Research task title

WiseKB : Developing self-learning knowledge-base and reasoning technology based on big data understanding

Research Overview

Development of an autonomous learning-type knowledge base and hybrid reasoning technology through big data understanding and implementation of the evolutionary knowledge augmented service platform which allows knowledge-augmented global service

The technical characteristics of the WiseKB platform are as follows :
  • Knowledge learning for expert knowledge and knowledge augmentation based on complex reasonin
  • Semantic analysis based on language processing technology and semantic reasoning technology
  • Management of the knowledge augmentation life cycle on the basis of dual spiral methodology
  • Semantic integration and validation of expert knowledge through automatic transformation of expert knowledge and knowledge curation
  • Expansion of common-sense knowledge through crowdsourcing
Inclusion of a system for semantically integrating and managing internal and external knowledge resources including formal and⋅informal big data collected from the website, sensor data collected from the terminal and the network, and core knowledge secured by the proposing organization, with legacy data in each application domain
WiseKB implements an autonomous collaboration architecture based on a large-scale dispersed and parallel treatment infrastructure and includes engines that allow multi-modal knowledge expression, collection/management of large-scale multiple domain expert knowledge, integration/management of knowledge resources, large-volume hybrid reasoning, and autonomous knowledge learning from large-scale expert knowledge.
The WiseKB platform product includes the knowledge base and reasoning functions developed in detailed task 2 and is directly connected to WiseQA in detailed task 1 to allow expert-level deep question answering through natural language understanding.
Ultimately, the WiseKB platform product promote innovation in the knowledge industry throughout the nation by fostering promotion of a creative economy based on the knowledge industry, optimizing national competition strategies, creating knowledge industry-related jobs, etc., and helps to ensure future national competitiveness.
In Step 2, knowledge resources and big data in expert fields such as finance, patents, etc. are to be collected, the multiple domain expert knowledge augmentation service platform is to be implemented by expanding the knowledge base through autonomous learning and validation and improvement of the reasoning system, and the base for expanding the multi-language service is to be provided by collecting and integrating English knowledge resources and constructing the appropriate knowledge base.

<Dual spiral methodology>

Research Goals


Knowledge resources and big data in expert fields such as finance, patents, etc. are to be collected, the multiple domain expert knowledge augmentation service platform is to be implemented by expanding the knowledge base through autonomous learning and validation and improvement of the reasoning system, and the base for expanding the multi-language service is to be provided by collecting and integrating English knowledge resources and constructing a knowledge base.
Construction of a knowledge base in expert fields and commercialization of knowledge augmentation technology for AI consulting
  • Commercialization of the technology for constructing a domain-specific professional knowledge base such as finance, patents, etc.
  • Commercialization of an AI consulting system utilizing the knowledge learning and reasoning technology developed in Step 1
Improvement of QA technology through the knowledge base and implementation of the intelligent consulting platform
  • Commercialization of technology of understanding a user’s question through knowledge base and automatic simple protocol and RDF query language (SPARQL) generation
  • Development of deep learning-based automatic consulting technology for implementation of an intelligent consulting platform
Collection and integration of multilingual knowledge resources and research in the technologies of construction and augmenting the English knowledge base
  • Research into augmenting Korean language knowledge utilizing multilingual knowledge resources
  • Collection of English knowledge resources and development of knowledge-based construction technology

Work Package

Knowledge Base
  • Expertise : Development of technology for KB construction and management for professional domains such as finance, etc.
  • Automation : Automation of new knowledge learning/updates and conflict resolution in near real–time
  • Multi-language : Utilization of English knowledge resources and expansion of multi-languages in the knowledge base
Complex Reasoning
  • Common sense reasoning : Time and default common sense reasoning, non-monotonic common sense distributed reasoning
  • Causal relationship : Complex/common sense reasoning, causal relationship reasoning for a SPARQL result
  • Analogical reasoning : Large-volume knowledge-distributed analogical reasoning based on reliability
Knowledge Learning
  • High precision : Commercial-level knowledge learning algorithms with high accuracy and a high recall rate
  • Complexity : Improvement of the high-quality knowledge learning algorithm for complex sentences
  • Openness : Generation and summarization of a large-volume knowledge graph for open QA
Question Answering
  • Semantic understanding : Improvement of SPARQL automatic transformation technology, based on semantic understanding
  • Situation adaptation : Generation of interactive natural language and development of situation-based ranking technology
  • deep learning : Development of deep learning-based KBQA automation technology