KR20200139008A - User intention-analysis based contract recommendation and autocomplete service using deep learning - Google Patents
User intention-analysis based contract recommendation and autocomplete service using deep learning Download PDFInfo
- Publication number
- KR20200139008A KR20200139008A KR1020190065509A KR20190065509A KR20200139008A KR 20200139008 A KR20200139008 A KR 20200139008A KR 1020190065509 A KR1020190065509 A KR 1020190065509A KR 20190065509 A KR20190065509 A KR 20190065509A KR 20200139008 A KR20200139008 A KR 20200139008A
- Authority
- KR
- South Korea
- Prior art keywords
- contract
- user
- service
- recommended
- learning
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Ceased
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
- G06Q50/18—Legal services
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/903—Querying
- G06F16/9032—Query formulation
- G06F16/90324—Query formulation using system suggestions
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/258—Heading extraction; Automatic titling; Numbering
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/268—Morphological analysis
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- General Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Mathematical Physics (AREA)
- Tourism & Hospitality (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Technology Law (AREA)
- Economics (AREA)
- Marketing (AREA)
- General Business, Economics & Management (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Evolutionary Computation (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
Description
본 발명은 딥러닝 기술의 활용 분야에 관한 것으로서, 보다 상세하게는 딥러The present invention relates to the field of application of deep learning technology, and in more detail,
닝 기술을 적용하여 사용자가 원하는 계약서의 종류를 추천해주고 해당 계약서에서 필수적인 내용만 알려주면 자동으로 계약서를 완성해주는 시스템과 그 서비스 방법에 관한 것이다.It is about a system that automatically completes the contract by applying the Ning technology to recommend the type of contract that the user wants and informs only the essential contents of the contract, and its service method.
본 발명의 근간이 되는 기술인 문서의 분류 및 범주화는 문서(계약서)의 내용을 기준으로 하여 하나 혹은 그 이상의 범주에 할당하는 것이다. 기계학습에서 사용되는 자동 분류 알고리즘들은 문서에 표현되는 단어로부터 문서 벡터를 생성하고, 벡터화된 학습 문서들을 예제로 사용하여 학습함으로써 관련된 문서에 범주를 할당한다. Classification and categorization of documents, which is the technology underlying the present invention, is assigned to one or more categories based on the content of the document (contract). Automatic classification algorithms used in machine learning create a document vector from words expressed in the document, and assign categories to related documents by learning using vectorized learning documents as examples.
또한 문서의 핵심 내용을 파악하도록 하는 문서 요약은 주어진 문서를 잘 표현할 수 있는 압축된 문장들을 만들어내는 과정인데 핵심 문장들을 선별하여 구성하는 추출적(extractive) 요약과 전체 내용을 잘 나타내는 문장을 직접 생성해내는 추상적(abstractive) 요약으로 분류할 수 있다. 이러한 문서 처리 관련 기술들을 근간으로 계약서의 내용을 분석 파악, 분류하도록 학습을 할 수는 있지만 계약서를 구성하는 필수 요소들이나 변호사, 법률 전문가의 대면적 지원이 필요한 부분까지를 판별하거나 알려주는 측면에서 한계점이 존재하였다. In addition, document summarization to grasp the core content of a document is a process of creating compressed sentences that can express a given document well. An extractive summary that selects and composes the core sentences and sentences that represent the entire content are directly generated. It can be categorized as an abstractive summary. You can learn to analyze, grasp, and classify the contents of a contract based on these document processing related technologies, but there is a limitation in discriminating or notifying the essential elements of the contract or parts that require large-scale support from lawyers and legal experts. Existed.
본 발명은 상술한 종래 기술의 문제점을 감안한 것으로서, 일반 문서보다 그 전문성이나 내용 파악의 난이도가 높은 계약서를 문자 인식하고 판독 및 분류, 특성 추출의 기술을 한 단계 높임은 물론 전문가의 도움 없이도 사용자가 원하는 계약서를 자동으로 생성해주는 기술 및 그 서비스 방법을 제공한다.The present invention takes into account the problems of the prior art described above, and enhances the technology of character recognition, reading, classification, and feature extraction of contracts with higher expertise or difficulty in grasping the contents than general documents, as well as enhancing the technology of reading, classifying, and extracting features, users can It provides the technology that automatically generates the desired contract and its service method.
