Performance analysis of some simple heuristics for computing longest common subsequences
- 105 Downloads
Although theLongest Common Subsequence (LCS)Problem has been studied by many researchers for years, heuristic methods have not been investigated before. In this paper we present a simple heuristic which guarantees to return a common subsequence of length at least 1/s that of the longest wheres is the number of different symbols in the input strings. Furthermore, we generalize the idea to several classes of heuristic algorithms. Surprisingly, we find that no other heuristic in these classes outperforms this simple algorithm. In other words, we show that any heuristic which uses only global information, such as number of symbol occurrences, might return a common subsequence as short as 1/s of the length of the longest. Analysis of the average performance of the simple heuristic fors=2 is also presented.
Key wordsLongest common subsequence Heuristics Performance analysis Scan algorithms
Unable to display preview. Download preview PDF.
- F. Chin and C. K. Poon, A fast algorithm for computing longest common subsequences of small alphabet size,J. Inform. Process., 13(4) (1990), 463–469. A preliminary version also appeared inProceedings of the International Workshop on Discrete Algorithms and Complexity, 1989, pp. 163–168.zbMATHGoogle Scholar
- F. Chin and C. K. Poon, Performance of heuristics for the longest common subsequences problem,Proceedings of the 1990 International Computer Symposium, Hsinchu, Taiwan, December 1990, pp. 164–169.Google Scholar
- G. R. Cross and S. Kuo, Two-Step String-Matching Procedure, Technical Report CS-89-198, Washington State University, 1989.Google Scholar