service request message is not particularly limited, but includes identification information for specifying a subject or object related to legal content. The identification information is, for example, information for specifying content, such as whether the type of intellectual property right is a patent, a trademark, an infringement dispute, or an application business. The object variables are as exemplified above. In addition, the service request message further includes a plurality of object variables for expressing the attribute for the service request as a multidimensional space vector. The identification information may include information for identifying a subject related to legal content as shown in (a) of FIG. 8 . Object variables may exist within the scope of a preset item, but it is also possible for the client to add and input new object variables. In this case, learning through a second learning unit to be described later is required.

Object variables include "cost" related to the legal content, "quality" required in relation to the legal content, "speed" related to the delivery schedule of the result, and "experience value" indicating the performance and experience of the service provider ", "size" of the service provider organization, "necessity of registration" according to whether it is essential to acquire IPR when legal content is IPR-related content, "importance" related to core items, etc. In addition, as object variables, there are issues related to whether the service provider is an agent of a competitor, that is, a "complete issue" and a "special matter" that can be freely entered. In the present embodiment, the "specificity" belongs to an object variable in a broad sense, but for this, different learning is performed, and therefore, it is referred to as a "residual variable" for conceptual distinction.

As shown in (b) of FIG. 8 , object variable values are assigned to each object variable, which is mainly assigned with object variable values according to a selective input from a client terminal. For example, when a client requests a service for legal content, "cost" selects "lower than average cost" or "less than 3,000 USD", "quality" selects "middle level or higher", and , "Speed" can select object variable values as "faster than average processing time" or "within 1 month", respectively. In addition, in terms of “experience”, select “rich in related experience” or “more than 10 related business processing experiences”, and in relation to “scale”, select “a corporation with 100 to 200 people” and “necessity of registration” Regarding , "must be registered" is selected, and with respect to "importance", object variable values can be selected as "very important", etc., respectively. In addition, you can select "Strictly apply" for "complete issue", and for "special matters", "as it is a trademark application with litigation issues, registration is required, and multiple registrations through divisional applications are also available." A message such as "required" may be input. For convenience, in this embodiment, the object variable is described as including the object variable value.

The above-described object variables are variables that can be expressed quantitatively, such as relatively high and low, or some numerical value, and at least some of these object variables can be expressed as multidimensional space vectors in a multidimensional space. In the present invention, an object variable that can be expressed as a multidimensional space vector in a multidimensional space among object variables included in the service request message is defined as a "first group object variable". Among the object variables, variables that do not correspond to the first group object variables are defined as "first group residual variables". In the case of a response message, it can be defined as "second group object variable" and "second group residual variable" with the same meaning.

Here, the "first group residual variable" is difficult to express directly as a multidimensional space vector. "Specificity" may be input in the form of a word or a short sentence. They are also often related to the core of service requirements. As can be seen in the above example, it is not possible to cover all of the client's situation only with object variables provided through the template. In this example, "specific matters" include "relevance to litigation" and "requires a large number of registration certificates", etc. In a way, these are things that can be key points in this service request. In the present embodiment, the processor 3200 differently performs arithmetic processing on the "object variable" and the "residual variable", calculates a matching score using the result of each execution, and uses the matching score to provide a service to the client and service provider. Match providers.

In step S200 , the processor receives a response message as a response to the service request message from the service provider terminal 2000 through the second interface 3120 . Prior to generating the response message, all information of the service request message may be transmitted to the service provider terminal, or the process may be implemented by transmitting only some important information. The service provider may generate a response message by selecting and inputting items that the service provider can provide in response to the object variables set through the client terminal through the terminal, and may transmit the generated response message to the server.

Steps S300 to S600 show a method of calculating an index for whether a match between a client and a service provider is based on a service request message and a response message. These steps are steps performed by the processor 3200 or the learning unit 3400 of the server, and since detailed operation contents of each step are distinguished from each other, for convenience, detailed components are described as the subject of the operation.

Of course, in FIG. 3 , each detailed component of the processor 3200 may be mainly implemented in the form of software. The source codes included in the software are stored in the memory 330 and may be implemented so that detailed operations are executed by the processor. In addition, the above components are implemented in the form of a signal processing device having an arithmetic capability, and the processing device may be implemented in a form in which these signal processing devices are integrated.

In step S300, the object variable separation unit 3210 divides the service request message into first group object variables and first group residual variables, and divides the response message into second group object variables and second group residual variables. distinguish

In step S410 , the requirement feature value set generating unit 3220 generates a requirement feature value set for expressing the first group object variables as vectors on multidimensional spatial coordinates. Also, the response characteristic value set generating unit 3230 generates a response characteristic value set according to the second group object variables. Here, the multidimensional space may be a two-dimensional or three-dimensional or more multidimensional space. The multidimensional space may be implemented in one or a plurality of forms.

In S420 , the spatial projection unit 3250 generates a multidimensional spatial distribution image using a set of requirement feature values that can be expressed as vectors having coordinate values in a multidimensional space and a set of response feature values. Since the requirement feature value set can be expressed as an ordered pair of two or three or more object variables selected from the object variables, a plurality of such requirement feature value sets exist, and accordingly, a plurality of multidimensional spatial distribution images may exist. there is.

