This project is about the investigation of a relatively new implementation of finite automata using hardcoding. In hardcoding, the transition function of the automaton is directly implemented using instructions that form part of the string recognizer algorithm as opposed to the traditional table-driven method whereby the transition function is represented in form of a table. The project consists of:
- A general empirical performance evaluation and comparison of both hardcoded and table-driven implementations.
- A toolkit that directly generates hardcoded finite automata based on the results of the above step.
- A taxonomy based performance analysis of hardcoded finite automata (and therefore pattern matchers) using the new taxonomy of keyword pattern matching algorithms, by Loek Cleophas, Gerard Zwaan and Bruce Watson.
- A general toolkit based on the above taxonomy following the same approaches used for SPARE Time.
|