AbstractWritten text is considered as one of the oldest methods to represent knowledge. A text can be defined as a logical and consistent sequence of symbols which encodes information in a certain language. A straightforward example are natural languages, which are typically used by humans to communicate in spoken or written form.
Other underlying examples are DNA, RNA and proteins sequences; DNA and RNA are nucleic acids that carry the genetic instructions, specifies the sequence of the amino acids within proteins, regulate the development and functionality of living organisms specifies the sequence of the amino acids within proteins. Proteins are molecules consisting of one or more chains of amino acids participate in virtually every process within cells.
DNA and RNA can be represented as sequences of the nucleo-bases of their nucleotides and proteins and can be represented by the sequence of amino acids encoded in the corresponding gene. A natural problem which emerges when processing such sequences is determine weather a specific patterns occur within another string (known as exact string matching problem); as far as natural language texts are concerned, an important problem in computational linguistics is finding the occurrences of a given word or sentence in a volume of text; Similarly, in computational biology identifying given features in DNA sequences is a important of great significance, on the other side, one is often interested in quantifying the likelihood that two pairs of strings have the same
underlying features based on explicit similarity/dissimilarity measurement (known as approximate string matching). Both instance of the string matching problem have been studied thoroughly since early 1960s.
This thesis contributes several efficient novel and derived solutions (algorithms
and/or data structures), for complex problems which have been originated either out of theoretical considerations or practical problems, and study their experimental performance and compare the proposed solutions with some existing solutions.
Among the latter originated introduced solution several ones motivated by realworld problems in the fields of molecular biology and computational linguistics.
Despite the fact that studied problems and their proposed solutions differs in research motivation paradigm, yet still utilise similar tools and methodologies for solving the corresponding problems. For example the seminal “Aho-Corasick” Automaton is employed for finding a set of motifs in a biological sequence and detecting spelling mistakes in Arabic text. Similarly, employing the bit-masking trick to extend the DNA symbols to accelerate equivalency testing of degenerate characters in the same way to extend the Arabic alphabet to measure similarity between a stem and derived/inflected forms a given word.
|Date of Award||2015|
|Supervisor||Costas Iliopoulos (Supervisor) & Max Crochemore (Supervisor)|