In S430 , the first interface 3110 transmits visualization information for displaying the generated multidimensional spatial distribution image on the screen of the client terminal. Here, the visualization information is information about the multidimensional spatial distribution image generated by the image processing unit 3500, and refers to information that can be visually expressed through a touch screen screen, a monitor, or the like.

In S440, the first interface 3110 receives the correction vector from the client terminal. In S450 , the correction variable determiner 3240 determines a correction variable using the received correction vector, and transmits the determined correction variable to the spatial projection unit 3250 . The spatial projection unit 3250 is configured to perform a multidimensional space according to the requirement feature value set corrected according to the correction variable determined by the correction variable determining unit 3240 and the response feature value set generated by the response feature value set generating unit 3230 . An image is generated and transmitted to the output unit 3600 . The output unit 3600 displays the multidimensional space image through an output device such as a monitor.

The matching score calculation unit 3260 may calculate a matching score by correcting the requirement feature value set according to the correction variable according to S450 and comparing it with the response feature value set. Also, the matching score calculating unit 3260 may A matching score may be calculated by further considering a degree of similarity between the first image and the second image reconstructed through S520 or a fully connected feature value extracted from the images. The above-described S410 to S450 will be described in more detail with reference to FIGS. 5 and 6 below.

5 shows an example of a multi-dimensional spatial distribution image and a set of requirement feature values that can be displayed on the screen of the client terminal. In the example of FIG. 5 , the requirement feature value set includes three different first group selection object variables selected from the first group object variables, and the response feature value set is provided to two different service providers. and second group selection object variables for In this example, the types of the first group selection object variable and the second group selection object variable are the same. Hereinafter, it will be described in more detail focusing on the set of requirement feature values.

In the present embodiment, the requirement feature value set includes the coordinates and attribute values of the requirement feature point for specifying the positions of the first group selection object variables in the multidimensional space. Here, the attribute value further includes a position vector and inverse phase information for the matchable region. Here, the position vector for the matchable area is for specifying the existence direction of the matchable area. The attribute value is information on the setting of the matchable region further including reverse image information. The attribute value indicates for each object variable whether the direction is opposite to a predetermined reference direction or the same direction as the reference direction. Calculating the matching score may be calculated using a distance between the set of requirement feature values according to the first group selection objects and the response feature value sets according to the second group selection objects.

5 shows an example of a multidimensional spatial distribution image that can be displayed on a screen of a client terminal. The example of FIG. 5 is an example in which quality, cost, and speed are selected as object variables among the first group object variables, and object variable values according to the selected object variable are expressed in a three-dimensional space. In this embodiment, the object variables thus selected are referred to as first group selection object variables, and a coordinate expression such as (quality, cost, speed) or a spatial vector expression is defined as a "requirement feature value set". The coordinate axis of the three-dimensional space consists of the coordinate axis with cost, speed, and quality as object variables. In the example of FIG. 5 , it can be seen that the client has input the maximum cost value for the cost. In the case of quality, a minimum quality value is input, and a higher quality is requested. In the case of speed, since the maximum speed value is entered, it is included in the request message that there is no need to rush it. Normally, the speed is selected through the minimum value setting, but in this case, the maximum value is set for the speed. This is because quality is more important in this case, so the range of allowable speed is set in the reverse direction from the standard of the general case (the reference value of the speed object variable).

The requirement feature value can be defined as a multidimensional space vector through the actual value of an object variable, or a relative value (high, medium, low) or a relative ratio with a reference value (average related to the task). For example, in consideration of a relative ratio to the maximum scale that can be displayed on the screen of the client terminal, it may be expressed in spatial coordinates, and of course, there is no particular limitation to this expression method.

In FIG. 5 , the first service provider meets the service requirements for speed and cost but does not meet the requirements for quality, and the second service provider meets the service requirements for speed and quality, Cost requirements are not met.

In the multidimensional space image of FIG. 5 displayed on the client terminal, it is possible to correct the object variable values initially input through service requirements through touch input or mouse input. In the example of FIG. 5 , the client may further adjust the first adjuster (maximum cost value) through a screen manipulation for a touch screen or the like. In particular, when a touch manipulation is input in the direction of increasing the cost value on the coordinate axis of “cost”, it may be defined in the form of a position vector, and this position vector (correction vector) can be easily detected through the touch panel. . According to the correction vector, in particular, in consideration of the direction and magnitude (the length of the touch input) that the correction vector faces, the correction variable determiner may calculate an adjustment value (correction variable) of the set value of the first group selection object variable. The client may make an adjustment to increase the first coordinator (maximum cost value) above the cost value suggested by the second service provider.

The requirement feature point (three-dimensional coordinates consisting of “cost, speed, and quality”) defined by each object variable included in the requirement feature value set of FIG. 5 is an area that satisfies the requirements requested by the client This information is essential for definition. By adjusting the correction vector, the coordinates of the feature point of the requirement are also changed, and it is possible to adjust so that more service providers can be included in the corresponding area. A requirement feature point is defined as a "matching coordinator" in this embodiment. As shown in FIG. 5 , the matching coordinator may be displayed on a client screen in a multidimensional space.

