Resolution of Geographical String Name through Spatio-Temporal Information

Autor:	Mauro Pedone, Pedro Chahuara, Aris Tsois, Luca Mazzola
Rok vydání:	2014
Předmět:	Data processing Information retrieval Computer science Location Description Bayesian network Leverage (statistics) Contextual information Data mining String metric computer.software_genre Location Temporal information computer
Zdroj:	Machine Learning and Data Mining in Pattern Recognition ISBN: 9783319089782 MLDM
DOI:	10.1007/978-3-319-08979-9_38
Popis:	Entity resolution of geographical references is essential for the analysis of spatial temporal information when data come from heterogeneous sources. In the ConTraffic project we work on the analysis of trajectory of moving objects (e.g. commercial containers and vessels) and, as part of data processing step, the right location for textual references must be determined. In this paper we present an application of Bayesian networks that leverage the information of already resolved references in order to estimate the right entity corresponding to a geographical location. Contextual information of objects that have followed similar trajectories is used as well. Our approach is suitable to perform entity resolution efficiently even when the database contains millions of movements. The results we obtained prove that our method is useful in cases where string similarity methods are unable to provide a solution.
Databáze:	OpenAIRE
Externí odkaz:	https://explore.openaire.eu/search/publication?articleId=doi_________::dce0d77def88dbae2021387167edd486 https://doi.org/10.1007/978-3-319-08979-9_38 Zobrazit plný text záznamu