Ristad, E. S., and Yianilos, P. N. 1998. Learning string edit distance. IEEE Transactions on Pattern Analysis and Machine Intelligence 20(5):522--532.  Home/Search   Document Details and Download   Summary   Related Articles   Check

.... according to the class to which y belongs; but in the latter we are interested in a string of L(F ) that is most likely transformed by C into the string x actually received by R according to E [5] It is expected superior performance when using adequately estimated stochastic parameters [8, 1]. However, although there are well known robust methods to estimate the parameters of stochastic finite state models [4] this is not the case for the estimation of the parameters of E. As far as we know this issue has been addressed only in [10, 2, 8, 1] Henceforth, symbols (primitives) are ....

....using adequately estimated stochastic parameters [8, 1] However, although there are well known robust methods to estimate the parameters of stochastic finite state models [4] this is not the case for the estimation of the parameters of E. As far as we know this issue has been addressed only in [10, 2, 8, 1]. Henceforth, symbols (primitives) are represented by lowercase letters, with n being the null symbol; alphabets (collections of symbols) are represented by capital Greek letters; and strings (concatenations of symbols) are represented by lowercase letters with a bar on top, with being the null ....

[Article contains additional citation context not shown here]

E. S. Ristad and P. N. Yianilos. Learning string-edit distance. IEEE PAMI, 20(5):522--532, May 1998.

....in the extracted data into a canonical form, e.g. mapping NT, WinNT , Windows NT, and Microsoft Windows NT all to a unique term. In many cases, such collapsing could be automated by clustering slot fillers using a distance metric based on textual similarity, such as character edit distance (Ristad Yianilos 1998). Currently, we only consider discrete valued slots. However, real valued slots, such as desired years of experience, could also be provided to the rule miner as additional input features when predicting other slots. Predicting such continuous values using regression methods instead of ....

Ristad, E. S., and Yianilos, P. N. 1998. Learning string edit distance. IEEE Transactions on Pattern Analysis and Machine Intelligence 20(5).

....2 shows a very specific PHMM that generates the pair RING , RANG with probability one. The domain and range of this transduction are thus singleton sets. 3.2 Relationships to other formalisms Regular syntax directed translation schemes. Aho Ullman, 1969a, 1969b) Memoryless transducers. (Ristad Yianilos, 1998) Stochastic Inversion Transduction Grammars. Wu, 1995, 1997) Non deterministic stochastic finite state transducers. Mohri, 1997a, 1997b) 4 s 0 q 11 (R) 1 q 10 (I) 1 q 01 (A) 1 q 11 (N) 1 q 11 (G) 1 s 1 1.0 1.0 1.0 1.0 1.0 1.0 Figure 2: A simple PHMM that maps RING to ....

Ristad, E. S., & Yianilos, P. N. (1998). Learning string-edit distance. IEEE Transactions on Pattern Analysis and Machine Intelligence, 20 (5), 522--532.

....two strings. The most commonly accepted measure is the edit distance. Ristad and Yianilos describe in detail a method for determining the string edit distance. They define edit distance as the minimum number of insertions, deletions and substitutions required to transform one string into another [10]. Figure 4 gives an example of insertion, deletion and substitution on the text string casablanca : 5 Figure 4. Examples of (a) Insertion, b) Deletion and (c) Substitution 3.2.2 Music String Searching In order to provide content based retrieval of music, it is necessary to search on the ....

E. Ristad and P. Yianilos, "Learning String Edit Distance", Princeton University, October 1997.

....x into y. This condition of irreducible operator ensures for example that the search space can be split into a partition of basins of attraction. On the other hand, other distances have been suggested to take into account the probability of visiting different neighbors of x. Ristad and Yanilos [45] propose to relate the distance between two strings to the probability that a random sequence of operator applications transforms a string into another, or to the probability of the most likely sequence that transforms a string into another. We are now ready to define the notion of a basin of ....

E. S. Ristad and P. N. Yanilos. Learning string edit distance. Technical Report CS-TR532 -96, Princeton University, 1996.