The "requirement feature value set" of the present invention, exemplified through FIG. 5 , includes coordinates of a matching coordinator (requirement feature point), and attribute values. The attribute value is information for indicating an attribute for such a requirement or a matchable region, and may include a position vector and reverse phase information. As shown in the lower right of FIG. 5 , the position vector is a vector for determining the direction in which the matchable area exists. The reverse phase information is information for identifying a variable determined as an opposite region unlike a universal standard. In the case of FIG. 5 , in the case of speed, the maximum value is set instead of the minimum value. Therefore, a general case can be expressed as "1", and the opposite case can be expressed as "-1". That is, in the example of FIG. 5 , the reverse phase information may be expressed as (1, -1, 1).

In FIG. 5 , each of the selection object variables, such as cost, speed, and quality, may be defined in the form of a scalar value according to a normalized value or a relative value with a predetermined reference value (average value, etc.). When the ratios when compared with the average value are the first adjuster (cost), second adjuster (speed), and third adjuster (quality) values, according to the example of FIG. 5 , it can be exemplified as (1.2, 1, 1.5) above. there is. According to the coordinates of the feature points above, the cost can be paid about 20% more than the average, the speed can be up to the average processing speed of the task, and the quality is 50% higher than the average quality because the event is important means to be high. The shaded area in FIG. 5 is a matchable area, and if a feature point of a service provider is within the corresponding area, the corresponding service provider may be a matching candidate. When there are a plurality of matching candidates, in general, the distance between the requirement feature point and the service provider feature point may be a valid value for defining the matching score. In this example, the requirement feature value set can be expressed as (1.2, 1, 1.5) (-1.2, -1, -1.5), (1, -1, 1). An object variable including a negative number in the reverse image information may be excluded from calculating the distance between the feature points. In this case, it is preferable to orthogonally project the values of “cost” and “quality” on the plane of the “speed” object variable, and calculate the matching degree of matching through calculating the distance between the projected feature points.

The example of FIG. 6 is an example of a multidimensional spatial distribution image in which scale, cost, and experience are selected as the first group selection variables from among the first group object variables. According to FIG. 6 , the client has already specified the minimum scale, the maximum cost value, and the minimum experience value through the service request message. In the example of FIG. 6 , in the case of the first service provider, the cost and experience values satisfy the service request requirements, but the scale does not satisfy them. In the case of the second service provider, the scale and cost meet the service requirements, but the experience does not. Even in this case, it is possible to modify one of the above first group selection object variables through manipulation (correction vector) on the screen of the client terminal.

4 , steps S510 and S520 show preprocessing for calculating a matching score using residual variables. Since the above process is performed by the learning unit 3400, the present steps will be described with reference to the detailed components of the learning unit shown in FIG. 7 .

The learning unit 3400 is partially similar to the form of a convolutional neural network, but in that the first learner in the form of a convolutional neural network and the second learner in the form of a generative adversarial learner are combined, It can be distinguished from a conventional neural network. In this embodiment, the learning unit 3400 may be implemented in the form of software, or may be implemented in the form of a processor capable of image processing and executing software for learning. The learning unit serves to generate a feature value in which the dimension of the input value is reduced, but the feature of the input value is well reflected. The learning unit 3400 may be implemented as a processor separate from the processor 3200 , or may be implemented through a GPU for image processing. When the learning unit 3400 is implemented through a separate processor, the meaning of the metric transformation unit, the convolution filter, and the activation unit in this embodiment can be understood as a divided concept for sequentially expressing the operations of the processor. Of course, as another embodiment of the present invention, it will be apparent to those skilled in the art that it is possible to implement each of the convolution filter, the activation unit, and the complete connection unit as a digital signal processing device capable of arithmetic operation.

The learning unit 3400 includes a first learning unit 3410 , a second learning unit 3420 , and a summary unit 3430 . The first learning unit 3410 includes a metric transformation unit 3411 , a convolution filter 3412 , an activation unit 3413 , and a full connection unit 3414 . The convolution filter and the activator may be repeatedly arranged. Meanwhile, the second learner 3420 may include a noise generator 3421 , a summation unit 3422 , a generator 3423 , a training sample 3424 , a selector 3425 , and a classifier 3426 . .

The summary unit 3430 generates first and second summary messages by distinguishing a meaningful variable from a variable with less meaning from the first and second group residual variables. For example, in the first group residual variable, the residual variable may be a "special item", and the residual variable value may be a matter described in the "special item". If it is written in the special item, "Because it is a trademark application with a litigation issue, registration is required, and multiple registrations through divisional application are also required" You can summarize the residual variables as ". The summary message is preferably obtained in the form of a metric so that it can be processed by the neural network.

The second learning unit 3420 may be implemented as a separate processor operating as a generative adversarial network (GAN), or may be implemented in the form of an application program executed in an existing processor. The second learning unit 3420 receives the metric from the summary message, synthesizes the image, and transmits the synthesized image to the metric converter through the classifier 3426 .

The noise generator 3421 generates random noise. The summing unit Adder 3422 sums the generated random noise and the summary message. The summed feature values are transferred to the constructor 3423 . The generator 3423, whose generative adversarial learning has been completed through prior learning, outputs a newly reconstructed image by inputting the summed feature values. The reconstructed image generated by the creator is derived from text information and cannot be viewed as a real image in nature. However, since text information and the reconstructed image have a unique matching relationship, and the similarity between images is related to the similarity of the context included in the text, the reconstructed image is a context image. When a matching score is calculated using such a contextual image, it is possible to match a service provider that reflects the requirements expressed in text.

