KR20230136449A - Legal assistance system using artifical intelligence and method of operation of the system - Google Patents

Abstract

A legal support system using artificial intelligence and a method of operating the system are provided. The method of operating a legal support system using artificial intelligence includes receiving a legal question including natural language text from a questioner, analyzing the content of the legal question, and matching a plurality of vocabularies included in the natural language text to a plurality of legal categories. generating a temporary answer from a pre-stored legal database based on the correspondence results of the plurality of vocabularies, providing the temporary answer as an answer subject to expert review, receiving expert review results for the answer subject to expert review. generating an expert answer from the provisional answer based on the expert review results, and providing the expert answer to the questioner.

Description

Legal support system and method of operation using artificial intelligence {LEGAL ASSISTANCE SYSTEM USING ARTIFICAL INTELLIGENCE AND METHOD OF OPERATION OF THE SYSTEM}

The present invention relates to a legal support system using artificial intelligence and a method of operating the system. More specifically, in response to a user's legal inquiry, artificial intelligence extracts the answer corresponding to the inquiry from a pre-stored legal database and automatically provides an answer. It concerns the legal support system that can be provided and how the system operates.

When users who are unfamiliar with legal knowledge encounter a legal problem, they seek opinions through the service of the knowledge portal site to resolve the problem, or seek opinions from experts such as lawyers, paralegals, and certified labor attorneys (hereinafter referred to as experts with a recognized qualification in the relevant legal field). It is common to obtain assistance from an 'expert'). However, in the case of knowledge portal sites, anyone, including non-experts, can reply, making it difficult to guarantee the accuracy of answers, and when seeking expert assistance, accessibility is limited due to issues such as the inconvenience of visiting, reluctance to meet face-to-face, and cost. It is falling.

Meanwhile, legal issues can be divided into things that the general public generally wants to know about and things that are very peripheral but not general. The general point is, 'Other people may be curious about the things I'm curious about, too.' For example, in 2019, the number of calls received by the call center dedicated to telephone inquiries at the Ministry of Employment and Labor was 9.78 million (2020 White Paper on Employment and Labor), and the number of inquiries received through other channels such as the website in addition to the call center is also large. Many of these legal questions are the same or very similar, and although early expert intervention may be necessary for cases with such a high degree of similarity, users can get faster answers by using the answer data accumulated in the process of answering multiple questions. Organizations that respond to these complaints will also be able to save a lot of time and manpower.

Considering these points, we are developing a system that can provide quick and accurate answers or advice using artificial intelligence and big data accumulated from public data such as laws, precedents, and administrative interpretations and guidelines regarding the legal knowledge that users want to obtain. This is required.

The technical problem that the present invention aims to solve is to automatically understand the contents of legal inquiries provided by users using artificial intelligence, and to use a legal database (hereinafter referred to as 'legal law') containing pre-stored laws, precedents, regulations, administrative interpretations, and guidelines. Through the expert verification process and deep learning based on the 'database'), we will ultimately develop a legal support system and operation method that can generate answers corresponding to the inquiry with a reliability level of 99% or higher without the need for expert verification procedures. It is provided.

The technical problems of the present invention are not limited to the technical problems mentioned above, and other technical problems not mentioned will be clearly understood by those skilled in the art from the description below.

A method of operating a legal support system using artificial intelligence according to an embodiment of the present invention to solve the above-described technical problem includes the steps of receiving a legal question including natural language text from a questioner (user), and the content of the legal question. Analyzing and matching a plurality of vocabularies included in the natural language text to a plurality of legal categories, Comparing the provided legal question with legal questions stored in an existing question database to calculate similarity, The calculated similarity and the plurality Generating a provisional answer from a pre-stored legal database based on the corresponding result of the vocabulary, providing the provisional answer as an answer subject to expert review, receiving an expert review result for the answer subject to expert review, the expert Generating an expert answer from the tentative answer based on a review result, and providing the expert answer to the questioner.

