Masura Tomita. Efficient Parsing for Natural Languages. A Fast Algorithm for Practical Systems. Kluwer Academic Publishers, 1985.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....the attachment ambiguity. Hence, for D theory to be useful in a system that does semantic interpretation, it would need to be coupled with a semantic interpreter that recognizes the difference between these two examples. The approach taken in this paper is to combine an all path parsing algorithm [5, 7, 11, 16, 17] with routines for generating logical representations in order to create a shared packed parse forest annotated with the logical representations for the constituents in the forest (i.e. an annotated shared packed parse forest) Before discussing the benefits of this approach, we describe the ....

.... representations in order to create a shared packed parse forest annotated with the logical representations for the constituents in the forest (i.e. an annotated shared packed parse forest) Before discussing the benefits of this approach, we describe the properties of a shared packed parse forest [16, 17, 18]. A shared packed parse forest is a data structure which stores all parses of a sentence in a compact form. Consider the packed parse forest produced by an implementation of Tomita s parser [17] for the sentence Every man saw the boy with his binoculars shown in Figure 1. The forest stores both ....

[Article contains additional citation context not shown here]

M. Tomita. Efficient Parsing for Natural Language. Kluwer Academic Publishers, Boston, MA, 1985.

....the LFG specification for Turkish which we will present in the next section. 4.1.1 Generalized LR Parser Compiler The Generalized LR Parser Compiler is the syntactic part of the Universal Parser used in the CMU Machine Translation project. It is based on Tomita s Generalized LR Parsing Algorithm [18, 17], and is implemented in Common Lisp. Tomita s Generalized LR Parsing Algorithm is an efficient context free parsing algorithm and emphasizes its practical value in natural language processing. It produces all possible parse trees in an efficient representation, and seems more efficient than any of ....

M. Tomita, Efficient Parsing for Natural Language. Kluwer Academic Publishers, 1985.

....or the parse information must be translated to a form consistent with the standard grammar before the metric is calculated. Difficulties such as these can be avoided if a stronger parsing approach capable of recognising the standard grammar is used. We have found that Tomita (GLR) parsing [34] is well suited for this purpose. It can parse complex ambiguous programming languages (including C )usingthe standard grammar, in near linear time. Tomita parsing has the further advantage of simplifying parsing that relies on semantic disambiguation. Weaker parsing techniques do not ....

M. Tomita. Efficient Parsing for Natural Language.Kluwer Academic Publishers, 1986.

....commonly used for compilers, such as the LALR(1) algorithm in Yacc [6,7] are deterministic and cannot generally cope with the fact that input involving a Schrodinger token may (temporarily) not have a unique parse. Instead, we use more general parsing techniques such as generalized LR parsing [8] or Earley s algorithm [9,10] which effectively simulate non determinism if necessary to handle ambiguity in the grammar. In the past, these types of parsers have not been widely used in compilers due to efficiency concerns: all other things being equal, the more general parsing algorithms tend ....

....in Figure 2 would require the parser to be in two distinct automaton states simultaneously, which is not possible in a deterministic automaton. This is where general parsing algorithms come in. Where Yacc s deterministic LALR(1) algorithm would fail, an algorithm like generalized LR (GLR) parsing [8] simulates non determinism, effectively a = input( switch (action[s,a] foreach t#a.types switch (action[s, t] a) b) Figure 4. Pseudocode for (a) the LALR(1) parsing algorithm, and (b) conceptual modifications for ....

Tomita M. Efficient Parsing for Natural Languages. Kluwer Academic, 1986.

....cannot accept arbitrary CFGs. They are limited to subsets of unambiguous grammars. In general, the LR parsing table constructed for an arbitrary CFG will contain conflicts. That is, one or more states will provide a choice of actions to perform for some inputs. A parsing method due to Tomita [6,4] overcomes the limitations of the LR methods. It uses LR tables that may contain conflicts. Whenever the parser encounters a choice of parsing actions, it in effect clones new copies of itself to track each of the conflicting actions simultaneously. Some copies of the parser may subsequently reach ....

Tomita, M. Efficient Parsing for Natural Language. Kluwer Academic Publishers, Boston, 1986.

....foundation for building a subsymbolic parser of realistic language. 2 The Task: Shift Reduce Parsing The task taken up in this study, shift reduce (SR) parsing, is one of the simplest approaches to sentence processing that nevertheless has the potential to handle a substantial subset of English [Tomita, 1986] . Its basic formulation is based on the pushdown automata for parsing context free grammars, but it can be extended to context sensitive grammars as well. The parser consists of two data structures: the input buffer stores the sequence of words remaining to be read, and the partial parse results ....

