S. Wu and U. Manber. Agrep -- a fast approximate patternmatching tool. In Proceedings USENIX Winter 1992.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....can easily be packed into 1MB of memory. Our approximate string matching algorithm is based on BaezaYates shifter algorithm [1] with our extensions for allowing errors and handling simple wildcard characters. The shifter algorithm is also used in an approximate string matching program agrep[7] (an extension to grep on UNIX) where wildcard characters are treated differently from ours. In our algorithm, we limit the wildcard to the basic . pattern in order to achieve simple and fast processing. Figure 18: A state transition machine which accepts ab. ca . z x g n q z x ....

Wu, S., and Manber, U. Agrep - a fast approximate patternmatching tool. In Proceedings of USENIXTechnicalConference (San Francisco, CA, January 1992), pp. 153--162.

....usually requires a large storage, about 21u bytes) although the construction is in O(u) time. The search time used by the two algorithms are shown in Figure 2 (left column) For k approximate matches, we tested the proposed method with two popular approximate pattern matching algorithms: agrep [13], and nrgrep [10] Both algorithms are based on bit wise operations using the patterns and text. The two algorithms operate on the raw (uncompressed) text. The results for the search time are shown in Figure 2, right column. Results for the proposed BWTbased approach is labeled bwt dfa. Here, the ....

Wu, S. and Manber, U. (1992a). agrep --- A fast approximate pattern-matching tool. In USENIX Association, editor, Proceedings of the Winter 1992 USENIX Conference: January 20 --- January 24, 1992, San Francisco, California, pages 153--

....Second, the creators of the benchmark suite also produced a version of each application s source code that has been manually modified to initiate prefetching by issuing TIP hints (which I used without change) 7.3.1 Agrep The Agrep benchmark uses version 2. 04 of the Agrep application [71]. Agrep, a variant of the standard UNIX Grep utility, is a fast full text pattern matching utility that allows matching errors. The benchmark searches a bunch of text files for a simple string that does not occur in any of the files. In terms of data requests, it loops through the files specified ....

S. Wu and U. Manber. AGREP - a fast approximate pattern-matching tool. In Proceedings of the USENIX Winter 1992.

.... an index for that purpose) dynamic text collections (where the cost of keeping an up to date index is prohibitive, including the searchers inside text editors and Web interfaces ) for not very large texts (up to a few hundred megabytes) and even as internal tools of indexed schemes (as agrep [29] is used inside glimpse [15] or cgrep [17] is used inside compressed indexes [21] Dept. of Computer Science, University of Chile. Blanco Encalada 2120, Santiago, Chile. gnavarro dcc.uchile.cl. Work developed while the author was at postdoctoral stay at the Institut Gaspard Monge, Univ. de ....

....a large class of string matching algorithms in the literature (see, for example, 26, 8, 4] but not all of them are practical. There is also a wide variety of fast online string matching tools in the public domain, most prominently the grep family. Among these, Gnu grep and Wu and Manber s agrep [29] are widely known and currently considered as the fastest string matching tools in practice. Another distinguishing feature of these software systems is their flexibility: they can search not only for simple strings, but they also permit classes of characters (that is, a pattern position matches a ....

[Article contains additional citation context not shown here]

S. Wu and U. Manber. Agrep -- a fast approximate pattern-matching tool. In Proc. USENIX Technical Conference, pages 153--162, 1992.

....be on line or it may be indexed. In the former case, all documents in the system are searched sequentially for matches. This approach is required when space or time does not allow the documents to be pre processed. Agrep is a Unix tool developed by Wu and Manber which supports fast online search [Wu92] The second case pre processes the documents, building an index, to improve subsequent approximate searches on those documents. A document may be indexed at the word level, where the exact location of each word is stored in the index, in blocks or as a whole document. Word level indices tend to ....

Wu, S. and Manber, U., Agrep - A Fast Approximate Pattern-matching Tool, in Proc. USENIX Technical Conference, 1992, pp. 153-162.