In some embodiments of the present invention, the step of providing the temporary answer as an answer subject to expert review includes, if the similarity is greater than or equal to a predetermined first similarity, the answer generated in response to the legal question stored in the existing question database and the temporary answer. A step of providing together as an answer subject to expert review, wherein the expert review results may include an answer generated in response to a legal question stored in the existing question database and an expert review result for the temporary answer.

In some embodiments of the present invention, after calculating the similarity by comparing the provided legal question with the legal question stored in the existing question database, if the similarity is greater than or equal to a predetermined second similarity, the legal question stored in the existing question database is responded to. A step of providing the generated answer as the temporary answer may be further included.

In some embodiments of the present invention, the step of generating the temporary answer includes inputting the plurality of vocabularies and the corresponding results of the plurality of vocabularies into an answer learning model generated based on existing query history and response history according to the query. and generating the temporary answer by extracting searched content from the legal database as an output for the input.

In some embodiments of the present invention, the step of training the answer learning model by using the plurality of vocabularies and the corresponding results of the plurality of vocabularies as input and the expert answer as output may be further included.

In some embodiments of the present invention, the pre-stored legal database may include at least one of laws, precedents, regulations, rules, and administrative interpretations and guidelines of competent ministries.

A legal support system using artificial intelligence according to another embodiment of the present invention to solve the above-described technical problem includes a processor; In the legal support system using artificial intelligence, including a memory connected to the processor, the memory is configured to: when executed, the processor receives a legal question including a natural language text from an interrogator; the content of the legal question; Analyzing and matching a plurality of vocabularies included in the natural language text to a plurality of legal categories, generating a temporary answer from a pre-stored legal database based on the correspondence results of the plurality of vocabularies, subjecting the temporary answer to expert review providing an answer as an answer, receiving an expert review result for the answer subject to expert review, generating an expert answer from the tentative answer based on the expert review result, and providing the expert answer to the questioner. Save the command to be executed.

Specific details of other embodiments are included in the detailed description and drawings.

The legal support system and method of operating the system using artificial intelligence according to an embodiment of the present invention classifies legal categories for major vocabulary by artificial intelligence in response to questions provided by users, and answers legal inquiries from a pre-stored legal database. By extracting corresponding content and generating an answer, quick and accurate answers can be provided to users. Through this, each step is ultimately learned several times, and when the similarity is above a certain level, it reaches the stage where an automatically generated answer is provided to the questioner (user) without an expert answer.

The effects of the present invention are not limited to the effects mentioned above, and other effects not mentioned will be clearly understood by those skilled in the art from the description of the claims.

1 is a diagram illustrating the configuration of a legal support system using artificial intelligence according to an embodiment of the present invention.
Figure 2a is a flowchart for explaining the operation method of a legal support system using artificial intelligence according to an embodiment of the present invention.
Figure 2b is a reference diagram to help understand the structure of the legal support system of the present invention.
Figure 3a is a diagram showing an example of a legal inquiry provided by a user by a legal support system using artificial intelligence according to an embodiment of the present invention, and Figure 3b is a diagram illustrating the classification of vocabulary included in the provided legal inquiry using a plurality of classification criteria. This is a drawing that shows the results so that they can be visually recognized.
Figure 4 is a diagram showing the criteria for determining the first classification by which the corresponding vocabulary can be classified in relation to the name of the related law by a legal support system using artificial intelligence according to an embodiment of the present invention.

The advantages and features of the present invention and methods for achieving them will become clear by referring to the embodiments described in detail below along with the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below and will be implemented in various different forms. The present embodiments only serve to ensure that the disclosure of the present invention is complete, and those skilled in the art It is provided to make it easier for those who understand 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 the specification.

One component is said to be “connected to” or “coupled to” another component when it is directly connected or coupled to another component or with an intervening other component. Includes all cases. On the other hand, when one component is referred to as “directly connected to” or “directly coupled to” another component, it indicates that there is no intervening other component. “And/or” includes each and every combination of one or more of the mentioned items.

The terms used in this specification are for describing embodiments and are not intended to limit the invention. As used herein, singular forms also include plural forms, unless specifically stated otherwise in the context. As used herein, “comprises” and/or “comprising” refers to the presence of one or more other components, steps, operations and/or elements. or does not rule out addition.

