Information recognition from Chinese documents is an important task of typhoon disaster database foundation. Disaster metadata has the Place-Event structure, which has diversity in itself. Meanwhile, in the system of paragraphs and sentences of information sources, this structure also reflects the high clustering, as well as a degree of granularity. According to these characteristics, this article constructs a rule base correspondingly on the Event classification, then, it rewrites a regular expression matching procedure to extract the disaster information in the Event. It also establishes a place name database for Place recognition, but in this part it utilizes the hierarchy feature of place. Evaluation on test shows a reliable recall and precision rates of the technology.