Showing papers on "Tree-adjoining grammar published in 2012"

The Logic of Categorial Grammars

Abstract: A method for harvesting invention fowl which includes the steps of horizontally extending beneath the fowl, in a confined area, a plurality of lifting fingers; raising and pivoting the fingers to lift the fowl and supporting them at least in part upon a continuously moving structure; continuing to said the fowl on the continuously moving structure to convey the fowl to a cooping location; and moving the fowl into a coop from the continuously moving structure.

The Logic of Categorial Grammars: A deductive account of natural language syntax and semantics

Abstract: This book is a contemporary and comprehensive introduction to categorial grammars in the logical tradition initiated by the work of Lambek. It guides students and researchers through the fundamental results in the field, providing modern proofs of many classic theorems, as well as original recent advances. Numerous examples and exercises illustrate the motivations and applications of these results from a linguistic, computational and logical point of view. The Lambek calculus and its variants, and the corresponding grammars, are at the heart of these lecture notes. A chapter is devoted to a key feature of these categorial grammars: their very elegant syntax-semantic interface. In addition, we adapt linear logic proof nets to these calculi since they provide efficient parsing algorithms as exemplified in the Grail parser. This book shows how categorial grammars weave together converging ideas from formal linguistics, typed lambda calculus, Montague semantics, proof theory and 1 linear logic, thus yielding a coherent and formally elegant framework for natural language syntax and semantics.

Inference of Program Properties with Attribute Grammars, Revisited

Abstract: A programming language is an essential ingredient for writing concise, maintainable, and error-free computer programs. A compiler takes a text written in such a language and compiles into machine instructions, and is usually implemented as a number of traversals over the abstract syntax of the program. Attribute Grammars (AGs) are a powerful tool for the description of such traversals and thus the implementation of a compiler: AGs offer aspect-oriented programming, abstraction over common traversal patterns, and automatic inference of a sound and efficient traversal algorithm. Over the years, computer languages have become more complex and harder to implement. Notoriously difficult to implement is type checking or type inference, which for complex languages is not only specified in terms of the abstract syntax but also in terms of the inferred types, and requires traversals that are hard to describe in general, and in particular with AGs. The reason is that the traversals mutually depend on types, which are not known apriori, thus the traversal represents some strategy that decides during the process in which order type information is collected and processed. In this thesis, we investigated the application of attribute grammars to the description of inference algorithms, with the goal of being able to describe such algorithms while retaining the good properties of AGs. To this end, this thesis builds on higher-order attribute grammars to describe traversals over semantics instead of syntax, conditional attribute grammars to describe the decisions to be made during the inference process, and ordered attribute grammars to express inference strategies. Visits, a concept from ordered attribute grammars, play an essential role in our work. A visit to a node in the abstract syntax tree represents a unit of evaluation. We make visits explicit in our AG descriptions so that we can express the order of evaluation and conditionally iterate visits. Moreover, we show that on top of visits, we can express decision points, and present a stepwise evaluation strategy that allows us to explore alternative choices until reaching a decision. Our work is a conservative extension of attribute grammars, and preserves their good aspects. We declaratively express properties of the evaluation order, while keeping the automatic scheduling of attribute computations. Our work integrates well with various attribute grammar extensions, such as parallel evaluation. In particular, our work facilitates an integration with dependently typed programming, which paves the way to prove and enforce the correctness of compilers described with attribute grammars.

Simple robust grammar induction with combinatory categorial grammars

Abstract: We present a simple EM-based grammar induction algorithm for Combinatory Categorial Grammar (CCG) that achieves state-of-the-art performance by relying on a minimal number of very general linguistic principles. Unlike previous work on unsupervised parsing with CCGs, our approach has no prior language-specific knowledge, and discovers all categories automatically. Additionally, unlike other approaches, our grammar remains robust when parsing longer sentences, performing as well as or better than other systems. We believe this is a natural result of using an expressive grammar formalism with an extended domain of locality.

Implicit learning of recursive context-free grammars.

