GraphDB Application Cases


Establishment of a compliance system

Development of an intelligent crime prevention cooperation system – Prosecution Service

Although “the safety of the people” is the government’s main government administration strategy, it is difficult to prevent crimes continuously in the long term by only temporary personnel and budget investment (increase in public safety manpower, CCTV installation, reinforcement campaigns, etc.) and crime prevention activities carried out by each government agency, An intelligent system that can reinforce the preemptive prevention-based criminal justice system through the establishment of scientific crime analysis system for ·effective crime prevention by analyzing a crime that has occurred in detail, sharing that information, and cooperating with relevant authorities,· is necessary, and close joint cooperation systems between people, the government and and regions for education, ·investigation post management and realization of victims’ and assailants’ normal rehabilitation within society by analyzing crime information, administrative and social information scientifically and sharing such information between relevant authorities is also necessary. In order to solve large problems from these two viewpoints, the scientific crime analysis-based system was established, the crime graph data-based intelligent crime analysis system was established, and the basis for the organization of crime prevention cooperation system was created.


< Interlligent crime prevention analysis system >

① Business Content

  1. Establishment of the scientific crime analysis base system.
    • Establishment of structured and unstructured data convergence infrastructure for five violent crimes (murder, sexual assault, robbery, arson, death resulting from bodily injury through assault).
    • Text analysis of approximately 800 semi-structured and unstructured documents.
    • Designing of various scenarios that support crime analysis.
    • Analysis of the evidence of crime, such as a criminal’s voice, video or picture.
    • Technical methodology research (external POC execution).
  2. Establishment of an intelligent crime analysis knowledge base.
    • Establishment of base materials for deep crime analysis.
    • Establishment of a concept dictionary and crime classification system for deep crime analysis.
    • Establishment of machine learning data and a knowledge expression system for the automation of in-dept crime analysis.
  3. Preparation of a base for a crime prevention cooperation system between relevant authorities.
    • dentification of the information required for crime prevention establishment by relevant authorities.
    • Preparation of a linkage base method with relevant authorities.
② Applied technologies and solutions.
  1. Application of the GraphDB Suite product
    Applies the GraphDB Suite product for graph data generation and storing, reasoning and analysis of the crime dictionary, information based on the concept dictionary, taxonomy, crime type information, and analysis model necessary for crime analysis.
  2. Application of the Big Data Suite product
    Applies big data storing and the search engine (DISCOVERY) for an intelligent semantic search service and unstructured analysis engine (TMS) and cognitive analysis engine (CAS) for collected unstructured data analysis among the Big Data Suite products.
  3. Application of Apache SPARK, STORM, KAFKA, etc.
    Configures the intelligent analysis platform that guarantees the best quality through the linkage with the GraphDB Suite product and the Big Data Suite products by reflecting OpenSource Apache SPARK, STORM and KAFKA in the intelligent crime prevention analysis system.
③ Validity of product selection
  1. Ontology-based real-time analysis and the visualization of relationships between objects are necessary.
  2. Visualization of scientific investigations, criminal policies, and crime prevention-related analysis results is necessary.
  3. Configuration of crime taxonomy specializing in Korean crimes is necessary.

Establishment of a graph data integration platform.

Establishment of an intelligent KMS system – NongHyup Bank

This knowledge management system to manage and share necessary business knowledge for customer consultation was improved in the Customer Happiness Center and all business branches of the NongHyup Bank. The convenience of functions such as knowledge generation, management and search was improved and when knowledge content is generated, such content is managed specifically according to particular fields of business and properties and these contents are saved after being converted into a Knowledge-Graph based knowledge base. This converts natural language content into knowledge data in a machine-readable format that can be comprehended by the system. AI-based automatic knowledge extraction is also applied as part of the process.


< Intelligent KMS system UI/UX >


① Business Content

The previous knowledge management system was a system designed to produce, save and manage business knowledge for customer consultation and was used for the purpose of sharing knowledge between bank consultants and/or branch employees. However, there is a continuously growing demand of an AI-based service since bank customers now use various digital channels more frequently, and it is necessary to establish and manage AI-based knowledge data separately for an AI service.
The previous knowledge contents and AI-based knowledge data take different shapes and structures, so it was difficult to maintain the same information if they were managed separately, and duplicated knowledge generation, management and use causes inconvenience to users in many ways.
The aim is to improve the convenience in management and sharing of consultation knowledge and information between employees by reforming the old knowledge management system, in use for ten years or more, into an intelligent knowledge management system that corresponds to the latest AI technology trends, and also to improve management efficiency and utilization in regard to such knowledge by creating and managing knowledge data that can be used by AI-based services.