Although first, second, etc. are used to describe various components, it goes without saying that these components are not limited by these terms. These terms are merely used to distinguish one component from another. Therefore, of course, the first component mentioned below may also be the second component within the technical spirit of the present invention.

Unless otherwise defined, all terms (including technical and scientific terms) used in this specification may be used with meanings that can be commonly understood by those skilled in the art to which the present invention pertains. Additionally, terms defined in commonly used dictionaries are not interpreted ideally or excessively unless clearly specifically defined.

1 is a diagram illustrating the configuration of a legal support system using artificial intelligence according to an embodiment of the present invention.

Referring to Figure 1, the legal support system 100 using artificial intelligence according to an embodiment of the present invention includes an interface unit 110, a storage unit 120, a processor 130, a learning processor 140, and a memory ( 150) may be included.

The interface unit 110 may be connected to the user terminal 10 and the expert terminal 20 connected to the legal support system 100 to perform a communication function. Accordingly, the interface unit 110 may include software components such as an API (Application Programming Interface) for communicating with the terminals 10 and 20 along with hardware components including wired and wireless data communication.

Accordingly, when a user who wishes to make a legal inquiry accesses the legal support system 100 through the user terminal 10, the interface unit 110 receives a legal inquiry including natural language text from the user terminal 10, The answer generated through a series of processing can be transmitted to the user terminal 10. Additionally, the interface unit 110 may transmit a temporary answer generated through a process described later to the expert terminal 20 and transmit a result of expert review thereof.

The processor 130 may perform each operation performed by the legal support system 100 of the present invention by executing instructions temporarily or non-temporarily stored in the memory 150.

The processor 130 may control the operation of the legal support system 100 according to an embodiment of the present invention using data analysis and machine learning algorithms. For example, with respect to a legal query containing natural language text received from the user terminal 10, operations such as natural language processing using artificial intelligence, syntax analysis, and category matching of words constituting the sentence can be performed.

The processor 130 controls the learning processor 140 and may request, search, receive, and utilize classified and analyzed data from the learning processor 140. In addition, the processor 130 can control components included in the legal support system 100.

Learning processor 140 may be configured to perform data mining, data analysis, intelligent decision-making, and machine learning algorithms and to receive, classify, store, and output information to be used for technology. For example, learning processor 140 may use expert review results received from expert terminal 20 to identify, index, categorize, and manipulate data for use in supervised and unsupervised learning, data mining, predictive analytics, or other machines. , may be configured to store data in a legal database stored in the storage unit 120 for storage, retrieval, and output.

To this end, the learning processor 140 may be used by the processor 130, etc. using any of various types of data analysis algorithms and machine learning algorithms. In addition, the learning processor 140 receives data received from the user terminal 10 and the expert terminal 20 and undergoes a pre-processing process, and data pre-processed from the processor 130 to learn an internal artificial neural network or a storage unit ( An artificial neural network can be learned by obtaining preprocessed input data from a legal database stored in 120).

The memory 150 may store data supporting various functions of the legal support system 100. The memory 150 may store application programs running on the legal support system 100, data for operation of the legal support system 100, and commands.

The storage unit 120 may store programs and data necessary for the operation of the legal support system 100. For example, the storage unit 120 can store a legal database used in the legal support system 100, and can update the legal database based on analysis and learning results provided from the processor 130 and the learning processor 140. there is.

The legal database stored in the storage unit 120 is information that serves as the basis for answers generated to legal inquiries, and may consist only of information generated by public institutions, such as laws, precedents, regulations and rules, and administrative interpretations and guidelines of the competent ministries. there is. The data constituting this legal database may be classified and stored in the storage unit 120 based on criteria such as categories for each ministry in charge of the data and categories for each law.

Additionally, the storage unit 120 may include an answer learning model generated through machine learning from existing query details and corresponding answer details. Specifically, the legal support system 100 uses legal questions provided by existing users and the answers provided in response to them using an answer learning model created through machine learning such as deep learning to provide temporary information, which will be described later. You can extract the content necessary for the answer generation process.

