Oommen, B.J., and loke, R.K.S., Pattern Recognition of Strings with Substitutions, Insertions, Deletions, and Generalized Transpostions, Pettern Recognition, vol. 30, No. 5, pp. 789-800, 1997.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....W are suggested as correction candidates. If necessary, appropriate statistical data can be used for re nement of ranking. Similarity between two words can be measured in several ways. Most useful are (dis)similarity measures based on variants of the Levenshteindistance [Lev66, WF74, WBR95, SKS96, OL97] or on n gram distances [AFW83, Ukk92, KST92, KST94] In this paper, we take the Levenshtein distance as a basis. The standard algorithm for computing the Levenshtein distance between two words by Wagner and Fisher [WF74] uses a dynamic programming scheme that leads to quadratic time complexity. ....

B.J. Oommen and R.K.S. Loke. Pattern recognition of strings with substitutions, insertions, deletions, and generalized transpositions. Pattern Recognition, 30(5):789-800, 1997.

....that are most similar to W are good suggestions for correction. Similarity between two words can be measured in several ways. Most popular are (dis)similarity measures based on n gram distances [AFW83, Ukk92, KST92, KST94] or on variants of the Levenshtein distance [Lev66, WF74, WBR95, SKS96, OL97] In this paper, we take the Levenshtein distance as a basis. 1 The standard algorithm for computing the Levenshtein distance between two words by Wagner and Fisher [WF74] uses a dynamic programming scheme that leads to quadratic time complexity. Even with more sophisticated algorithms (cf. ....

B.J. Oommen and R.K.S. Loke. Pattern recognition of strings with substitutions, insertions, deletions, and generalized transpositions. Pattern Recognition, 30(5):789-800, 1997.

.... both optimal and quadratic. The only reported result for traditional transpositions is the the one proposed by Lowrance and Wagner [LW75,SK83] The difference between our algorithm and the scheme presented by Lowrance and Wagner for traditional transpositions is given in the unabridged paper [OL94]. We formalize the problem as follows. We are given a string Y which is the noisy version of some unknown string X chosen from a finite dictionary, H. Apart from Y containing SID errors, it also contains transposed characters which are themselves subsequently substituted. The intention is to ....

.... is true : D(X i , Y j ) Min [ D(X i 1 , Y j ) d e (x i ) D(X i , Y j 1 ) d i (y j ) D(X i 1 , Y j 1 ) d s (x i , y j ) D(X i 2 , Y j 2 ) d t (x i 1 x i , y j 1 y j ) The proof, which differs from proofs traditionally used in the literature can be found in the unabridged paper [OL94]. 3. THE COMPUTATION OF D(X,Y) To compute D(X,Y) we make use of the recursive properties given above. The idea is essentially one of computing the distance D(X i , Y j ) between the prefixes of X and Y. The computation of the distances has to be done in a schematic manner, so that any quantity ....

[Article contains additional citation context not shown here]

B. J. Oommen and R. K. S. Loke, Pattern recognition of strings with substitutions, insertions, deletions and generalized transpositions. Unabridged Paper. Will be soon available as a Carleton University technical report.

.... Fischer [11] algorithm, ii) SID and traditional transposition errors were assumed and corrected using Lowrance and Wagner [5,10] algorithm and (iii) SID and generalized transposition errors were assumed and corrected using a recent unconstrained editing algorithm for all the four operations [7]. The dictionary, H, used consisted of a hundred strings taken from the classical book on pattern recognition by Duda and Hart and were randomly truncated so that the length of the words in H was uniformly distributed in the interval [40, 80] Using random deletions as in [6] a set of 500 ....

....are tabulated in Table 2. Note that our scheme far outperforms the traditional string correction algorithm (94.0 instead of 64.2 ) It also outperforms the Lowrance and Wagner algorithm (which had an accuracy of 75.6 ) Our recent unconstrained distance criterion for all the four errors [7] yielded an accuracy of 74.6 . The power of the strategy in PR is obvious 6 Conclusions In this paper we have considered the problem of recognizing strings by processing their noisy subsequences. The solution which we propose is the only known solution in the literature when the noisy ....

B. J. Oommen and R. K. S. Loke, Pattern recognition of strings with substitutions, insertions, deletions and generalized transpositions. Unabridged Paper. Available as a Carleton University technical report (1994).

.... Fischer [WF74] algorithm, ii) SID and traditional transposition errors were assumed and corrected using Lowrance and Wagner [LW75] algorithm. iii) SID and generalized transposition errors were assumed and corrected using a recent unconstrained editing algorithm for all the four operations [OL94]. The dictionary used consisted of a hundred strings taken from the classical book on pattern recognition by Duda and Hart [DH73] Each string was the first line of a section or sub section of the book, starting from Section 1.1 and ending with Section 6.4.3. Further, to mimic a UNIX nroff (or ....

....93.00 2.7 93.20 3.0 93.40 3.3 93.40 3.6 93.80 4.0 94.00 4.3 93.60 4.5 93.40 5. 0 93.40 Table V: The variation of the recognition accuracy as a function of K t The four algorithms, Wagner Fischer (WF) Lowrance Wagner (LW) the unconstrained algorithm for generalized transpositions (SID GT) [OL94] and our constrained algorithm (Const SID GT) were tested with the 500 noisy subsequences. In the case of our algorithm, rather than have the constraint set use only a single feasible integer for the number of insertions and transpositions, the algorithm was marginally modified to include a small ....

[Article contains additional citation context not shown here]

B. J. Oommen and R. K. S. Loke, Pattern recognition of strings with substitutions, insertions, deletions and generalized transpositions. Submitted for Publication. Available as a Carleton University Technical Report.

No context found.

Oommen, B.J., and loke, R.K.S., Pattern Recognition of Strings with Substitutions, Insertions, Deletions, and Generalized Transpostions, Pettern Recognition, vol. 30, No. 5, pp. 789-800, 1997.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC