Binary-Coded Ternary Number Representation in Natural Language Text Compression

Efficient Integer Retrieving from Unordered Compressed Sequences

“…can be either 'true' or 'false' with high probability) and this slows down the algorithm on modern processors. However, we can avoid performing conditional statements by exploiting an idea similar to the fast decoding of a binary-encoded ternary code given in [15]. We use this approach in Algorithm 2 described in the next Section, which is a part of general integer retrieving Algorithm 1.…”

“…• When we found the leftmost byte of a required codeword, decode it using a fast byte-aligned decoding technique, e.g. as discussed in [15]. Note that in general the j-th L2-block position can be approximated by the formula kj + b, where parameters k and b are calculated with the ordinary least squares technique.…”

“…• When we found the leftmost byte of a required codeword, decode it using a fast byte-aligned decoding technique, e.g. as discussed in [15].…”

“…2. Its idea resembles the fast decoding method given in [15]. Algorithm 2: Function Decode number(i, s) -decoding the (s + 1)-th codeword starting from the right edge of the i-th byte of a code (little-endian machine)…”