FIG. 2A is a flowchart for explaining the operation method of a legal support system using artificial intelligence according to an embodiment of the present invention, and FIG. 2B is a reference diagram to help understand the structure of the legal support system of the present invention.

Referring to Figure 2a, the method of operating a legal support system using artificial intelligence according to an embodiment of the present invention includes the step of receiving a legal inquiry from a user (S110), analyzing the content of the legal inquiry and analyzing the content included in the natural language text. A step of matching vocabulary to a legal category (S120), a step of determining the similarity between the contents of the legal inquiry and existing inquiries (S130), and generating temporary answers and expert answers according to the similarity between the contents of the legal inquiry and existing inquiries. It may include steps (S140 to S180) and a step (S190) of providing the generated expert answer or the final answer by the system to the user.

Describing the operating method of the legal support system of the present invention in detail, first, in performing the step (S110) of receiving a legal inquiry from the user, the legal inquiry is, for example, at least one as shown in FIG. 3A. It may contain natural language text consisting of sentences. The interface unit 110 of the legal support system 100 may receive a legal inquiry from the user terminal 10 connected to the legal support system 100 and transmit it to the processor 130 or store it in the memory 150. .

Next, a step (S120) is performed in which the content of the legal query is analyzed and the vocabulary included in the natural language text corresponds to the legal category. As described above, natural language text constituting a legal query may include multiple vocabularies. In order to match each vocabulary, a preprocessing process may be performed in advance, including extracting the vocabulary included in the natural language text and selecting key vocabulary to select the corresponding target to the legal category.

For example, the processor 130 vectorizes each word of the vocabulary constituting the legal query through an algorithm such as word embedding, and calculates the frequency of appearance among the vocabulary constituting the legal query and between one vocabulary and another vocabulary. Key vocabulary can be selected based on the distance between vector spaces of .

Subsequently, the legal support system 100 may match a plurality of vocabularies included in the query to predetermined legal categories in order to analyze and classify the content of the legal query. At this time, the plurality of vocabularies included in the query may include key vocabularies selected through the preprocessing process described above.

Predetermined legal categories can be classified into, for example, a first category in which the vocabulary can be classified in relation to the name of the statute to which it relates, a second category in which the vocabulary can be classified in relation to the content of the statute to which it relates, and other vocabulary. May include a third category.

Figure 4 is a diagram showing the criteria for determining the first classification in which the corresponding vocabulary can be classified in relation to the relevant legal name.

Referring to Figure 4, for example, the Ministry of Employment and Labor category may include multiple laws such as the Labor Standards Act, the Minimum Wage Act, and the Employee Retirement Benefits Security Act. These laws may have a unique index and be stored in the legal database stored in the storage unit 120. Although not shown in FIG. 4, precedents, regulations and rules, and administrative interpretations and guidelines of competent ministries stored in the legal database may also be stored, including link information to related laws classified according to the first classification.

The processor 130 may match the vocabulary included in the query to the first classification based on whether the vocabulary included in the query has a similarity greater than or equal to a predetermined standard value with the names of a plurality of laws stored in the legal database.

As shown in Figure 3b, in the case of the vocabulary "labor union", it is judged to have a degree of similarity greater than a predetermined value with the legal name of the Trade Union and Labor Relations Adjustment Act, and may correspond to the first classification. The legal support system 100 stores in the storage unit 120 or memory 150 the results of locating the names of laws and regulations included in the legal database on word vectors, and the word vectors of major vocabulary included in the user's legal inquiry. By calculating the similarity using the distance from the law name in the above, the corresponding vocabulary can be matched to the first classification.

If the vocabulary of the first category is classified among the main vocabulary constituting the query, a second category, that is, a series of words that can be classified in relation to the content of the law corresponding to the vocabulary of the first category, can be classified.

For example, in the case of words such as "provisions", "change", and "organization" among the vocabulary shown in Figure 3b, contents related to the relevant laws and regulations, for example, Article and administrative interpretation may correspond to a second classification related to at least one content. The processor 130 may select a vocabulary of the second category based on the similarity between the data constituting the data under the Trade Union Act category, that is, the data constituting each of the laws, precedents, and administrative interpretations, and the vocabulary constituting the legal inquiry, and may select the vocabulary of the second category in advance. Vocabulary with a similarity level greater than a predetermined value or a vocabulary with a predetermined number of higher levels of similarity can be assigned to the second classification.

