Information Extraction
We investigate new methods for extracting information from texts at various levels of granularity (named entities, domain templates, general relations). Research interests include the combination of different shallow and deep methods, as well as machine learning approaches for improving information extraction performance. Domains include German business news from two leading newspapers, and Chinese sports news. There is close cooperation with the projects Whiteboard at the DFKI and Deep Thought at the Dept. of Comp. Linguistics & Phonetics.
|