- Keywords: Knowledge base, information extraction, entity match, knowledge base refinement
- Abstract: Knowledge bases play crucial roles in a wide variety of information systems, such as search engines and intelligent personal assistants. For responding constantly fluctuating user information demands, we aim to construct a large-scale and well-structured global knowledge base from the world’s evolving data. In this paper, we discuss enterprise-specific issues with knowledge base construction and present how to deal with these issues in our construction system called “MERMAID.” To maintain the quality of our knowledge base at the production-level, MERMAID is carefully designed to incorporate various automatic and manual validation methods. We partly leverage manual validation methods to deal with business requirements and user feedbacks quickly since it is difficult to filter out all incorrect facts automatically in practice. Moreover, we propose a novel information extraction method that obtains reliable factual information from Web-crawled data on the basis of distant supervision. Our constructed knowledge base is already utilized in real-world Japanese Web services, and the number of entities in it keeps growing steadily.
- Archival status: Non-Archival
- Subject areas: Information Extraction, Databases, Knowledge Representation, Semantic Web