If the second category of vocabulary is classified among the main vocabulary constituting the query, the third category, that is, the vocabulary to be used for data extraction to generate temporary answers in the legal database among the key words, can be additionally classified. Among the major vocabularies, these may be vocabulary words that are not classified into the first or second classification, but are recognized as having similarity to data such as laws, precedents, and administrative interpretations that make up the legal database.

Next, a step (S130) may be performed to determine the similarity between the content of the legal inquiry and the existing inquiry.

The legal support system 100 may store legal questions provided by existing users, temporary answers, expert answers, and final answers generated in response to the legal questions in the storage unit 120. The legal support system 100 receives a new legal query from the user, analyzes it, matches the included vocabulary to a legal category, and then determines the degree of similarity (S) that the new legal query has with the existing query. You can.

Judgment of the similarity between a new legal query and an existing query can be performed based on whether the main vocabulary and particles included in the queries are common. In particular, since the corresponding results of the legal categories of major vocabulary were derived through the content analysis process of the previous legal inquiry, if the same vocabulary and particle are included in the same legal category as the analysis result of the existing inquiry, weight can be assigned to the similarity calculation. . For example, if the new legal query and the existing query contain the common "union" vocabulary in the first category, the similarity calculation may be weighted.

In some embodiments of the present invention, a new legal query and an existing query used to determine similarity may include multiple query details. Therefore, among a plurality of existing queries that have similarity to the content of the new query, the similarity (S) of the new legal query can be calculated based on the existing query with the highest similarity.

The legal support system 100 may determine the range of similarity (S) of the new legal inquiry and decide whether to use the answer to the existing inquiry based on the similarity (S).

If the similarity (S) of a new legal query is a new type that did not previously exist or if the query content does not match the content stored in the legal database, the similarity (S) may have a value smaller than the first value (S1). . At this time, the first value (S1) may be, for example, a value of less than 50%, but the present invention is not limited thereto. For these legal inquiries, a process may be required in which raw data based on a legal database is generated as a temporary answer and provided to the expert terminal 20.

In general, a step (S140) of generating a temporary answer may be performed by extracting content corresponding to the response result of the legal category of the vocabulary constituting the query from the legal database.

By analyzing each main vocabulary in the previous step (S120), a temporary answer can be generated based on the name of the law determined by the vocabulary corresponding to the first classification and the content of the law determined by the vocabulary corresponding to the second classification. . For example, the processor 130 records content corresponding to the response results of legal categories such as change in regulations, division of organization, and effect of collective agreement (second classification) within the Trade Union Act (first classification) into the legal database stored in the storage unit 120. You can generate a temporary answer by extracting from .

The process of extracting content corresponding to the corresponding result of the classified legal category from the legal database can also be performed by determining the similarity between the classified words and the content stored in the legal database. Specifically, the answer learning model pre-stored in the storage unit 120 is loaded, and key words included in the user's legal inquiry are provided as input to the answer learning model.

As described above, the answer learning model is a model created from existing question details and corresponding answer details. The processor 130 or the learning processor 140 searches the legal database using key words provided as input, and then selects at least one data with the highest similarity based on the similarity between the content of the data stored in the legal database and the key words. , that is, raw data selected from laws, precedents, regulations/rules, and administrative interpretations and guidelines of the competent ministries can be selected and created as temporary answers.

Subsequently, a step (S150) may be performed in which the provisional answer is provided as an answer subject to expert review and the expert review result thereof is received.

Although the temporary answer is based on legal grounds by selecting data recognized as similar to the legal question provided by the user by artificial intelligence, the legal support system 100 of the present invention provides a temporary answer to improve the completeness of the answer. A separate expert review process can be conducted.

