Improving Precision in Information Retrieval for Swedish using Stemming

We will in this paper present an evaluation of how much stemming improves precision in information retrieval for Swedish texts. To perform this, we built an information retrieval tool with optional stemming and created a tagged corpus in Swedish. We know that stemming in information retrieval for English, Dutch and Slovenian gives better precision the more inflecting the language is, but precision depends also on query length and document length. Our final results were that stemming improved both precision and recall with 15 respectively 18 percent for Swedish texts having an average length of 181 words.