Original title: Syllable-Based Compression
Translated title: Syllable-Based Compression
Authors: Lánský, Jan ; Pokorný, Jaroslav (advisor)
Document type: Rigorous theses
Year: 2010
Language: cze
Abstract: Classic textual compression methods work over the alphabet of characters or alphabet of words. For languages with rich morphology as well as for compression of smaller files it can be advantageous to use an alphabet of syllables. For some compression methods like the ones based on Burrows-Wheeler transformation the syllable is a reasonable solution also for large files - even for languages having quite simple morphology. Although the main goal of our research is the compression over the alphabet of syllables, all implemented methods can compress also over the alphabet of words. For small files we use the LZW method and Huffman coding. These methods were improved by the use of initialized dictionary containing characteristic syllables specific for given language. For the compression of very large files we implemented the project XBW allowing combination of compression methods BWT, MTF, RLE, PPM, LZC, and LZSS. We have also tried to compress XML files that are not well-formed. When compressing over a large alphabet, it is necessary to compress also the used alphabet. We have proposed two solutions. The first one works well especially for small documents. We initialize the compression method with a set of characteristic syllables whereas other syllables are coded when necessary character by character. The...

Institution: Charles University Faculties (theses) (web)
Document availability information: Available in the Charles University Digital Repository.
Original record: http://hdl.handle.net/20.500.11956/24515

Permalink: http://www.nusl.cz/ntk/nusl-441329


The record appears in these collections:
Universities and colleges > Public universities > Charles University > Charles University Faculties (theses)
Academic theses (ETDs) > Rigorous theses
 Record created 2021-05-30, last modified 2024-01-26


No fulltext
  • Export as DC, NUŠL, RIS
  • Share