본 발명에 이용될 핵심 기술인 딥러닝(Deep Learning)은 여러 비선형 변환기법의 조합을 통해 높은 수준의 추상화를 시도하는 기계학습의 한 분야로서 심층 신경망(Deep Neural Network, DNN), 합성곱 신경망(Convolutional Neural Network, CNN), 순환 신경망(Recurrent Neural Network, RNN), 제한 볼츠만 머신(Restricted Boltzmann Machine, RBM), 심층 신뢰 신경망(Deep Belief Network, DBN), 심층 Q-네트워크(Deep Q-Networks) 등 주어진 데이터 집합에 적합하게 적용할 수 있는 다양한 알고리즘이 존재하며, 영상 인식, 자연어 처리, 자동 음성 인식 등 다양한 응용 분야에 활용되고 있다.Deep Learning, a core technology to be used in the present invention, is a field of machine learning that attempts a high level of abstraction through a combination of several nonlinear transducers, and is a deep neural network (DNN) and a convolutional neural network. Neural Network (CNN), Recurrent Neural Network (RNN), Restricted Boltzmann Machine (RBM), Deep Belief Network (DBN), Deep Q-Networks, etc. There are various algorithms that can be appropriately applied to a data set, and are used in various application fields such as image recognition, natural language processing, and automatic speech recognition.
본 발명에 따르면, 문서 인식 및 분류 기술 중에서 계약서 분야에 대한 진보된 기술의 구현은 물론 계약서 작성이 필요한 고객이 고가의 변호사 비용 때문에 전문적인 법률 서비스를 받지 못 하거나 계약상의 실수가 발생하는 경우를 대폭 줄일 수 있다. 또한 변호사나 법률 전문 서비스의 입장에서도 본 발명의 기술을 이용하면 고객을 위한 계약서 작성을 일일이 수작업으로 진행하던 방식에서 벗어나 자동 작성 후 검수 및 수정만 해도 됨으로써 시간과 비용, 업무량을 획기적으로 절감할 수 있다.According to the present invention, among document recognition and classification technologies, customers who need to implement advanced technologies in the field of contracts as well as contracts can not receive professional legal services due to expensive attorney fees, or a contractual mistake occurs. Can be reduced. In addition, from the standpoint of lawyers or legal professional services, if the technology of the present invention is used, it is possible to significantly reduce time, cost, and workload by simply reviewing and modifying after automatic writing, instead of manually writing contracts for customers. have.
도 1은 기존의 계약서로부터 키워드를 추출하는 절차를 나타내는 도면이다.
도 2는 본 발명을 통한 계약서 자동완성 서비스를 구현하기위한 실질적인 방법과 절차를 나타내는 도면이다.
도 3은 서비스 이용자의 의도를 분석 후 최적의 계약서 템플리트 제시 및 추가적인 질의·응답을 통해 계약서를 완성해가는 과정의 예시를 나타내는 도면이다.1 is a diagram showing a procedure for extracting a keyword from an existing contract.
2 is a diagram showing a practical method and procedure for implementing a contract automatic completion service through the present invention.
3 is a diagram showing an example of a process of completing a contract through analysis of the intention of a service user, presentation of an optimal contract template, and additional questions and answers.
도 1은 비정형 이미지 또는 문서 파일 형태의 계약서를 문자 인식하고 텍스트 전처리 및 키워드 추출을 위한 알고리즘을 적용하는 절차이다. 인공지능 학습을 위한 기존의 계약서 내용을 인식하기 위하여, 문자 영역을 분할 변환하여 판별 및 분류를 하고 학습 모델을 통해 단어 인식 정확도를 높이게 된다. 이 때 계약서(이미지)가 회전 되었거나 구겨진 문서도 보정을 통해 인식률을 높이도록 할 수 있다.1 is a procedure for recognizing a contract in the form of an unstructured image or document file and applying an algorithm for text pre-processing and keyword extraction. In order to recognize the contents of the existing contract for artificial intelligence learning, the character area is divided and transformed to identify and classify, and the accuracy of word recognition is improved through a learning model. At this time, the recognition rate can be increased by correcting even the document with a rotated or wrinkled contract (image).