The selector 3425 selects one of the training sample 3424 or the synthesized image (in the learning process) from the generator and delivers it to the classifier.

The classifier 3426 has a filter coefficient specified through learning to discriminate whether the input image is a real image or a fake image, and through this, discriminates whether the actually input image is a real image or a fake image. In the learning process, the generator 3423 and the discriminator 3426 first update each filter coefficient so that the image reconstructed by the generator 3423 can be recognized as a real image. Following the first update, the classifier 3426 receives a training sample, recognizes the training sample as a real image, and sets its filter coefficient to 2 so that the synthesized image generated by the generator 3423 can be recognized as a fake image. update by car. Following the secondary update, the generator 3423 updates its filter coefficients tertiarily so that the images generated by it can be determined as real images by the classifier. This update process is iteratively performed for various training samples. It will be apparent to those skilled in the art that the process of updating the filter coefficients of the generator and the classifier is not limited to the above example and can be variously modified.

The learner 3426 may distinguish the real image into a first real image and a second real image. Since the image generated by the creator is an image reconstructed from text information, it is difficult to view such an image as an image identical to an actual image. However, through comparison of the feature value calculated through the discriminator, the first reference value, and the second reference value, the discriminator has a first real image with a low relative probability and a second with a high relative probability in terms of a probability that the discriminator is similar to the real image. Like the real image, the classifier can distinguish the reconstructed first image and the second image in detail.

The metric converter 3411 (Matrix Convertor) receives the first and second images reconstructed according to the first group residual variables and the second group residual variables from the classifier 3426, and uses them to convolution It is converted into a matrix suitable for applying a neural network. The first metric and the second metric transformed through the metric converter are image signals that can be processed by the convolution filter. The metric converter 3411 may include removing unnecessary noise included in the image, normalizing the image, or adjusting the image resolution before applying the convolution filter to the image. Of course, it is also possible to directly apply the convolutional neural network to the input image.

The convolution filter 3412 moves the input image up, down, left, and right to continuously perform a weighted sum operation. The result obtained through the convolution filter is in the form of a reduced-dimensional image (also referred to as a feature value), and this image is an image obtained to well represent the local properties of some of the properties of the input image. Coefficients of a convolution filter specify the attributes of the filter. Since the present embodiment is implemented to calculate a feature value using an artificial neural network that has already been trained, the filter coefficient of the convolution filter 3412 may be implemented as a predetermined value.

The activation unit 3413 applies activation functions such as ReLU, Sigmoid, Hard limit, TanH, etc. to the image that has passed through the convolution filter 3413 . An additionally applied activation unit operates in the same manner. The output by applying such an activation function is a value obtained by simplifying the input value or adjusting the input value to a value within a specific range.

In the present embodiment, the first learning unit 3410 has a structure in which a plurality of pairs of such a convolution filter and an activator are arranged. The output value obtained while passing through such an iterative arrangement is obtained so that the dimension of the first metric (information in the form of an image) is reduced and local features are also well represented. The feature value obtained from the activation unit located at the rearmost position is finally input to the complete connection unit 3414 . The fully connected unit 3414 performs an operation with a predetermined connection weight on the feature values input from the activator, and outputs first and second fully connected features obtained by connecting the operation results as a whole.

The configuration of the learning unit combining the first learning unit and the second learning unit presented in this embodiment is to reconstruct residual variables such as singularities into an image, and to extract feature values again from the reconstructed image. Of course, semantic interpretation of residual variables such as singularities is possible in a traditional way. However, if the matching score between the service request message and the response message is calculated through the image reconstructed from the text as in this embodiment, the system stability can be improved even for various and unusual request messages that are difficult to grasp through semantic interpretation. Matching that can satisfy the needs of the client is possible while securing it.

In step S600 , the matching score calculator 3260 of the processor 3200 displays the first and second images reconstructed according to the comparison result between the first and second object variables and the first residual variable and the second residual variable. A matching score is calculated using the comparison result between the two groups.

In step S700, the first interface transmits the matching score to the client terminal as a matching result. In step S800, the second interface similarly transmits the matching result to the service provider terminal.

Hereinafter, a detailed method of calculating the matching score will be described in more detail. The matching score may be a single one, or may include a plurality of first matching scores and second matching scores. The first matching score is calculated according to the similarity between the requirement feature value set and the response feature value set, and the second matching score is calculated according to the similarity between the reconstructed first image and the second image. The final matching score is a result of fusion of the first matching score and the second matching score, and for example, the final matching score can be calculated through a sum operation according to a predetermined weight.

To calculate the first matching score, the matching score calculation unit 3260 of the processor determines whether the response feature value set is included in the request area defined by the requirement feature value set, and the service provider included in the request area. In the case of , the distance between the requirement feature point and the response feature point can be calculated. In this case, the larger the distance value, the larger the matching score, and the smaller the distance, the smaller the matching score. In the case of a service provider that is not included in the requested area, the distance according to the calculation result can be expressed as a negative value. In the case of a service provider located outside such a requested area, a service provider with a large distance value (absolute value is small) may have a higher matching score than a service provider with a small distance value (absolute value large). there is.

