Globala och lokala sökmotorer
en utvärdering av Google, MSNSearch och Svesök
The purpose of this thesis is to evaluate three search engines with regard to retrieval effectiveness of information specific to Swedish users. Coverage and the precision in result descriptions are also examined. The three selected search engines are Google, MSN Search and Svesök. 30 queries were tested against the search engines and the first 20 result descriptions and results were judged for relevancy. Relevancy was judged by a binary scale. Micro- and Macro-precision were used to measure the retrieval effectiveness of the search engines. The number of duplicate and inactive links for each search engine was recorded. Precision in result descriptions was used to measure the accuracy of the links. That is if a link was judged as relevant/not relevant it should ideally lead to a relevant page/not relevant page. Coverage was measured by calculating the total amount of retrieved hits for each search engine. The results show that Google reached the highest precision values. The results also show that about half of the links judged as relevant did not lead to relevant pages. That Google also had the best coverage. The conclusions I draw in this paper is that Google is the best choice of search engine for searching and finding relevant information specific to Swedish users. Global search engines are better at finding this information than local ones.