GIR'05 2005 ACM workshop on geographical information retrieval

  • Authors:
  • Christopher B. Jones;Ross Purves

  • Affiliations:
  • Cardiff University;University of Zurich

  • Venue:
  • ACM SIGIR Forum
  • Year:
  • 2006

Quantified Score

Hi-index 0.00

Visualization

Abstract

Geographical information retrieval (GIR) is concerned with the problems of finding information resources that relate to particular geographical locations. Until recently most web search engines have treated geographical terminology within user queries in the same way as other terminology. This can often result in failure to find relevant documents and in the retrieval of irrelevant documents. There are several reasons for this. For example, there are many different places with the same name, so unless the query contains a unique set of geographical terms, documents referring to the wrong place may be retrieved. There are many uses of place names in the names of people and organisations with the consequence that search depending on naive exact matching of terms may retrieve such documents even though they may not relate to the named place. It is also the case that queries may include spatial prepositions, such as near or outside, which will not be interpreted intelligently by the search engine. Thus relevant documents may not be found because the user's query did not contain explicit references to the places that match the query expression, in the sense of being, for example, near, inside or north of a named place. It is also the case that conventional search engines lack other desirable facilities for geographical search such as relevance ranking that takes account of geographical proximity and the ability to use interactive maps to specify a query and to visualise the results.