An Efficient Compression Scheme for Large Natural Language Text

Abstract

Data compression is the route towards adjusting, encoding or changing the bit structure of information so that it requires less space. Data compression is a decrease in the quantity of bits expected to demonstrate the data. Compacting data can spare stockpiling limit, accelerate record exchange, and lessening costs for capacity equipment and system transfer speed. Data compression covers a huge space of jobs including data correspondence, data putting away and database improvement. In the same way, Text compression can be as straightforward as expelling every unneeded character, embedding a solitary recurrent character to demonstrate a string of rehashed characters and substituting a little piece string for a habitually happening bit string. The fundamental standard behind compression is to build up a strategy or convention for utilizing less bits to express the actual data. Character encoding is fairly identified with data compression which represents a character by a type of encoding system. In this thesis, an efficient and simple compression algorithm for large natural text named n-Sequence based m Bit Compression (nSmBC) is proposed which can able to beat WinZip and WinRAR in terms of compression ratio. WinZip and WinRAR are two well-known compression techniques used for text compression in the industry. The scheme provides an efficient encoding algorithm that converts an 8 bit character by 5 bits utilizing a look up table. The look up table is produced by using Zipf’s distribution which is a discrete distribution of commonly used characters in different languages. 8 bit characters are converted to 5 bits by partitioning the characters into 7 sets. After converting the characters into 5 bit, an n-sequence scheme is developed to logically calculate the location number of a particular combination of characters. The reverse algorithm to recover the actual input is further demonstrated. The algorithm is finally compared with the well-known WinZip, WinRAR, Huffman and LZW techniques. Promising performance is demonstrated both by theoretical and experimental analysis.