| Text Mining has been the focus of the academic research in the field of Data Mining.Users can mine more meanings behind the words, which will provide high-qualityinformation, by doing text filtering, text classification, text prediction, etc. Also, as thegeographic information is mostly show or stores in text mode, in recent years, thecombination use of text mining and spatial databases has gradually becomes one of thehottest research field of Data Mining. To make the location-based applications meet users’requirements, or to make the query in spatio-temporal databases accurately, it is necessaryto construct the geographic properties properly. However, for unstructured data such asnews reports or section of text, in most cases users can only get the names of the places,and the properties of them are unable to be obtained. Therefore, users can only judge andcomprehend according to their own understanding. As one place name might correspond tomultiple regions, for users, it is likely to cause comprehension-ambiguity, thus affectingthe correctness of judgment. So, to preprocess the place name by using geotagging methodand analyze the ambiguity of them is particularly important in data processing inspatio-temporal databases.Through the fusion of multiple algorithms and functions to construct, store, judge andprocess the name attributes for different geographical entities. Returns the most likelyoutcomes to users by recognizing and analyzing the geographic information appear inunstructured data to remove ambiguity according to query in local databases and innetwork. In the same time, optimize the process of data processing to improve queryefficiency and reduce resource overheads. Experimental results show that the algorithmand systems raised in this thesis can recognize place names accurately in texts, can judgeand remove the names with geographical ambiguity, and returns accurate results for systemusers. |
PDF Full Text Request