....the search scheme proceeds in two steps: first the queried word w is searched in the vocabulary V , then all candidate blocks of L(w) are sequentially examined to find all the w s occurrences. Complex queries (e.g. approximate or regular expression searches) can be supported by using Agrep [28] both in the vocabulary and in the block searches. Clearly, the search is efficient if the vocabulary is small, if the query is enough selective, and if the block size is not too large. The first two requirements are usually met in practice, so that the main constraint to the effective use of ....

S. Wu and U. Manber. AGREP - A fast approximate patternmatching tool. In Proceedings of the Usenix Winter

....or as an implementation of a DFA (where the identifier of each deterministic state is the bit mask as a whole) Pushing even more on this direction, one may resort to pure bit parallelism and forget about the modules. This was done in [13] by Wu and Manber, and included in their software Agrep [12]. A computer word is used to represent the active (1) and inactive (0) states of the NFA. If the states are properly arranged and the Thompson construction [10] is used, then all the arrows carry 1 s from bit positions i to i 1, except for the transitions. Then, a generalization of Shift Or ....

.... [10] which builds an NFA with at most 2m states (and at least m 1) This NFA has some particular properties (e.g. O(1) transitions leaving each node) that have been extensively exploited in several regular expression search algorithm such as that of Thompson [10] Myers [7] and Wu and Manber [13, 12]. Another particularly interesting NFA construction algorithm is by Glushkov [6] popularized by Berry and Sethi in [3] The NFA resulting from this construction has the advantage of having just m 1 states (one per position in the regular expression) Its number of transitions is worst case ....

[Article contains additional citation context not shown here]

S. Wu and U. Manber. Agrep -- a fast approximate pattern-matching tool. In Proc. of USENIX Technical Conference, pages 153--162, 1992.

.... provides the new set of active states (another computer word) This can be considered either as a bit parallel simulation of an NFA, or as an implementation of a DFA (where the identifier of each deterministic state is the bit mask as a whole) This idea was first proposed by Wu and Manber [31, 30]. Later, Navarro and Raffinot [23] used a similar procedure, this time using Glushkov s [7] construction of the NFA. This construction has the advantage of producing an automaton of exactly m 1 states, while Thompson s may reach 2m states. A drawback is that the structure is not so regular and ....

S. Wu and U. Manber. Agrep -- a fast approximate pattern-matching tool. In Proc. of USENIX Technical Conference, pages 153--162, 1992.

....We have also considered the naive approach of decompressing then searching. Two choices are shown: DS uses our LZ78 format and decompresses the file in memory while applying a Sunday [19] search algorithm over it; D Agrep first decompresses the text and then then runs agrep over it. Agrep [21, 22] is considered the fastest text searching tool, and we recall that the decompression time of our format is the fastest. As can be seen, our algorithms are significantly faster than Bit parallel (up to 30 ) and than both decompress then search approaches (up to 50 ) even for short patterns (m ....

S. Wu and U. Manber. Agrep -- a fast approximate pattern-matching tool. In Proc. USENIX Technical Conference, pages 153--162, Berkeley, CA, USA, Winter 1992.

.... an index for that purpose) dynamic text collections (where the cost of keeping an up to date index is prohibitive, including the searchers inside text editors and Web interfaces 1 ) for not very large texts (up to a few hundred megabytes) and even as internal tools of indexed schemes (as agrep [26] is used inside glimpse [14] or cgrep [16] is used inside compressed indexes [20] There is a large class of string matching algorithms in the literature (see, for example, 8, 4] but not all of them are practical. There is also a wide variety of fast online string matching tools in the Dept. ....

....Gaspard Monge, Univ. de Marne la Vall ee, France, partially supported by Fundaci on Andes and ECOS Conicyt. 1 We refer to the search in page facility, not to confuse with searching the Web. 1 public domain, most prominently the grep family. Among these, Gnu grep and Wu and Manber s agrep [26] are widely known and currently considered the fastest string matching tools in practice. Another distinguishing feature of these softwares is their flexibility: they can not only search for simple strings, but they also permits classes of characters (that is, a pattern position matches a set of ....