Abstract: Context-free grammars are fundamental for the description of linguistic syntax. However, most artificial grammar learning experiments have explored learning of simpler finite-state grammars, while studies exploring context-free grammars have not assessed awareness and implicitness. This paper explores the implicit learning of context-free grammars employing features of hierarchical organization, recursive embedding and long-distance dependencies. The grammars also featured the distinction between left- and right-branching structures, as well as between centre- and tail-embedding, both distinctions found in natural languages. People acquired unconscious knowledge of relations between grammatical classes even for dependencies over long distances, in ways that went beyond learning simpler relations (e.g. n-grams) between individual words. The structural distinctions drawn from linguistics also proved important as performance was greater for tail-embedding than centre-embedding structures. The results suggest the plausibility of implicit learning of complex context-free structures, which model some features of natural languages. They support the relevance of artificial grammar learning for probing mechanisms of language learning and challenge existing theories and computational models of implicit learning.

Vietnamese Parsing with an Automatically Extracted Tree-Adjoining Grammar

Abstract: This paper presents the construction and evaluation of a deep syntactic parser based on Lexicalized Tree-Adjoining Grammars for the Vietnamese language. This is a complete sys- tem integrating necessary tools to process Vietnamese text, which permits to take as input raw texts and produce syntactic struc- tures. A dependency annotation scheme for Vietnamese and an algorithm for extracting dependency structures from derivation trees are also proposed. At present, this is the first Vietnamese parsing system capable of producing both constituency and dependency analyses with encouraging performances: 69.33% and 73.21% for constituency and dependency analysis accuracy, respectively.

Strong Lexicalization of Tree Adjoining Grammars

Abstract: Recently, it was shown (KUHLMANN, SATTA: Tree-adjoining grammars are not closed under strong lexicalization Comput Linguist, 2012) that finitely ambiguous tree adjoining grammars cannot be transformed into a normal form (preserving the generated tree language), in which each production contains a lexical symbol A more powerful model, the simple context-free tree grammar, admits such a normal form It can be effectively constructed and the maximal rank of the nonterminals only increases by 1 Thus, simple context-free tree grammars strongly lexicalize tree adjoining grammars and themselves

Integration of the dual approaches in the distributional learning of context-free grammars

Abstract: Recently several "distributional learning algorithms" have been proposed and have made great success in learning different subclasses of context-free grammars. The distributional learning models and exploits the relation between strings and contexts that form grammatical sentences in the language of the learning target. There are two main approaches. One, which we call primal, constructs nonterminals whose language is supposed to be characterized by strings. The other, which we call dual, uses contexts to characterize the language of each nonterminal of the conjecture grammar. This paper shows how those opposite approaches are integrated into single learning algorithms that learn quite rich classes of context-free grammars.

On Watson-Crick automata

Abstract: Motivated by Watson-Crick automata, we introduce here a grammar counterpart called a Watson-Crick regular grammar that has rules as in a regular grammar but involves double stranded strings. The language generated by this grammar consists of strings in the upper strands of the double stranded strings related by a complementarity relation. We obtain some properties of these grammars and also indicate an application to generation of chain code pictures.

A Formal Model for Plausible Dependencies in Lexicalized Tree Adjoining Grammar

Abstract: Several authors have pointed out that the correspondence between LTAG derivation trees and dependency structures is not as direct as it may seem at first glance, and various proposals have been made to overcome this divergence. In this paper we propose to view the correspondence between derivation trees and dependency structures as a tree transformation during which the direction of some of the original edges is reversed. We show that, under this transformation, LTAG is able to induce both ill-nested dependency trees and dependency trees with gap-degree greater than 1, which is not possible under the direct reading of derivation trees as dependency trees.