To calculate the second matching score, the processor may calculate a similarity between the first fully connected feature value and the second fully connected feature value, which are the result values from the fully connected unit 3414 . For example, a service provider having a second fully connected feature value most similar to the first fully connected feature value may be determined through an average operation of the sum of squares of differences between feature values for each corresponding element.

Of course, the results (matching results) obtained by calculating the first matching score and the second matching score may be different, and the processor performs a final A matching score can be calculated. Here, the weight may be a predetermined value or may be a weight that is adaptively calculated according to a real-time situation. For example, when the image reconstructed by the generator belongs to the level of the first real image, it is preferable to lower the weight of the second matching score. Conversely, when the reconstructed image belongs to the level of the second real image, it is preferable to increase the weight of the second matching score.

In the present embodiment, there is no particular limitation on the method in which the processor calculates the first and second matching scores, and a person skilled in the art may define and use similarity values in various ways in consideration of specifications required in the system. For example, through vector analysis of the first fully connected feature value and the second fully connected feature value, the similarity value may be defined using the cosine distance, the Euclidean distance, and the Mahalanobis distance.

In calculating the matching score between the client terminal and the service provider terminal, the processor 3200 may calculate the matching score by considering similarity or relevance between various variables differently according to priorities. In particular, the processor 3200 may calculate the final matching score using the above-described first matching score and the second matching score through the following equation. Further extensions to the third matching score are possible.

[Equation]

Matching Score = A*1 _st Matching Score + B*2 _nd Matching Score

Here, the Matching Score is a score indicating the degree of matching between the client terminal and the service provider. A and B are set weights for the first matching score and the second matching score, respectively.

Preferably, the processor may first perform normalization on the first and second matching scores, and calculate the matching score by substituting the normalized similarity value into the above equation. This is because, in some cases, a deviation between similarity values may be large depending on the detailed situation of the system. The setting weight can be set by the client through a detailed request message. The third matching score may be a value based on absolute priority. The third matching score is used as a cutoff means for selecting service provider candidates, and may be implemented to be excluded from calculation of the final matching score.

For example, when "response speed" is included in the object variable, the cross-correlation between the request time series analysis waveform for the time when the service request message is generated and the response time series analysis waveform for the response time of the service provider is considered very important can be The average time delay value between when a response message is generated and when a request message is generated is a measure of how quickly a response is made. In addition, a large cross-correlation value indicates that the patterns between the two analysis waveforms are similar, and can be a measure to indicate that the service provider's response pattern is continuous and steady. Considering the average time delay value and the cross-correlation value, it is possible to define a third matching score that reflects service speed/stability.

In the present invention, it will be apparent to those skilled in the art that the matching score is not limited to being calculated using the above-described first to third matching scores.

In step S800, the processor 3200 provides a matching result value between the client terminal and the provider terminal. The matching result value between the client terminal and the provider terminal is, for example, a matching score and a service provider ranking accordingly. The processor 3200 may transmit information on the service provider ranking to the client terminal through a communication interface. In addition, the processor 3200 may transmit the service provider ranking information to the image processing unit 3500 , and may provide or directly display the image generated by the image processing unit 500 to the outside through the output unit.

According to another embodiment of the present invention, the processor 3200 generates an input request item message necessary for service provider matching required for the client in consideration of the semantic interpretation result, and uses the message to generate an input request item message through the communication interface 3100 to the client terminal The step of transmitting to 1100 may be further performed. Here, the input request item message is, for example, when the content to be provided is a content related to a trademark search, the necessary items are due date, mark, goods/service, target country (target country), priority (priority), conflict (conflict), may include at least one or more of the budget (budget). Here, the priority is information indicating the weight of important matters from the client side in the service provider matching. For example, depending on the client's circumstances, in some cases the budget may be the most important, and in other cases meeting the deadlines may be the most important. Priority is the information needed to define a weight for the importance of these necessary items. The priority is selected through the client terminal, and as described above, the priority may be included in the detailed request message input by the client. In addition, factor weights such as cost, speed, history, and scale of the service provider expert may be preset through the client terminal. In this case, the server may recommend the service provider or the expert to the client by aggregating expert information of the service provider managed through the platform.

Next, the processor 3200 receives a detailed request message that is a response to the input request item message from the client terminal 1100 . The specified request message may be implemented in the form of a message in which information necessary for the input request item message is written. Information for generating the detailed request message is input through the I/O interface 1300 of the client device 1000 . The I/O interface 1300 includes, for example, an input device such as a camera, a microphone, and a keyboard. More specifically, when the content to be provided is the content for the trademark research, the logo image for the corresponding trademark may be an image acquired through a camera or already acquired and stored in a memory. The logo image may be transmitted to the processor 3200 through the communication interface 3100 . In addition, it may be implemented to receive a voice signal through a microphone and transmit the voice signal to the processor 3200 .

In this embodiment, the detailed request message may be divided into request material information and request process information according to the type or property of the operation performed by the processor 3200 . The requested material information is related to the actual content of the content, and may be, for example, a logo image in the case of trademark research content. As another example, in the case of content for a patent search, it may be a drawing image of a specific structure of the invention. Request process information includes procedural matters, not practical matters, such as due date, budget, or proximity (when the location information of the service provider terminal is physically close to the location information of the client terminal) related information.