< Sumary of intelligent KMS system >

② Applied technologies and solutions.
  1. Knowledge graph-based knowledge data generation and management
    An intelligent knowledge management system can create knowledge content and generate and manage knowledge data in a machine-readable format that can be comprehended by the system, in line with the knowledge graph structure. Such a knowledge base can be used for various AI-based services.
  2. Automatic knowledge extraction from unstructured text.
    Knowledge contents are prepared as a sentence or a paragraph in a format (TEXT or HTML) which is easily comprehended by users. The KENT (Knowledge Extraction from Natural Language Text) technology automatically extracts and converts knowledge data from such unstructured text based on a learned model.
③ Validity of Product Selection
  1. The selection of the repository for storing knowledge graphs that have the best quality and performance in the country.
  2. The selection of a machine learning-based knowledge extraction solution for extracting knowledge from an unstructured document.
④ Main Performance Points
  1. Linkage between NongHyup Bank and NongHyup Card’s intelligent knowledge management system and the Q&A system.
    Intelligent knowledge management systems were established separately for NongHyup Bank and NongHyup Card with the established systems linked to enable management of separate contents and sharing of knowledge. Also, an environment to deliver knowledge data to employees, and the system at the same time through the linkage with the AI Q&A system serviced by the consultation advisers in the Call Center was established.
  2. Commercialization of automatic knowledge extraction technology.
    Automatic knowledge extraction is a technical field that still needs a lot of R&D effort. Various measures, including language processing and machine learning, were researched in search of quality improvement, obtaining satisfactory results within a given range in the extraction of required knowledge from the contents held by NongHyup Bank. It can be considered a successful case of the application of automatic knowledge extraction to an actual service.
Safe Food Q&A System – Korea Agency of HACCP Accreditation and Services
① Business Content
  1. It aims at “the development of AI-based personalized knowledge consultation service for HACCP intelligent information service” that can establish a knowledge base based on close HACCP certified safe food using raw materials and nutritional content data and provide a theme-based personalized safe food recommendation service.

< Block diagram of safe food Q&A system goal >


< Safe food Q&A UI/UX >

② Applied technologies and solutions.
  1. Q&A Manager: Provides user query processing and an interface (Open API)
  2. KBQA: FBQA & TBQA &Plugin
  • FBQA: Fact-based query processing
  • TBQA: Template-based query processing function, processing of a complicated or complex query
  • Plugin: Query processing for data, such as weather, news and world time, etc. changed in real time.
  1. IRQA: Finds another question similar to the question asked through text analysis and provides a ranked answer.
  2. KB(Ontology): Open domain knowledge base necessary for query processing (OWL ontology: RDF + Axioms)
  3. NLU: Natural language query analysis and pattern recognition.
  4. NLG: Creates a natural language query to obtain a query processing result.
  5. Analysis dictionary management: Manages resources necessary for query analysis and query patterns.
  6. Q&A management: KB management, query testing, Q&A evaluation, KBQA analysis dictionary management.
③ Validity of Product Selection
  1. Uses the repository for storing a knowledge graph.
  2. Uses the machine learning solution for extracting knowledge from an unstructured document.
Main Performance Points
  1. Possible to provide theme-based customized food information to users.
  2. Reinforces the accessibility for information vulnerable people and improves the right to know safe food.
  3. Reduces time and saves costs to obtain information regarding intelligent safe food service.
  4. Expected cost savings in terms of civil consultation and technical support of intelligent HACCP consultation service

Establishment of a linked data (LOD) service

Development of a customized IP-Biz information sharing platform – Korean Intellectual Property Office
① Business Content

An information channel that can be used in linkage with various business information possessed by other government departments and public institutions based on the intellectual property right information possessed by Korean Intellectual Property Office was established, and convenient functions to allow employees in small and medium enterprises and ·start-up companies to obtain information promptly and conveniently from the user’s viewpoint was provided.

  1. Business information possessed by other government departments and public institutions is linked.
  2. Converges intellectual property right information possessed by Korean Intellectual Property Office and business information and provides customized information.
  3. Provides customized reports based on established intellectual property rights information and business information.

< Block diagram of customized IP-Biz information sharing platform >

② Applied technologies and solutions.
  1. Extraction of text mining keyword
  2. Trend analysis and TopN analysis
  3. Application of knowledge graph for correlation analysis
③ Validity of Product Selection
  1. Uses the repository for storing a knowledge graph.
  2. Uses the machine learning solution for extracting knowledge from an unstructured document.
  3. Text mining and time series analysis technology solution for trend analysis.

< Patent trend/technology/market analysis service UI/UX >

Main Performance Points
  1. Opens more than five million cases of domestic and overseas patent information.
  2. Provides three million cases of patent analysis DB linkage information from thirteen institutions.
  3. Provides analysis service for each product type.
  4. Provides a customized analysis report for a company.