CRPP - Conference Papers

In this paper, we attempt to give spatial semantics to web pages by assigning them place names. The entire assignment task is divided into three sub-problems, namely place name extraction, place name disambiguation and place name as- signment. We propose our approaches to address these sub- problems. In particular, we have modified GATE, a well- known named entity extraction software, to perform place name extraction using a US Census gazetteer. A rule-based place name disambiguation method and a place name as- signment method capable of assigning place names to web page segments have also been proposed. We have evalu- ated our proposed disambiguation and assignment methods on a web page collection referenced by the DLESE meta- data collection. The results returned by our methods are compared with manually disambiguated place names and place name assignment. It is shown that our proposed place name disambiguation method works well for geo/geo ambi- guities. The preliminary results of our place name assign- ment method indicate promising results given the existence of geo/non-geo ambiguities among place names.