[Article contains additional citation context not shown here]

S. Wu and U. Manber. Agrep -- a fast approximate pattern-matching tool. In Proc. USENIX Technical Conference, pages 153--162, 1992.

....BYP [3] i.e. the original version of this algorithm) BYN [2] and Myers [6] the other fastest algorithms, based on bit parallelism) and Ours (our modification to BYP with hierarchical verification) The code is from the authors in all cases. On English text we add two extra algorithms: Agrep [13] (the fastest known approximate search software) and a version of our algorithm that includes the splitting optimization. On English text the code Ours corresponds to our algorithm with hierarchical verification and splitting optimization, while Ours NO shows hierarchical verification and no ....

....that includes the splitting optimization. On English text the code Ours corresponds to our algorithm with hierarchical verification and splitting optimization, while Ours NO shows hierarchical verification and no splitting optimization. The algorithms included in this comparison are: Agrep [13] (the fastest known approximate search software, only for English text) BYP [3] i.e. the original version of this algorithm) BYN [2] and Myers [6] the other fastest algorithms, based on bit parallelism) Ours (our modification to BYP with both improvements) and Ours NO (the same, including the ....

S. Wu and U. Manber. Agrep -- a fast approximate pattern-matching tool. In Proc. of USENIX Technical Conference, pages 153--162, 1992.

....of the blocks where the word appears is kept. See Figure 1. 2 b words block of b words block of b words block of r blocks Text words occurrences Index Figure 1: The word indexing scheme. To search a word allowing errors, the vocabulary is sequentially scanned, word by word, with Agrep [9]. Agrep is an on line approximate search software, which will treat the vocabulary as a simple piece of text. For each matching word, all the blocks where it appears in the text are marked. Then, for every marked block (i.e. where some matching word is present) a new sequential search is ....

S. Wu and U. Manber. Agrep -- a fast approximate pattern-matching tool. In Proc. USENIX, pages 153--162, 1992.

....1= log n) otherwise it is O( p mk= log n n) i.e. O( p k n) for m = O(log n) else O(kn) It involves also a cost to verify potential matches, which is shown to be not significant for ff ff 1 1 Gamma m 1= p log n = p oe. This algorithm is a generalization of an earlier heuristic [23, 5], that reduces the problem to subproblems of exact matching and is shown to be O(n) for ff ff 0 = 1= 3 log oe m) The second one partitions the automaton in sub automata, being O(k 2 n= p oe log n) on average. For ff 1 Gamma 1= p oe its worst case, O( m Gamma k)kn= log n) ....

....the special case j = k 1 is different, since in that case we search with 0 errors, and a faster algorithm is possible (i.e. exact multipattern search) Thus, it is possible that in some cases we may prefer to increase j beyond its minimum value, setting it to k 1. This is the heuristic used in [23, 5]. To analyze this algorithm, we assume the use of an Aho Corasick machine [1] which guarantees O(n) search time. However, in practice we find that an extension of Boyer Moore Horspool Sunday [15] to multipattern searching is faster. This extension consists of building a trie with the ....

[Article contains additional citation context not shown here]

S. Wu and U. Manber. Agrep -- a fast approximate pattern-matching tool. In Proc. of USENIX Technical Conference, pages 153--162, 1992.

....This expresses the distance as closer to 1.0 for similar frequencies, and closer to 0.0 for less similar frequencies. Comparison of path profiles is accomplished using an adaptation of an edit distance or Levenshtein measure algorithm. This algorithm is widely used for approximate string matching [73, 74], and other approximate pattern matching tasks. My method includes a frequency distance measure as a second point of comparison. Experimentation is needed to determine the quality of this comparison technique for my purposes, and to discover appropriate threshold values for quantifying similarity ....

