Download now Free registration required
The paper describes automatic summarization of the XML documents in Croatian language. The goal of the summarizer is to generate extracts with high percent of extract-worthiness. The research shows that extracts generated using the algorithm is well formed, but it also shows that algorithm is very domain dependant. The results of the evaluation process proved that the technique of identifying cue phrases and bonus/stigma words in the training corpus significantly improves the text summarization for Croatian language.
- Format: PDF
- Size: 672.2 KB