Synchronous Context-Free Tree Grammars

Abstract: We consider pairs of context-free tree grammars combined through synchronous rewriting The resulting formalism is at least as powerful as synchronous tree adjoining grammars and linear, nondeleting macro tree transducers, while the parsing complexity remains polynomial Its power is subsumed by context-free hypergraph grammars The new formalism has an alternative characterization in terms of bimorphisms An advantage over synchronous variants of linear context-free rewriting systems is the ability to specify tree-to-tree transductions

One-sided forbidding grammars and selective substitution grammars

Abstract: In one-sided forbidding grammars, the set of rules is divided into the set of left forbidding rules and the set of right forbidding rules. A left forbidding rule can rewrite a non-terminal if each of its forbidding symbols is absent to the left of the rewritten symbol in the current sentential form, while a right forbidding rule is applied analogically except that this absence is verified to the right. Apart from this, they work like ordinary forbidding grammars. As its main result, this paper proves that one-sided forbidding grammars are equivalent to selective substitution grammars. This equivalence is established in terms of grammars with and without erasing rules. Furthermore, this paper proves that one-sided forbidding grammars in which the set of left forbidding rules coincides with the set of right forbidding rules characterize the family of context-free languages. In the conclusion, the significance of the achieved results is discussed.

Termination Analysis for Higher-Order Attribute Grammars

Abstract: This paper describes a conservative analysis to detect non-termination in higher-order attribute grammar evaluation caused by the creation of an unbounded number of (finite) trees as local tree-valued attributes, which are then themselves decorated with attributes. This type of non-termination is not detected by a circularity analysis for higher-order attribute grammars. The analysis presented here uses term rewrite rules to model the creation of new trees on which attributes will be evaluated. If the rewrite rules terminate then only a finite number of trees will be created. To handle higher-order inherited attributes, the analysis places an ordering on non-terminals to schedule their use and ensure a finite number of passes over the generated trees. When paired with the traditional completeness and circularity analyses and the assumption that each attribute equation defines a terminating computation, this analysis can be used to show that attribute grammar evaluation will terminate normally. This analysis is applicable to a wide range of common attribute grammar idioms and has been used to show that evaluation of our specification of Java 1.4 terminates.

Tree-Adjoining Grammars Are Not Closed Under Strong Lexicalization

Abstract: A lexicalized tree-adjoining grammar is a tree-adjoining grammar where each elementary tree contains some overt lexical item. Such grammars are being used to give lexical accounts of syntactic phenomena, where an elementary tree defines the domain of locality of the syntactic and semantic dependencies of its lexical items. It has been claimed in the literature that for every tree-adjoining grammar, one can construct a strongly equivalent lexicalized version. We show that such a procedure does not exist: Tree-adjoining grammars are not closed under strong lexicalization.

One-Sided Random Context Grammars with Leftmost Derivations

Abstract: In this paper, we study the generative power of one-sided random context grammars working in a leftmost way. More specifically, by analogy with the three well-known types of leftmost derivations in regulated grammars, we introduce three types of leftmost derivations to one-sided random context grammars and prove the following three results. (I) One-sided random context grammars with type-1 leftmost derivations characterize the family of context-free languages. (II) One-sided random context grammars with type-2 and type-3 leftmost derivations characterize the family of recursively enumerable languages. (III) Propagating one-sided random context grammars with type-2 and type-3 leftmost derivations characterize the family of context-sensitive languages. In the conclusion, the generative power of random context grammars and one-sided random context grammars with leftmost derivations is compared.

Defining contexts in context-free grammars

Abstract: Conjunctive grammars (Okhotin, 2001) are an extension of the standard context-free grammars with a conjunction operation, which maintains most of their practical properties, including many parsing algorithms. This paper introduces a further extension to the model, which is equipped with quantifiers for referring to the left context, in which the substring being defined does occur. For example, a rule A → a & ◺B defines a string a, as long as it is preceded by any string defined by B. The paper gives two equivalent definitions of the model--by logical deduction and by language equations--and establishes its basic properties, including a transformation to a normal form, a cubic-time parsing algorithm, and another recognition algorithm working in linear space.

Object Grammars: Compositional & Bidirectional Mapping Between Text and Graphs

Abstract: Object Grammars define mappings between text and object graphs. Parsing recognizes syntactic features and creates the corresponding object structure. In the reverse direction, formatting recognizes object graph features and generates an appropriate textual presentation. The key to Object Grammars is the expressive power of the mapping, which decouples the syntactic structure from the graph structure. To handle graphs, Object Grammars support declarative annotations for resolving textual names that refer to arbitrary objects in the graph structure. Predicates on the semantic structure provide additional control over the mapping. Furthermore, Object Grammars are compositional so that languages may be defined in a modular fashion. We have implemented our approach to Object Grammars as one of the foundations of the Ensō system and illustrate the utility of our approach by showing how it enables definition and composition of domain-specific languages (DSLs).