The processor 3200 transmits a matching request message to the service provider. The matching request message is a message obtained as a result of semantic analysis of the detailed request message and includes important matters from the point of view of the service provider. Here, the matching request message may be the same as the detailed request message, but preferably a message converted into the language of the service provider or a message including a summary of process information and material information (excluding confidential information). there is.

Next, the processor 3200 transmits at least a portion of the material information and the process information to the learning unit, and obtains a service request feature value determined by the learning unit. In addition, the processor 3200 transmits a response message according to the matching request message to the learning unit, and obtains a response message characteristic value determined by the learning unit.

Next, the processor 3200 calculates a matching score in consideration of the relation between the service request characteristic value and the response message characteristic value. There is no particular limitation on the method in which the processor 3200 finds a matching score. Preferably, the processor 3200 performs an operation on the sum of the squares of errors between the service request characteristic value and the service response characteristic value, or a matching score between the service request characteristic value and the high response message characteristic value through an operation on the correlation. can be calculated.

Here, an example of relevance is the degree of similarity between reconstructed images when a message is reconstructed into an image. A client that requires a quick response and a service provider with quick response characteristics are highly related to each other. For this correlation calculation, the server's processor correlates the distribution curve (x-axis is the time axis, y-axis is the number of events) for the request request request characteristics of the client terminal and the distribution curve to the response of the service provider terminal Judgment on the relationship can determine whether the relationship is high or low. In particular, the cross correlation between the contrasting distribution curves can be used as a reference value for the correlation. The client profile and the service provider profile may further include a distribution curve representing a behavioral pattern in terms of time as their unique characteristic value, and the processor further calculates the cross-correlation value through comparison between the distribution curves included in the profile. can be calculated easily.

Referring back to FIG. 2 , the memory 3300 stores data transmitted through the communication interface or data transmitted from the processor. There is no particular limitation on the form and type of memory, but cache memory or flash memory is preferable in terms of operation speed. Although not separately shown in FIG. 2 , a history storage unit for storing history information related to matching between the client terminal and the provider terminal is provided, but the server stores a large amount of cumulative history information separately from the memory 3300. can be provided

The image processor 3500 generates a visual image from the operation result of the processor 3200 . The image generated by the image processing unit 3500 may be displayed on a user interface (display screen) through the output unit 600 . Also, the output unit 600 may provide the image generated by the image processing unit to a separately provided monitor device, and reproduce the image through the monitor.

According to another embodiment of the present invention, the detailed request message transmitted from the client terminal may further include a client profile. The client profile may further include identification information for identifying the client, preference information, and characteristic information. The preference information may include, for example, priority information that the client has set when matching a provider. The characteristic information is unique information generated by the client. For example, there is information on how many requests were made to the server during the entire process of generating one legal content.

Like the client profile, the service response message transmitted from the service provider may further include a service provider profile for the service provider. Similarly, the service provider profile may further include characteristic information. The characteristic information may include, for example, response characteristic information of a service provider terminal when a request is generated from a client. In the case of a certain service provider, it may have a quick response characteristic, and on the contrary, it may have a delayed response characteristic of always responding late. The server may identify response characteristics of each service provider terminal through analysis of history information on the service provider. The response characteristic may be managed in the form of a response characteristic distribution curve, the response characteristic distribution curve is stored as history information, and the "characteristic information" is a summary of the response characteristic distribution curve. In this embodiment, the metric converter of the learning unit 3400 may normalize the response characteristic distribution curve and extract or summarize characteristic information through learning of the normalized image.

8C is a reference diagram illustrating an example of a screen providing a matching result according to an embodiment of the present invention. FIG. 8D is a reference diagram illustrating an example of a screen showing a service provider ranking according to a matching result. In (c) of FIG. 8 , it is indicated by a black dot, and a point corresponding to the center of the circle indicates a point where the client terminal is located. The small circular dots indicate the locations of service provider terminals on the map. Although the present invention is proposed as a network system for providing legal content, an offline process may be further required separately from the client. When the present invention is operated in conjunction with a location-based service, it is possible to derive a result that is more consistent with a client's request.

9 illustrates response content according to an embodiment of the present invention. More specifically, the response content of FIG. 9 is a case in which legal opinions are typified and expressed. The template of such typed response content is pre-formulated and provided by the server.

In the example of FIG. 9 , matters related to the probability, possibility, and right to apply the factor 1 are displayed. In addition, legal conclusions regarding the viability of the litigation case, the possibility of registering an intellectual property right application, etc. may be further included. Also, the server 140 may further perform a translation process to enable compatibility between different languages. Through the translation engine provided inside the matching system between the client terminal and the provider terminal or the translation engine accessible from the outside, the response content written in language A is translated into language B, language C, and language D, and the translated language is provided to the client terminal can do.

10 illustrates response content according to another embodiment of the present invention. The response content of FIG. 10 includes considerations for the identification of the trademark AlfredQ, prior trademark, registration possibility, and the like. Such response content is generated by the service provider terminal through the platform and provided to the client terminal as a result.

11 is a flowchart illustrating a matching method between a security-enhanced client terminal and a provider terminal according to another embodiment of the present invention. The matching method between a client terminal and a provider terminal illustrated in FIG. 11 includes the following steps performed in time series in a server of a matching system between the client terminal and the provider terminal.