S. Wu and U. Manber. AGREP -- a fast approximate pattern-matching tool. In Proceedings of the Winter 1992 USENIX Conference, pages 152--162, Berkeley, California, Jan. 1992.

....We measure preprocessing and searching time together, since preprocessing time is totally negligible. The slowest preprocessing time was 2 milliseconds, which is less than 1 of the fastest searching time on 10 Mb. The algorithms included in this comparison are (in alphabetical code order) Agrep [32] is a widely distributed exact and approximate search software oriented to natural language text. It is limited (although not intrinsically) to m 32 and k 8. BM is a filter based on applying a Boyer Moore type machinery [27] The code is from the authors. BPM (bit parallel matrix) is a recent ....

S. Wu and U. Manber. Agrep -- a fast approximate pattern-matching tool. In Proc. of USENIX Technical Conference, pages 153--162, 1992.

....text about 10 seconds Partially supported by the University of Illinois Critical Research Initiatives on a SUN SPARC 10. Due to this efficiency, multiple string pattern matching techniques have many applications. Wu and Manber proposed an efficient approximate pattern matching software, agrep [12] and a text searching tool, glimpse [7] based on their multiple string matching algorithm [13] Kim and Segre proposed a fast, reliable DNA sequence assembly algorithm using their multiple string pattern matching technique [9] We propose an efficient multiple string pattern matching algorithm ....

....position computed by logically ANDing T and H. If the hash entry at the position is empty, skip the pattern testing procedure (see Section 3.1) and scan the next text character. EXPERIMENTS To evaluate the performance of our two algorithms, MULTI1, we compare our algorithms to grep and agrep [12] on a SUN Ultra 10 workstation (333MHz processor) with 128MB memory. MULTI1 used a hash table of size up to 1M entries. All times reported are user times measured by a Unix command time. We performed two different kinds of experiments with two different texts; one text is the King James Bible ....

Wu, S. and Manber, U., "Agrep -- A Fast Approximate Pattern-matching Tool," Usenix Winter Technical Conference, 1992

.... implementations of recognizers have been described [20, 24] The UNIX operating system in particular provides a number of software tools constructed around regular expression recognizers; these include grep, awk [1] and lex [21] and research into extensions to these tools continues (e.g. agrep [30] and TLex [19] Whilst many of the above systems and languages offer a great deal of functionality to the user, the pattern specifications created are often extremely cryptic and terse for all but the most basic of patterns. The differentiation between normal characters and metacharacters is ....

WU, S., ANDMANBER, U. agrep --- A Fast Approximate Pattern-Matching Tool. In Proceedings of Winter USENIX'92 Conference (1992). 13

No context found.

S. Wu and U. Manber. Agrep -- a fast approximate patternmatching tool. In Proceedings USENIX Winter 1992.

No context found.

S. Wu and U. Manber. Agrep -- a fast approximate pattern-matching tool. In Proc. USENIX Technical Conference, pages 153--162, 1992.

No context found.

S. Wu and U. Manber. Agrep -- a fast approximate pattern-matching tool. In Proc. of the USENIX Technical Conference, pages 153--162, 1992.

No context found.

S. Wu and U. Manber. Agrep -- a fast approximate pattern-matching tool. In Proc. of the USENIX Technical Conference, pages 153--162, 1992.

No context found.

S. Wu and U. Manber. agrep - A fast approximate pattern-matching tool. In Proceedings of the Usenix Winter 1992.

No context found.

S. Wu and U. Manber. Agrep -- a fast approximate pattern-matching tool. In Usenix Winter 1992 Technical Conference, pages 153--162, 1992.

No context found.

S. Wu and U. Manber. Agrep -- a fast approximate patternmatching tool. In Proc. of USENIX Technical Conf, pages 153--162, 1992.

No context found.

Wu S., and U. Manber, "Agrep --- A Fast Approximate Pattern-Matching Tool," Usenix Winter 1992 Technical Conference, San Francisco (January 1992), pp. 153-162.

First 50 documents  Next 50

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