이후 자연어 이해를 위한 형태소 분석 알고리즘을 이용하게 되는데, 형태소 분석은 표층형 (surface level form)인 어절로부터 의미가 있는 최소 단위인 형태소 (morpheme)를 추출하는 작업으로써 이를 위해서는 어절을 분석하여 형태소의 결합으로 분리하고, 각 형태소에 품사정보를 할당하고, 형태소 결합 시 발생하는 음운 변화를 원형(root form)으로 복원하는 것이 필요하다.Afterwards, a morpheme analysis algorithm is used to understand natural language, and morpheme analysis is a task of extracting a morpheme, which is the smallest meaningful unit, from a word that is a surface level form. It is necessary to separate the morphemes, allocate part-of-speech information to each morpheme, and restore the phonological changes that occur when morphemes are combined into a root form.
계약서의 핵심 내용 파악을 위해서 계약서 내의 가변 항목들을 표준화 하고, 유형화, 법률 사항에 적합 하도록 특수 조항들에 대한 분석을 유형화하고, 각 단어의 가중치를 계산한 후 집단 간 텍스트 특성의 차이나 토큰 사이의 관계 등을 분석하여 상위 적당 K개수의 가중치를 가지는 키워드를 선정한다.To identify the core contents of the contract, standardize the variable items in the contract, categorize them, and categorize the analysis of special provisions to suit legal matters, calculate the weight of each word, and then calculate the difference in text characteristics between groups or the relationship between tokens. Etc. are analyzed to select keywords having a weight of the highest suitable K number.
도 2는 본 발명에 의한 계약서 추천 및 자동 완성 서비스의 흐름도로써 사용자가 서비스 시스템과의 질의 응답을 통해서 필요한 분야와 종류의 계약서를 추천 받고 이후 계약 당사자 및 주요 계약 내용에 대한 입력을 받은 후에 자동 작성 및 보정과 사용자 검수를 거쳐 최종본이 완성되게 된다.2 is a flow chart of a contract recommendation and automatic completion service according to the present invention, which is automatically created after a user receives a recommendation for a contract of a required field and type through a query response with a service system, and after receiving the input of the contracting party and main contract details. And the final version is completed through correction and user inspection.
도 3은 도2의 내용을 예시를 통해 알기 쉽게 표현한 것으로 변호사와 상담과정을 통해 계약서를 작성하는 실제 과정과 동일한데, 1항의 '무엇을 도와 드릴까요?'라는 질문과 2항의 추가 질의에 사용자가 현재 상황에 대한 기본 내용을 입력하면 그 텍스트와 답변을 분석하여 가장 적합한 계약서 양식을 추천하여 제시하게 된다. 이 때 추천 알고리즘의 작동은, 여러 문서로 이루어진 문서군이 있을 때 어떤 단어가 특정 문서 내에서 얼마나 중요한 것인지를 나타내는 통계적 수치인 TF-IDF 가중치를 활용하여 계약서에 포함된 단어들과 그 중요도를 벡터공간에 저장하고, 계약서 내 단어들의 벡터를 학습시킬 때 특정 조항에서 유추할 수 있는 단어의 확률을 최대화 할 수 있는 방법으로 학습시켜 유사도를 높인 후에 질의어의 의미를 확장하여 질의 의미와 유사도가 높은 계약서를 추천하게 된다.Fig. 3 is an easy-to-understand representation of the contents of Fig. 2, which is the same as the actual process of creating a contract through a counseling process with an attorney, but the user responds to the question of ‘What can I do for you?’ in
추천에 따라 제공된 양식에 맞춰 계약서가 자동으로 1차 완성 및 교정이 수행되고 유사 사례 및 관련 법조항을 비교할 수 있도록 추천 제시가 되므로 사용자가 보다 정확한 법 적용을 이해할 수 있도록 한다. 마지막으로 사용자의 최종 확인을 거치면 계약서 완성본이 문서 파일이나 원하는 포맷으로 제공된다. In accordance with the form provided according to the recommendation, the first completion and correction of the contract is automatically performed, and the recommendation is presented so that similar cases and related laws can be compared, so that users can understand the application of the law more accurately. Finally, after the user's final confirmation, the completed contract is provided in a document file or in a desired format.
이상, 본 발명의 상세한 설명에서는 구체적인 실시예에 관해서 설명하였으나, 본 발명의 범위에서 벗어나지 않는 한도 내에서 여러 가지 변형이 가능함은 당해 분야에서 통상의 지식을 가진 자에게 있어서 자명하다 할 것이다.As described above, in the detailed description of the present invention, specific embodiments have been described, but it will be apparent to those of ordinary skill in the art that various modifications can be made without departing from the scope of the present invention.
