H. Jung, C.-N. Seon, J.C. Sohn, W.-K. Sung, and D.-I. Park (Korea)
Query Analysis, Information Extraction, Lexico-Semantic Pattern, Semantic Web, Ubiquitous Robotics Companion, Web Services
Information extraction is an appropriate methodology to analyze purposeful queries which require information or actions. We apply information extraction to find meaningful arguments and a predicate from queries on weather, traffic, and home automation for an intelligent robot service, and to generate a robot-understandable query structure. To deal with various ambiguities and flexible utterances of arguments, we separate extraction rules into non-context-based A-extraction rules and context-based revision rules. These extraction rules select and modify arguments by matching the input query with lexico-semantic patterns. We additionally add "linguistic information detection" to find conditional clause and modalities, "predicate extraction" to acquire query meaning type, and "query structure generation" to construct query-related information such as a query type, concepts, and auxiliary concepts in a query. Currently, we involve the project of Ubiquitous Robotics Companion technology development as a national project, and choose the above three domains as a test environment to experiment the adaptability of information extraction.
Important Links:
Go Back
