Extracting Landslide Disaster Information From Web Pages

Due to serious threats of landslide disasters, it is urgent to know how to obtain adequate, accurate and timely information on landslide hazards for both government and researchers. Therefore this topic is of great significance for study of landslide disaster and risk reducing. As a giant database, internet can be used to extract landslide disaster data temporally and spatially, and this way of extracting has the advantage of quickness, effectiveness and briefness. According to characteristics of landslide disaster information in Web texts, extracting method of landslide disaster information in Web text is proposed by analyzing extraction techniques of places names, time and attributes. Then an experimental verification is performed in an original corresponding prototype system which is developed by author. Main three research methods and results are as follows:(1) Extracting effective landslide disaster information from web texts:disaster information retrieval method is designed which achieve the extraction of the landslide disaster topic page effectively by using search engines and news pages; then the method of removing duplicates in landslide disaster information is proposed by analyzing structural characteristics of landslide disaster topic information and regularity of Internet information; finally use regular expressions and HTMLParse methods to extract effective landslide disaster information.(2) Classification and the corresponding extraction skill of landslide disaster information: first landslide disaster information are divided into categories of temporal, geographical and attributive information respectively by using text block and word segmentation techniques. Then different extraction methods are designed according to different types of information.(3) Landslide disaster system implementation and experiment verification:On the basis of those technical researches, landslide disaster information extraction system is developed by using.Net development platform and Html technology. Various functions in this system are achieved which contains landslide disaster information extraction, extraction rule library management, landslide information display spatially in maps and so on. Then take landslide disaster information in Sichuan province as the sample to conduct an experiment verification. This study indicate that the text data in the Internet can be used effectively to extract spatial and temporal landslide disaster information data. It is an effective auxiliary way to find useful landslide disaster information. Due to the complexity of disaster information text content, there are some limitations by using the rules and statistical approach of manual sorting to extract landslide disaster information. In addition, text disasters information in internet have the restriction of describing indirectly. These information are fuzzy and uncertain at certain extent, therefore this method need to be combined with other landslide disaster information extraction method for further data integration.