In S1100 , the processor 3200 calculates a service request characteristic value and a service response characteristic value from the received service request message and response message. More specifically, the processor 3200 calculates a first fully connected feature value as a service request feature value and a second response feature value as a response feature value for a service response. Prior to step S100 , the processor receives the detailed request message from the client terminal 1000 through the communication interface and receives the response message from the service provider terminal 2000 as described above.

In S1200, the processor 3200 generates a session key for encryption in communication for each network entity, and shares the session key table with the network entity. In the case of the matching system between the client terminal and the provider terminal according to the present embodiment, a continuous network between entities is maintained, and the characteristic value of communication varies over time according to the situation of each entity (by terminal device).

The change in the network situation for each entity detected over time is a unique value that cannot be known by a third party. If this value (network situation variable for each entity) is used as a variable for updating the session key table, security can be strengthened. there is. An example of the network condition variable for each entity includes network traffic information or response speed information of a service provider. The network traffic information of the service provider is related to the absolute amount of process performed through the service provider as the amount of traffic detected through the platform. In addition, the response speed may be defined as a relative ratio between the time difference between when a query is sent and when a response is reached, and the amount of transmitted data. Since these individual network situation variables can be shared between two entities and a third entity cannot be shared, using these individual network situation variables can improve the security of the matching system between the client terminal and the provider terminal. can be strengthened For example, the fact that the service provider terminal cannot transmit a message between the client terminal and the server is a clearly distinguishable difference compared to the existing one.

In S1300, the processor 3300 detects a situation in which the process of the matching system between the client terminal and the provider terminal needs to be stopped, and when such a situation is detected, the inter-object network is stopped. Contents provided through the service provider and information shared between entities may all require credentials. You can temporarily stop the service.

In S1400, the processor 3400 changes the session key table value, and shares the changed session key table value for each entity. Although the necessity of a continuous process still exists, since the information leakage of the past is suspected, the processor 3400 substantially resets the session key table.

In S1500, the processor 3400 detects whether or not the reason for stopping the network is released, and if the reason for stopping is released, restarts the session at S1600. Even if the session key table is reset in S1400, the session is not restarted immediately. After confirming the cancellation of the reason for network suspension, the session is resumed after sharing the changed session key with each entity.

Network entities constituting the matching system between the client terminal and the provider terminal of the present invention are a client terminal, a service provider, and a server. "Client terminal and server" and "service provider terminal and server" are directly connected, but "client terminal" and "service provider terminal" are not directly connected. The session key of FIG. 9 is divided into a first session key between "client terminal and server" and a second session key between "service provider terminal and server".

12 shows a frame of a message for message integrity in the message transmission of the matching system between the client terminal and the provider terminal according to the present invention. In the matching system between the client terminal and the provider terminal according to the present invention, confidentiality of the provided legal content is required. In such legal content, since the content of the original should not be changed by a third party who has no obligation of confidentiality, it is very important to ensure the integrity of the original.

In FIG. 12 , the first message 4100 is, for example, a “service request message”. Substantial data of the service request message is included in the first data 4102 field. The first address 4104 is information indicating the location of the first integrity data in the first message 4100 . Also, the first address may be information indicating the length of the first integrity data or identification information. The processor 3400 may generate the first integrity data 4106 using the first data and the 1-1 session key. Although there is no particular limitation on a method of generating the integrity data, the processor 3400 may generate the integrity data using a hash function. In particular, it is possible to summarize the first data into data of a predetermined number of digits and generate integrity data through an XOR operation with the 1-1 session key.

In FIG. 12 , the second message 4200 has a structure in which the first message further includes the second data cumulatively. The second message 4200 includes an existing first message 4100 , second data 4202 , a second address 4204 , and second integrity data 4206 . The second integrity data may be a hash value using at least the first integrity data, the second data, and the 1-2 session key as inputs.

In this embodiment, the second message may be an input request item message transmitted by the server to the client terminal. The third message (not shown) likewise further includes third data, a third address and third integrity data in the second message. For example, the third message is a message transmitted from the client terminal to the server, and may be a detailed request message. Through such a cumulative message structure and a structure in which integrity data is connected in a chain, the integrity of the message can be guaranteed. This is very important in the distribution of legal content that requires continuity and confidentiality.

Even if all the components constituting the embodiment of the present invention described above are described as being combined or operated in combination, the present invention is not necessarily limited to this embodiment. That is, within the scope of the object of the present invention, all the components may operate by selectively combining one or more.

In addition, all of the components may be implemented as one independent hardware, but a part or all of each component is selectively combined to perform some or all of the combined functions in one or a plurality of hardware program modules It may be implemented as a computer program having In addition, such a computer program is stored in a computer readable media such as a USB memory, a CD disk, a flash memory, etc., read and executed by a computer, thereby implementing the embodiment of the present invention. The recording medium of the computer program may include a magnetic recording medium, an optical recording medium, and the like.

The above description is merely illustrative of the technical idea of the present invention, and those of ordinary skill in the art to which the present invention pertains may make various modifications, changes and substitutions within the scope without departing from the essential characteristics of the present invention. will be. Accordingly, the embodiments disclosed in the present invention and the accompanying drawings are for explaining, not limiting, the technical spirit of the present invention, and the scope of the technical spirit of the present invention is not limited by these embodiments and the accompanying drawings . The protection scope of the present invention should be construed by the following claims, and all technical ideas within the scope equivalent thereto should be construed as being included in the scope of the present invention.

