Design of a Spatially Aware Amharic Web Content Retrieval
No Thumbnail Available
Date
2018-06-04
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Addis Ababa University
Abstract
It is obvious that we are living in the information era. WWW has been central to the development of the information age and is the primary tool billions of people use to interact on the Internet. It fundamentally changed how we connect with each other. Location information is a kind of information people are interested in. The analysis and use of geographical content from web resources is currently an area of increased interest and research. Almost everything that we do can be regarded as having some form of geographical context. Over the past few decades, accessing geographical information has focused on the combination of digital maps and databases that characterise the majority of geographic information systems. In geographic information systems, geographic objects are generalized into geometric points, lines and polygons.
The objective of this work is to design a model which uses Amharic geo-ontology for the development of a spatially-aware Amharic web content retrieval. The model receives and parses the user request in order to identify the existence of spatial features, spatial relationships to discover the exact location phrases. The Amharic geo-ontology maintains a place name, the geographical footprints which indicate its spatial extent, and its topological relationships with other places. It plays a key role in the process of geo-parsing of web documents and generation of spatial indexes. The geo-parsing process is identifying the presence of place names and spatial relationships in the document. While the geo-coding process involves disambiguating place names with multiple spatial references. Once significant place names have been detected in a document, the geographical ontology will be used to provide footprints to the document. A spatial index of web documents will be created based on the footprints. The spatial relevance ranking is based on measures of distance between the query footprint and the document footprint.
A prototype application is developed to evaluate the proposed solution. A query is formulated in three different formats, which are: <Place Name >, <A Place Name, Spatial Relationship, Geographic Feature Type>, and <Geographic Feature Type>. The evaluation is performed using 275 known place names, 38 cities in Ethiopia, and 35 geographical feature types.
Description
Keywords
Geo-Ontology, Place Name, Geographic Feature, Geographic Feature-Type, Spatial Relationship, Geo-Coding, Geo-Parsing, Spatial Indexing