To this end, a temporary answer can be provided to the expert terminal 20 as an answer subject to expert review, and the result of expert review of the answer subject to expert review can be received from the expert terminal 20. The expert review result is a type of method that improves the accuracy of the answer by reviewing the validity of the temporary answer generated from the legal support system 100 by an expert in the field, and providing the corrected answer as an expert answer if error correction or content addition is necessary. It can perform the function of supervised learning.

This may include a process of additionally training an answer learning model using the user's legal question and verified expert answers to the question. In other words, the answer reflects the expert's opinion by upgrading and learning the answer learning model based on the main vocabulary included in the user's legal inquiry, the legal category classification results of the main vocabulary, and expert answers that reflect the expert review results for the user's legal inquiry. You can create an accurate answer learning model using .

On the other hand, if the similarity (S) of the new legal inquiry does not contain the same content as the existing query content but is somewhat related, the similarity (S) is greater than the first value (S1) and smaller than the second value (S2). You can have At this time, the range of the first value (S1) to the second value (S2) may be 50 to 99%, but the present invention is not limited thereto.

If a query with a similarity greater than the first value (S1) and less than the second value (S2) is found among the stored existing legal queries, the existing legal query extracted from the content corresponding to the corresponding result of the legal category of the vocabulary from the legal database A step (S160) may be performed in which the expert answer to the inquiry is provided as a temporary answer to the expert terminal as an expert review answer and the expert review result is received.

In other words, if a query with similarity (S1<S<S2) within a certain range is found among the existing stored legal queries, an expert answer generated based on the content of the user's legal query, that is, generated according to the existing query, is sent. It is provided as a temporary answer to the expert terminal 20 as an answer subject to expert review. Through this process, expert review results can be referenced to review and supplement expert answers to existing questions.

Subsequently, a step (S170) of generating an expert answer using the expert review result may be performed. In other words, a more upgraded expert answer can be generated by using the answer data generated by providing already created expert answers to similar existing questions.

If a query with a similarity greater than the second value (S2) is found among the stored existing legal questions, an expert answer to the existing question can be generated as an expert answer to the user's legal question (S180). In other words, if the similarity is greater than a certain range, the existing expert answer retrieved through similarity judgment can be provided to the user as is without any additional processing. This ultimately allows answers to be provided by the system without intervention by experts.

Finally, a step (S190) of transmitting the generated expert answer or final answer through the user terminal 10 may be performed. The interface unit 110 may provide the generated expert answer or final answer to the connected user terminal 10.

As described above, the legal support system 100 and its operation method according to an embodiment of the present invention classifies legal categories for key vocabulary using artificial intelligence in response to questions provided by users, and pre-stores the legal database. By extracting content corresponding to legal questions and generating answers, quick and accurate answers can be provided to users.

In addition, the accuracy of the answer is improved by generating expert review results for the initially generated answer and ultimately providing the user with an answer that reflects the expert review result, and as described above, similar questions can be answered through continuous learning. Expert intervention is minimized, and ultimately, the system can generate and provide answers on its own without expert intervention.

In some other embodiments of the present invention, after generating an expert answer, a process of additionally training an answer learning model using the user's legal question and the expert answer may be further performed. In other words, the expert's opinion is reflected by further training the answer learning model based on the existing expert answers that reflect the main vocabulary included in the user's legal inquiry, the legal category classification results of the main vocabulary, and the expert review results for the user's legal inquiry. You can create an accurate answer learning model using the answers.

The present invention can also be implemented as computer-readable code on a computer-readable recording medium. Computer-readable recording media include all types of recording devices that store data that can be read by a computer device. Examples of computer-readable recording media include hard disks, ROMs, RAM, CD-ROMs, hard disks, magnetic tapes, floppy disks, and optical data storage devices, as well as carrier waves (e.g., transmission via the Internet). It also includes implementation in the form of.

Although embodiments of the present invention have been described above with reference to the attached drawings, those skilled in the art will understand that the present invention can be implemented in other specific forms without changing the technical idea or essential features. You will be able to understand it. Therefore, the embodiments described above should be understood in all respects as illustrative and not restrictive.

