pre-CODIE-Crosslingual On-Demand Information Extraction
暂无分享,去创建一个
Our research addresses two central issues of information extraction --- portability and multilinguality. We are creating information extraction (IE) systems that take foreign-language input and generate English tables of extracted information, and that can be easily adapted to new extraction tasks. We want to minimize the human intervention required for customization to a new scenario (type of facts or events of interest), and allow the user to interact with the system entirely in English. As a prototype, we have developed the pre-CODIE system, an experimental Crosss-lingual On-Demand Information Extraction system that extracts facts or events of interest from Japanese source text without requiring user knowledge of Japanese.
[1] Ralph Grishman,et al. Automatic Pattern Acquisition for Japanese Information Extraction , 2001, HLT.