M. Tomita. Efficient Parsing for Natural Language. Kluwer, Dordrecht; Boston, 1986.

....the new plan generation stage (9) The tokenization and grammatical subprocesses are implemented in the C programming language. Text strings are tokenized by employing a subsystem built around lex, a Unix lexical analysis tool [1] The grammatical processing is performed by a yacclike LR(1) parser [1, 16] extended to include backtracking, inheritance, token stream manipulation, and the use of semantic hierarchies, described in the next section. The semantic hierarchies (7) are also used by the later interpretation and generation modules. Most of the interpretation and generation modules are ....

....Other disadvantages are: there is no clear strategy for multi error detection and correction, and the fact that such a large corpus (20 megabytes in our mass change corpus) is required to create a good statistical model. 3. 2 Parsing For parsing, we use a generalized LR(1) shift reduce parser [16, 1, 10]. Like yacc (which, given a grammar, generates a parser for that grammar) our parser precompiles the CFG grammar into a state transition table. The parser exercises CFG grammar rules annotated with syntactic and semantic action routines, thus .allowing for synthesized and inherited attributes. In ....

Tomita, Masaru (1985). Efficient Parsing for Natural Language. Dordrecht: Kluwer Academic Publishers.

....language ASF [3] The parser generator produces generalized LR (GLR) parsers. Generalized parsing allows definition of the complete class of context free grammars instead of restricting it to a non ambiguous subclass such as LL(k) LR(k) or LALR(1) which is common to most other parser generators [24]. This allows for a more natural definition of the intended syntax because a grammar developer no longer needs to encode it in a restricted subclass. Moreover, since the full class of context free grammars is closed under composition (unlike restricted subclasses) generalized parsing allows for ....

M. Tomita. Efficient Parsing for Natural Languages. A Fast Algorithm for Practical Systems. Kluwer, 1985.

.... Although a classic nondeterministic top down parser in which the parse functions return lists of results [23, 8] could be memoized in a similar way, the resulting parser would not be linear time, and would likely be comparable to existing tabular algorithms for ambiguous context free grammars [3, 20]. Since nondeterministic parsing is equivalent in computational complexity to boolean matrix multiplication [12] a linear time solution to this more general problem is unlikely to be found. 5.2 Stateless Parsing A second limitation of packrat parsing is that it is fundamentally geared toward ....

Masaru Tomita. Efficient parsing for natural language. Kluwer Academic Publishers, 1985.

....of stacks places a high burden on both memory consumption and CPU time. However, many copies share the same suffix of tokens. Those suffixes can be shared and thus limit the resource consumption, i.e. a suffix tree is created. This is done in the same manner as the reuse of stack prefixes in [Tom86] The FDE can be used to fill the initial version of the meta index. But, the real benefit of a feature grammar shows when the feature detector algorithms change and the index has to be updated. To support index maintenance, the feature grammar can then be used with a different execution ....

Masaru Tomita. Efficient parsing for natural language. Kluwer Academic Publishers, 1986.

....search algorithm [4] is suitable for real time processing of speech recognition because of its frame synchronous processing and ease of implementation. On the other hand, use of Context Free Grammar (CFG) is advantageous in representing more generalized language constraints. Generalized LR parser [7] is one of most efficient parsers and is often used in natural language processing as well as in continuous speech recognition. It was combined with HMM phone models [1] first in the level building style and it provided one of the most accurate speech recognition methods producing Nbest candidates ....

M. Tomita, Efficient Parsing for Natural Language, Kluwer Academic Publishers, 1986.

....is supported by a parser generator that produces generalized LR (GLR) parsers. Generalized parsing allows definition of the complete class of context free grammars instead of restricting it to a non ambiguous subclass such as LL(k) LR(k) or LALR(1) which is common to most other parser generators [22, 21]. This allows for a more natural definition of the intended syntax because a grammar developer no longer needs to encode it in a restricted subclass. Moreover, since the full class of context free grammars is closed under composition (unlike restricted subclasses) generalized parsing allows for ....

M. Tomita. Efficient Parsing for Natural Languages. A Fast Algorithm for Practical Systems. Kluwer, 1985.

....generalized LR (GLR) parsers. Generalized parsing allows definition of the complete class of context free grammars instead of restricting it to a non ambiguous subclass of the contextfree grammars, such as the LL(k) LR(k) or LALR(1) class restrictions common to most other parser generators [36, 30]. This allows for a more natural definition of the intended syntax because a grammar developer no longer needs to encode it in a restricted subclass. Moreover, since the full class of context free grammars is closed under composition (the combination of two CF grammars is again a CF grammar) ....