....on strings. For instance, for personal names, a distance metric such as Soundex might be appropriate; in other situations, domainindependent edit distance metrics such as Smith Wasserman [ Monge and Elkan, 1997 ] TFIDF distance [ Salton, 1989; Cohen, 1998 ] or a learned edit distance metric [ Ristad and Yianilos, 1997 ] might be appropriate. In Section 4 we will derive a cost function that allows one to compare di erent hardenings of a given soft database (relative to a given potential interpretation set I pot ) More speci cally, we will want to minimize a cost function of the form w(I) 1 jI(S)j 2 ....

E.S. Ristad and P. N. Yianilos. Learning string edit distance. In Machine Learning: Proceedings of the Fourteenth International Conference. Morgan Kaufmann, 1997.

....of error may be more common so this weighting may not be the most appropriate. For query by humming, more research into tune recall is needed to identify suitable weights. It may be possible to automatically establish appropriate weights for the metric by learning from a reference set of data [31]. A brute force approach to collating the near match set would iterate sequentially through all contours. It is possible to implement a similar algorithm using a tree structure to visit each contour. Consider a tertiary tree which has a depth equal to the atomic length, L. Each node contains a ....

Ristad, E.S. and Yianilos, P.N., Learning String Edit Distance, Technical Report CS-TR-532-96, Department of Computer Science, Princeton University, 1996.

....selection produces large solution yields without compromising search. 7.2 Future Work There are several issues raised by this thesis which merit further attention. The treatment of edit operation weights in chapter 3 is unsophisticated. It is possible to learn edit weights for particular problems (Ristad and Yianilos 1998). Rather than arbitrarily setting all the weights to 1, it might be possible to learn weights from a training set of labelling problems. Similarly, the ambiguous measurement framework for graph matching, introduced in chapter 4, was somewhat ad hoc. In particular, the control of the ambiguity ....

E. S. Ristad and P. N. Yianilos (1998). Learning string edit distance. IEEE Transactions on Pattern Analysis and Machine Intelligence 20, 522--532.

....worked surprisingly well. It also shares with lllama the advantages of extensibility to long sequences and ability to recognize subsequence features. It also has the advantage of much shorter training time. Extending the method by tuning the SmithWaterman matching parameters (for example, see [16]) could yield significant improvements in classification accuracy with only minimal cost in training time. ....

E. S. Ristad and P. N. Yianilos, Learning string edit distance, Tech. Rep. TR-532-96, Princeton University, 1996.

No context found.

Ristad, E. S., and Yianilos, P. N. Learning string edit distance. In Machine Learning: Proceedings of the Fourteenth International Conference (San Francisco, July 8--11 1997), D. Fisher, Ed., Morgan Kaufmann, pp. 287--295.

No context found.

Ristad, E. S., and Yianilos, P. N. Learning string edit distance. In Machine Learning: Proceedings of the Fourteenth International Conference (San Francisco, July 8--11 1997), D. Fisher, Ed., Morgan Kaufmann, pp. 287--295.

No context found.

Ristad, E. S., and Yianilos, P. N. 1998. Learning string edit distance. IEEE Transactions on Pattern Analysis and Machine Intelligence 20(5):522--532.

No context found.

E. S. Ristad and P. N. Yianilos. 1997. Learning string edit distance. In Machine Learning: Proceedings of the Fourteenth International Conference, pages 287--295.

No context found.

E. S. Ristad, and P. N. Yianilos, "Learning stringedit distance", IEEE Transactions on Pattern Analysis and Machine Intelligence, 20(5):522--532, 1998.

No context found.

Ristad E. S. and Yianilos P. N.: Learning string edit distance. In Machine Learning: Proceedings of the 14th International Conference (San Francisco, July 8--11 1997), D. Fisher, Ed., Morgan Kaufmann, pp. 287--295.

No context found.

Ristad, E.S. and Yianilos, P.N., Learning String Edit Distance, Technical Report CS-TR-532-96, Department of Computer Science, Princeton University, 1996.

No context found.

Ristad, Eric Sven, and Peter N. Yianilos. 1997. Learning String Edit Distance. Princeton University Computer Science Department, Research Report CS-TR-532-96.

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