Claims (1)
사용자가 자연어로 입력하는 현재 상황 및 원하는 계약 내용을 분석하여 어떤 종류의 계약서가 필요한지를 근접도 순으로 추천하는 동작;
사용자가 자연어로 입력하는 현재 상황 및 원하는 계약 내용을 분석하여 추천이 될 수 있는 특정 종류의 계약서에 필요한 추가 항목을 스스로 판단하고 질의하는 동작;
사용자가 입력한 내용 및 추가 질의로 얻어진 정보들을 이용하여 가장 근접하도록 추천된 특정 종류 계약서의 템플릿을 완성시키는 동작;
계약서 별 학습을 통해 특정 계약서의 형식 뿐 아니라 통상적으로 자주 추가되는 권장 조항들을 습득하여 추천 계약서의 구성 과정에서 사용자에게 검토하도록 표시해주는 동작;
을 포함하는,
컴퓨터 판독가능 저장매체에 저장된 컴퓨터 프로그램
A computer program stored in a computer-readable storage medium, wherein the computer program, when executed on one or more processors, causes the following operations to be performed for providing an artificial intelligence service within an entire system, the operations:
Analyzing the current situation and the desired contract content entered by the user in natural language, and recommending what kind of contract is required in order of proximity;
Analyzing the current situation and desired contract content input by the user in natural language to determine and query additional items necessary for a specific type of contract that can be recommended;
Completing a template of a specific type of contract recommended to be closest by using the content input by the user and the information obtained from the additional query;
Learning each contract by learning not only the format of a specific contract, but also the commonly added recommended clauses, and displaying it to the user for review during the construction of the recommended contract;
Containing,
Computer programs stored on a computer-readable storage medium
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
KR1020190065509A KR20200139008A (en) | 2019-06-03 | 2019-06-03 | User intention-analysis based contract recommendation and autocomplete service using deep learning |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
KR1020190065509A KR20200139008A (en) | 2019-06-03 | 2019-06-03 | User intention-analysis based contract recommendation and autocomplete service using deep learning |
Publications (1)
Publication Number | Publication Date |
---|---|
KR20200139008A true KR20200139008A (en) | 2020-12-11 |
Family
ID=73786354
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
KR1020190065509A Ceased KR20200139008A (en) | 2019-06-03 | 2019-06-03 | User intention-analysis based contract recommendation and autocomplete service using deep learning |
Country Status (1)
Country | Link |
---|---|
KR (1) | KR20200139008A (en) |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR102356972B1 (en) * | 2021-04-23 | 2022-02-08 | 주식회사 엘젠아이씨티 | an agricultural machine sharing service system based on IOT and AI |
CN114663069A (en) * | 2022-04-11 | 2022-06-24 | 中国建筑第二工程局有限公司 | Engineering project contract full-process management method and system |
KR102478967B1 (en) * | 2022-03-11 | 2022-12-20 | 주식회사 이지트러스트 | Method and system for providing blockchain-based contract management service interlocked with metaverse |
KR102596533B1 (en) | 2023-06-16 | 2023-11-01 | 주식회사 태건리너텍 | Composting fuel treated with sewage sludge and its treatment method and system |
CN117851602A (en) * | 2024-03-07 | 2024-04-09 | 武汉百智诚远科技有限公司 | Automatic legal document classification method and system based on deep learning |
KR102663183B1 (en) * | 2023-05-16 | 2024-05-03 | 행정사법인 민행24 | Agency Affairs Automatic Document Creation System |
KR102721156B1 (en) * | 2023-11-15 | 2024-10-23 | 에스지앤 주식회사 | Method, apparatus and computer-readable medium for managing work plan and work based on natural language processing and policy analysis |
-
2019
- 2019-06-03 KR KR1020190065509A patent/KR20200139008A/en not_active Ceased
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR102356972B1 (en) * | 2021-04-23 | 2022-02-08 | 주식회사 엘젠아이씨티 | an agricultural machine sharing service system based on IOT and AI |
KR102478967B1 (en) * | 2022-03-11 | 2022-12-20 | 주식회사 이지트러스트 | Method and system for providing blockchain-based contract management service interlocked with metaverse |
CN114663069A (en) * | 2022-04-11 | 2022-06-24 | 中国建筑第二工程局有限公司 | Engineering project contract full-process management method and system |
CN114663069B (en) * | 2022-04-11 | 2022-12-23 | 中国建筑第二工程局有限公司 | Engineering project contract full-process management method and system |
KR102663183B1 (en) * | 2023-05-16 | 2024-05-03 | 행정사법인 민행24 | Agency Affairs Automatic Document Creation System |
KR102596533B1 (en) | 2023-06-16 | 2023-11-01 | 주식회사 태건리너텍 | Composting fuel treated with sewage sludge and its treatment method and system |
KR102721156B1 (en) * | 2023-11-15 | 2024-10-23 | 에스지앤 주식회사 | Method, apparatus and computer-readable medium for managing work plan and work based on natural language processing and policy analysis |
CN117851602A (en) * | 2024-03-07 | 2024-04-09 | 武汉百智诚远科技有限公司 | Automatic legal document classification method and system based on deep learning |
CN117851602B (en) * | 2024-03-07 | 2024-05-14 | 武汉百智诚远科技有限公司 | Automatic legal document classification method and system based on deep learning |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
KR20200139008A (en) | User intention-analysis based contract recommendation and autocomplete service using deep learning | |
KR102318103B1 (en) | Method for machine learning train set and recommendation systems to recommend the scores to match between the recruiter and job seekers, and to give the scores of matching candidates to recruiters and to give the pass scores to job seekers respectively | |
US11699034B2 (en) | Hybrid artificial intelligence system for semi-automatic patent infringement analysis | |
Biyani et al. | Using non-lexical features for identifying factual and opinionative threads in online forums | |
CN111985243B (en) | Emotion model training method, emotion analysis method, device and storage medium | |
JP7297458B2 (en) | Interactive content creation support method | |
CN118886519B (en) | Model training method, data processing method, electronic device and storage medium | |
CN116341519B (en) | Method, device and storage medium for extracting event causal relationships based on background knowledge | |
US20240160642A1 (en) | Systems and methods for categorization of ingested database entries to determine topic frequency | |
CN110610003A (en) | Method and system for assisting text annotation | |
Zhang et al. | Business chatbots with deep learning technologies: state-of-the-art, taxonomies, and future research directions | |
US20230205779A1 (en) | System and method for generating a scientific report by extracting relevant content from search results | |
CN115757831B (en) | Method and device for semi-automatically constructing domain knowledge graph | |
WO2023233392A1 (en) | Method and system for producing unified natural language processing objects | |
Tesfa et al. | Aspect-based sentiment analysis on amharic text for evaluating ethio-telecom services | |
CN119311843A (en) | Solution matching method and system for artificial intelligence large model | |
CN119397474A (en) | A judicial data file generation method, system and computer program product | |
Wang et al. | Educational big data analytics using sentiment analysis for student requirement analysis on courses | |
Baghbanzadeh | Job-Resume Compatibility Scoring Using Graph Neural Networks and Large Language Models | |
Vilalta et al. | Studying the impact of the full-network embedding on multimodal pipelines | |
CN112558913A (en) | Conversation method and device based on aggregated card, computer equipment and storage medium | |
Xiong et al. | Enhancing food safety review classification with large language models and label embedding | |
Chen | A Study on Employment Problems and Sentiment Analysis of College Students Based on Bert-BiLSTM | |
Abdullah et al. | Chinese Monolingual Sentiment Classifier for Social Media Data Using Corpus-Based Approach | |
Harris | Identifying student difficulty and frustration from discussion forum postings |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PA0109 | Patent application |
Patent event code: PA01091R01D Comment text: Patent Application Patent event date: 20190603 |
|
PA0201 | Request for examination |
Patent event code: PA02012R01D Patent event date: 20200601 Comment text: Request for Examination of Application Patent event code: PA02011R01I Patent event date: 20190603 Comment text: Patent Application |
|
PG1501 | Laying open of application | ||
E902 | Notification of reason for refusal | ||
PE0902 | Notice of grounds for rejection |
Comment text: Notification of reason for refusal Patent event date: 20220831 Patent event code: PE09021S01D |
|
E601 | Decision to refuse application | ||
PE0601 | Decision on rejection of patent |
Patent event date: 20221117 Comment text: Decision to Refuse Application Patent event code: PE06012S01D Patent event date: 20220831 Comment text: Notification of reason for refusal Patent event code: PE06011S01I |