Text

• ASCII
  • Just a table
  • 7 bits plus a check bit.
  • Really "english" only
  • Very much a standard
• Unicode
  • Allows for representation of different characters (alphabets)
  • another chart
  • 16 bits (2¹⁶ characters, 65,536)
• Text Compression
  • Keyword Encoding
    • The text

      	      The sum of the parts is greater than the whole.
      	   

    • Becomes

      	      The sum of + parts is greater than + whole.
      	   

    • Compression ratio: 43/47 = .91
    • Very simple idea, replace common words with a symbol
    • the is replaced with +
    • and is replaced with -
    • don't replace A, why?
    • Extra symbols limit effectiveness
    • Can't use some (most) symbols (We need $)
    • The is not the same as the, and it is not encoded
    • Efficiency Depends on text being sent.
  • Run Length Encoding
    • If a character is repeated more than three times, replace it with a code
    • aaaaa => *5a
    • * tells us that the encoding has started
    • 5 tells us how many times the character repeates
    • a is the repeated character
    • Not very efficient with english, but used in other places (images, data)
    • The text

      	    aaaaabbaaaaaaaababbbbbbabbbabbbb
      	  

    • Becomes

      	    *5abb*8aba*6babbba*4b
      	  

    • A compression ratio of 21/32 = .66
  • Huffman Codes
    • Build a variable length code based on the frequency of use
    • e, a, s, t get few bits
    • q, z, x get many bits
    • Need to build a unique code, but that is not hard.
    • For the alphabet

      Letter	Frequency	Code
      a	.25	10
      b	.25	01
      c	.26	11
      d	.14	001
      e	.10	000

    • The message abcacad encodes 10 01 11 10 11 10 001
    • Or could encode 000 001 010 000 010 000 011
    • A compression ratio of 15/21 = .71
• Come up with an encoding for

      Theirs was not to make reply,
      Theirs was not to reason why,
      Theirs was but to do and die:
      Into the valley of Death
        Rode the six hundred.