100: Legal support system 110: Interface department
120: storage unit 130: processor
140: Running processor 150: Memory

Claims (12)

receiving a legal question including natural language text from the questioner;
Analyzing the content of the legal question and matching a plurality of vocabulary words included in the natural language text to a plurality of legal categories;
Comparing the provided legal questions with legal questions stored in an existing question database to calculate similarity;
generating a temporary answer from a pre-stored legal database based on the calculated similarity and a corresponding result of the plurality of vocabularies;
providing the provisional answer as an answer subject to expert review;
Receiving an expert review result for the answer subject to expert review;
generating an expert answer from the provisional answer based on the expert review results; and
Including providing the expert answer to the questioner,
How a legal support system using artificial intelligence operates. According to claim 1,
The step of providing the provisional answer as an answer subject to expert review is:
When the similarity is greater than or equal to a predetermined first similarity, providing an answer generated in response to a legal question stored in the existing question database and the temporary answer together as an answer subject to expert review,
The expert review results include answers generated in response to legal questions stored in the existing question database and expert review results for the temporary answers,
How a legal support system using artificial intelligence operates. According to claim 1,
After calculating the degree of similarity by comparing the provided legal questions with legal questions stored in the existing question database,
If the similarity is greater than or equal to a predetermined second similarity, providing an answer generated in response to a legal question stored in an existing question database as the expert answer, further comprising:
How a legal support system using artificial intelligence operates. According to paragraph 1,
The steps for generating the temporary answer are:
The plurality of vocabularies and the corresponding results of the plurality of vocabularies are input into an answer learning model created based on the existing query history and the response history according to the query, and the raw data (raw) retrieved from the legal database is output for the input. Comprising the step of extracting data and generating the temporary answer,
How a legal support system using artificial intelligence operates. According to paragraph 4,
Further comprising training the answer learning model using the plurality of vocabularies and the corresponding results of the plurality of vocabularies as input and the expert answer as output,
How a legal support system using artificial intelligence operates. According to paragraph 1,
The pre-stored legal database is,
Containing at least one of laws, precedents, regulations, rules, administrative interpretations and guidelines of the competent ministry,
How a legal support system using artificial intelligence operates. A computer-readable recording medium on which a program for executing any one of the methods of claims 1 to 6 is recorded. processor; and
In a legal support system using artificial intelligence including a memory connected to the processor,
The memory, when executed, the processor,
receiving a legal question including natural language text from the questioner;
Analyzing the content of the legal question and matching a plurality of vocabulary words included in the natural language text to a plurality of legal categories;
Comparing the provided legal questions with legal questions stored in an existing question database to calculate similarity;
generating a temporary answer from a pre-stored legal database based on the similarity and correspondence results of the plurality of vocabularies;
providing the provisional answer as an answer subject to expert review;
Receiving an expert review result for the answer subject to expert review;
generating an expert answer from the provisional answer based on the expert review results; and
A legal support system using artificial intelligence that stores commands to execute steps for providing the expert answer to the questioner. According to clause 8,
The step of providing the provisional answer as an answer subject to expert review is:
For inquiries that have a degree of similarity within a certain range among existing legal inquiries, an expert answer generated according to the existing inquiry is provided as a temporary answer and provided as an answer subject to expert review,
The expert review results include answers generated in response to legal questions stored in the existing question database and expert review results for the temporary answers,
Legal support system using artificial intelligence. According to clause 9,
The memory is,
After calculating the degree of similarity by comparing the provided legal questions with legal questions stored in the existing question database,
Storing a command to further execute the step of providing an answer generated in response to a legal question stored in an existing question database as the expert answer when the similarity is greater than or equal to a predetermined second similarity,
Legal support system using artificial intelligence. According to clause 8,
The steps for generating the temporary answer are:
The plurality of vocabularies and the corresponding results of the plurality of vocabularies are input into an answer learning model created based on the existing query history and the response history according to the query, and the raw data (raw) retrieved from the legal database is output for the input. Comprising the step of extracting data and generating the temporary answer,
Legal support system using artificial intelligence. According to clause 11,
The memory is,
Storing a command to further execute the step of training the answer learning model using the plurality of vocabularies and the corresponding results of the plurality of vocabularies as input and the expert answer as output,
Legal support system using artificial intelligence.