Summarizers – Similarities, Differences and Application in the Lithuanian Language Cover Image

Skaitmenizuotų tekstų santraukų rengimo sistemos: panašumai, skirtumai ir taikymas lietuvių kalbai
Summarizers – Similarities, Differences and Application in the Lithuanian Language

Author(s): Jurgita Lasytė, Bronius Tamulynas
Subject(s): Language and Literature Studies
Published by: Kauno Technologijos Universitetas
Keywords: kompiuterinė lingvistika; automatinis santraukų rengimas; teksto glaudinimas; santraukos įskaitomumas; santraukos prasmingumas

Summary/Abstract: Summarizer is a useful system, which allows compressing text and represents it in a shorter way. In Lithuania summarizers are new technological developments and information about them is scarce, although these systems are created and developed by the specialists of computational linguistics in other countries. In the research English and Lithuanian texts were used, which allowed seeing differences among output summaries. In the English texts, summaries using all summarizers were quite informative, useful and did not miss the main information from the original texts, except cases when a high compression rate was used. While summarizing Lithuanian texts, some summarizers confront with language recognition problem – the output summaries in Lithuanian are with unrecognizable symbols, or in individual cases some symbols are missing. Summaries in the Lithuanian language are not of the same quality as the summaries in English. It was also noticed, that Lithuanian summaries with higher compression rate are loosing the meaning of the original text. To conclude, it was noticed that commercial summarizers give better summaries in the English language, but in the Lithuanian language better summaries are given by noncommercial summarizers. Furthermore, summaries received using high compression rate are loosing their information and meaning. Analyzed summarizers are best fitted for the English language, therefore using summarizer for other language than English stipulates various problems – from language recognition to rendering of text meaning. It is very important to adapt different languages to summarizers, as this would significantly improve summarizers. Findings of this paper can be used in creating a Lithuanian summarizer or improving other summarizers.

  • Issue Year: 2010
  • Issue No: 17
  • Page Range: 46-52
  • Page Count: 7
  • Language: Lithuanian