A framework for benchmarking FA-based string recognizers

  • Authors:

  • Ernest Ketcha Ngassam;Derrick G. Kourie;Bruce W. Watson

  • Affiliations:

  • University of South Africa, Pretoria;University of Pretoria, Pretoria;University of Pretoria, Pretoria

  • Venue:
  • SAICSIT '10 Proceedings of the 2010 Annual Research Conference of the South African Institute of Computer Scientists and Information Technologists
  • Year:
  • 2010

Quantified Score

Hi-index 0.00

Visualization

Abstract

Previous work on implementations of FA-based string recognizers suggested a range of implementation strategies (and therefore, algorithms) aiming at improving their performance for fast string recognition. However, an efficient exploitation of suggested algorithms by domain-specific FA-implementers requires prior knowledge of the behaviour (performance-wise) of each algorithm in order to make an informed choice. We propose a unified framework for frequently evaluating existing FA-based string recognizers such that FA-implementers could capture appropriate problem domains that guarantee an optimal performance of available recognizers. The suggested framework takes into consideration factors such as the kind of automaton being processed, the string and alphabet size as well as the overall behaviour of the automaton at run-time. It also forms the basis for further work on FA-based string recognition applications in specific computational domains such as natural language processing, computational biology, natural and computer virus scanning, network intrusion detection, etc. It is well-known that performance remains a significant bottleneck to the high-performance solutions required in such industrial applications.