110: client terminal 120: service provider terminal
130: communication network
1200: environmental sensor
1400: image display unit 3110: first interface
3200: Processor
3300: memory 3400: learning unit
3500: image processing unit 3600: output unit

Claims (11)

A matching method between a client terminal and a provider terminal performed by a server that performs a process for matching between a client terminal and a provider terminal requesting provision of legal content, the method comprising:
The communication interface receives, from the client terminal, identification information for specifying the legal content to be requested, and a service request message including object variables indicating attributes related to the service request, and corresponds to a response to the service request message. receiving a response message from the provider terminal;
The server includes a first group of object variables including a plurality of object variables that can be expressed as a multidimensional space vector in a multidimensional space among object variables included in the service request message, and among the object variables included in the service request message separating a first group residual variable including a plurality of object variables that do not belong to the first group object variables; and
calculating a matching score for matching between network entities using the first group object variables or the first group residual variable;
The server further includes a second learning unit generating a first image reconstructed by inputting the first group residual variable as an input, and a first learning unit generating a fully connected feature value from the first image,
The server calculates the matching score,
Second group object variables including a plurality of object variables that can be expressed as multidimensional space vectors in a multidimensional space among object variables included in the response message, or the second group object among object variables included in the response message The matching score is calculated by further using a second group residual variable including a plurality of object variables that do not belong to the variables, and is generated from the first image generated from the first group residual variable and the second group residual variable using the second similarity value between the second images,
The first group residual variable includes at least some textual information,
The object variables, characterized in that it includes at least two selected from the group consisting of cost, quality, speed, experience value of a service provider, and scale of a service provider related to the provision of legal content, for providing legal content A matching method between a client terminal and a provider terminal. delete delete The method of claim 1, wherein calculating the matching score comprises:
Matching between the network entities is performed using a first similarity value between the first group object variables and the second group object variables, or a second similarity value between the first group residual variable and the second group residual variable. A matching method between a client terminal and a provider terminal for providing legal content, characterized in that for calculating a matching score for The method of claim 1, wherein the second learning unit generates the first image,
obtaining a first summary message summarizing the first group residual variable; and
A matching method between a client terminal and a provider terminal for providing legal content, characterized in that it further comprises the step of generating a first image reconstructed from the first summary message. delete The method of claim 1, wherein the step of calculating the matching score by the server comprises:
generating a requirement feature value set for expressing the first group object variables as vectors on multidimensional spatial coordinates and a response feature value set for expressing the second group object variables as vectors on the multidimensional spatial coordinates;
Visualizing and expressing a multidimensional spatial distribution image according to the requirement feature value set and the response feature value set on the screen of the client terminal;
determining a correction variable for correction of a first group object variable in the multidimensional spatial distribution image visualized and expressed on the client terminal; and
The method further comprises correcting the first group object variable in consideration of the correction variable, wherein the server calculates a matching score using a first similarity value between the first group object variables and the second group object variables. A matching method between a client terminal and a provider terminal for providing legal content, characterized in that the calculation. 8. The method of claim 7,
The requirement feature value set includes two or three different first group selection object variables selected from the first group object variables, and the response feature value set includes: Including two or three different second group selection object variables to be selected,
Visualizing and expressing on the screen of the client terminal includes expressing the first group selection object variables and the second group selection object variables as images projected on a multidimensional space,
The server determines the correction variable by receiving a correction vector according to a user touch input on the screen for correcting the positions of the first group selection object variables that are visualized and displayed on the screen of the client terminal, A matching method between a client terminal and a provider terminal for providing legal content, characterized in that the correction variable is determined using the correction vector. 8. The method of claim 7,
The requirement feature value set includes two or three different first group selection object variables selected from the first group object variables, and the response feature value set includes: Including two or three different second group selection object variables to be selected,
The requirement feature value set includes coordinates and attribute values of the requirement feature point for specifying positions of the first group selection object variables in a multidimensional space;
The attribute value further includes a position vector for indicating the attribute for the matchable region, and reverse phase information for setting the matchable region,
Calculating the matching score is a matching method between a client terminal and a provider terminal for providing legal content, characterized in that the calculation is performed using a distance between the first group selection objects and the second group selection objects . According to claim 1,
The second learning unit obtains a first summary message summarizing the first group residual variable, adds the first summary message and random noise to generate a first summed feature value, and uses the first summed feature value to reconstruct the first image, and determine whether the reconstructed first image is a real image or a fake image,
The first learning unit applies a convolution filter to the reconstructed first image determined as a real image by the second learning unit, and uses the activated feature value obtained by activating the feature value according to the convolution filter. 1 Generate fully connected feature values,
The server calculates the matching score,
the first fully connected feature value obtained from the reconstructed first image;
A matching method between a client terminal and a provider terminal for providing legal content, characterized in that the calculation is performed using a second similarity value between second fully connected feature values obtained from the reconstructed second image. A computer program stored in a computer-readable recording medium in order to execute the matching method between the client terminal and the provider terminal according to any one of claims 1, 4, 5, and 7 to 10 on the computer.

Images