| With the development and popularization of the Internet,the number of Internet users is increasing rapidly.The Internet will produce a large number of text resources every day,which contain rich information,including important geographic information.Analyzing and mining the relationship between text resources and geographic information is one of the hottest research topics in science area.The traditional Internet is integrated based on Uniform Resource Location(url).However,extracting location information from text in web pages and reintegrating them based on location are of great significance.The idea of mapping text pages with geographical locations can help us to innovate the location service,to establish and develop a new generation of location service platform.In this paper,we proposed a method to extract geographic information from text in web pages.In this method,we first obtain all the place names mentioned in the text.Then we calculate an important score for each place name by referring to the nodes and relationships in a knowledge graph.Finally,according to the important score,the place names which are most related to the text are selected as the result of geographic information extraction.This paper mainly contains two aspect of work.First of all,we proposed a method to predict the location of web page text.By reference the relationships between geographic entities or relationships between geographic entities and non-geographic entities,the relationships between place names referred in the text and key content of the text can be inferred out.This can help us to judge which is the most relevant place name of the text content.We also did experiments on a text dataset and compared the predict results with expect results,which proves that the method in this paper did well in predicting on street level granularity.Secondly,we designed and implemented a text content location prediction system according to the method above and tested the function of the system. |
PDF Full Text Request