@seandalai

Improving search engine retrieval using a compound splitter for Swedish

. 15th Nordic Conference of Computational Linguistics, (2005)

Abstract

In this paper we have investigated 128 high frequent Swedish compound queries (6.2 per thousand) with no search results among 1.6 million searches carried out at nine public web sites containing all together 100,000 web pages in Swedish. To these compound queries we added a compound splitter as a pre-processor and we found that after decompounding these queries they gave relevant results in 64 percent of the cases instead of zero percent hits. We give also examples on some rules for optimal compound splitting in a search situation.

Links and resources

Tags

community

  • @dblp
  • @seandalai
@seandalai's tags highlighted