M. Tomita. Efficient Parsing for Natural Languages. A Fast Algorithm for Practical Systems. Kluwer, 1985.

....tree produced by the parser is, in fact, a parse tree whose shape is dictated by that same grammar specification. The Harmonia framework uses two techniques for abstracting grammars: a variant of the EBNF notation for expressing sequences and optional nonterminals, and the GLR parsing technology [5] which transparently handles ambiguity, including unbounded syntactic lookahead. This approach lets us express language grammars at an abstract level without the usual contortions associated with producing a grammar amenable to traditional parsing methods. The same mechanism lets us create an ....

M. Tomita. Efficient Parsing for Natural Languages. Kluwer Academic Publishers, 1985.

....develop a technique to filter out problematic features from models (or grammar rules) This scheme will essentially weaken the top down constraints, therefore could be incorporated into the function ae. Lastly, we can adopt another data structure, possibly based on some kind of stack similar to (Tomita, 1986, 1991) to keep the items local and to ensure the appropriate models are used to fill the expectation. However, coping with left recursion may become a major difficulty. This is the subject of future research. ....

Tomita, M. (1986). Efficient Parsing for Natural Language. Boston: Kluwer Academic Publishers.

.... Factor Factor ( Exp ) Factor Num 4.4. THE VIEWS PARSER 33 is difficult to update an already generated parse table incrementally if the grammar is modified. For a detailed description of LR parsing, see [1, ch. 4. 7] Generalized LR parsing This is an extended LR parsing algorithm, see [12], that requires a conventional (but possibly multi valued) LR parse table. The parser starts as an LR parser, but when it encounters a multi valued entry in the parse table (conventionally known as a conflict) it splits up into as many parsers working in parallel as there are conflicting ....

M. Tomita. Efficient Parsing for Natural Languages. Kluwer Academic Publishers, 1985. 47

....parsing (O(n 3 ) and perhaps worse, due to the additional work of performing unifications. Empirical examinations of unification based parsers have indicated nonlinear average case performance as well (Shann, 1991; Carroll, 1994) Other popular parsing algorithms, such as Tomita s algorithm (Tomita, 1986), also fail to achieve average case linear performance, even without the inclusion of semantic interpretation. Our hypothesis is that a uniform approach to processing will result in a more efficient parsing algorithm. To test this hypothesis, we have developed a further extension of left corner ....

....algorithm uses a chart (or table) to record the partial constituents in order to eliminate redundant search. Earley s algorithm (Earley, 1970) a variant of chart parsing, is proven to run in time O(n 3 ) for general context free grammars. Tomita s Generalized LR parsing algorithm (GLR) (Tomita, 1986, 1991) uses a precompiled table, an extension of LR parse table, to guide the search at any given point in the parse. GLR also employs other efficient encoding techniques such as graph structured stack and packed shared forest. However, the worst case complexity of GLR is proven to be no better ....

Tomita, M. (1986). Efficient Parsing for Natural Language.

....in Lexicalized TAG. The relationships between the elementary trees in the final syntactic structure can be described as relations between their lexical anchors, thus producing a dependency tree. 1 There are no infinitely ambiguous or cyclic lexicalized grammars, such as S SS, S a, S ffl (Tomita, 1985). Pure lexicalized formalisms, however, have also several disadvantages. Linguistically, the abstract syntactic level is necessary to avoid the loss of generalization which would arise if class level information were repeated in all lexical items. In parsing, a predictive component is required to ....

M. Tomita. 1985. Efficient Parsing for Natural Language. Kluwer Academic Publishers.

No context found.

Masura Tomita. Efficient Parsing for Natural Languages. A Fast Algorithm for Practical Systems. Kluwer Academic Publishers, 1985.

No context found.

Masaru Tomita. Efficient Parsing for Natural Language.Kluwer Academic Publishers, Dordrecht, 1986.

No context found.

M. Tomita. 1985. Efficient parsing for natural language. Kluwer Academic Publishers, Boston, MA.

No context found.

Masaru Tomita. Efficient Parsing for Natural Language, A Fast Algorithm for Practical Systems. Kluwer Academic Publishers, 1985.

No context found.

M. Tomita, Efficient Parsing for Natural Language, Boston, MA: Kluwer Academic, 1986.

No context found.

M. Tomita. Efficient Parsing for Natural Language, A Fast Algorithm for Practical Systems. Kluwer, 1986.

No context found.

Masaru Tomita. 1986. Efficient Parsing for Natural Language. Kluwer Academic Publishers.

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