A Logic for Categorial Grammars: Lambek’s Syntactic Calculus

Abstract: Our second chapter is a rather complete study of the Lambek calculus, which enables a completely logical treatment of categorial grammar.

Logical grammars, logical theories

Abstract: Residuated lattices form one of the theoretical backbones of the Lambek Calculus as the standard free models. They also appear in grammatical inference as the syntactic concept lattice, an algebraic structure canonically defined for every language L based on the lattice of all distributionally definable subsets of strings. Recent results show that it is possible to build representations, such as context-free grammars, based on these lattices, and that these representations will be efficiently learnable using distributional learning. In this paper we discuss the use of these syntactic concept lattices as models of Lambek grammars, and use the tools of algebraic logic to try to link the proof theoretic ideas of the Lambek calculus with the more algebraic approach taken in grammatical inference. We can then reconceive grammars of various types as equational theories of the syntactic concept lattice of the language. We then extend this naturally from models based on concatenation of strings, to ones based on concatenations of discontinuous strings, which takes us from context-free formalisms to mildly context sensitive formalisms (multiple context-free grammars) and Morrill's displacement calculus.

On shape grammars, color grammars and sortal grammars: a sortal grammar interpreter for varying shape grammar formalisms

Abstract: Grammar formalisms for design come in a large variety, requiring different representations of the entities being generated, and different interpretative mechanisms for this generation. Most examples of shape grammars rely on labeled shapes, a combination of line segments and labeled points. Color grammars extend the shape grammar formalism to allow for a variety of qualitative aspects of design, such as color, to be integrated in the rules of a shape grammar. Sortal grammars consider a compositional approach to the representational structures underlying (augmented) shape grammars, allowing for a variety of grammar formalism to be defined and explored. In this paper, we revisit and explore an exemplar shape grammar from literature to illustrate the use of different grammar formalisms and consider the implementation of rule application within a sortal grammar interpreter

Exi decoder and computer readable medium

Abstract: There is provided with an EXI decoder, including: a grammar store storing a first set of type grammars and a second set of type grammars, the first set of type grammars being type grammars generated according to an EXI specification from a basic schema of an XML and the second set of type grammars being type grammars that, among a set of type grammars generated according to the EXI specification from an extension schema of XML, type grammars common to the first set of type grammars are excluded; a stream input unit to receive an EXI stream; and a parser unit decoding the EXI stream, when the EXI stream is compatible with the basic schema, based on the first set of type grammars, and, when the EXI stream is compatible with the extension schema, based on the second set of type grammars and the common type grammars.

Beyond Semilinearity: Distributional Learning of Parallel Multiple Context-free Grammars

Abstract: Semilinearity is widely held to be a linguistic invariant but, controversially, some linguistic phenomena in languages like Old Georgian and Yoruba seem to violate this constraint. In this paper we extend distributional learning to the class of parallel multiple context-free grammars, a class which as far as is known includes all attested natural languages, even taking an extreme view on these examples. These grammars may have a copying operation that can recursively copy constituents, allowing them to generate non-semilinear languages. We generalise the notion of a context to a class of functions that include copying operations. The congruential approach is ineective at this level of the hierarchy; accordingly we extend this using dual approaches, dening nonterminals using sets of these generalised contexts. As a corollary we also extend the multiple context free grammars using the lattice based approaches.

Estimating Compact Yet Rich Tree Insertion Grammars

Abstract: We present a Bayesian nonparametric model for estimating tree insertion grammars (TIG), building upon recent work in Bayesian inference of tree substitution grammars (TSG) via Dirichlet processes. Under our general variant of TIG, grammars are estimated via the Metropolis-Hastings algorithm that uses a context free grammar transformation as a proposal, which allows for cubic-time string parsing as well as tree-wide joint sampling of derivations in the spirit of Cohn and Blunsom (2010). We use the Penn treebank for our experiments and find that our proposal Bayesian TIG model not only has competitive parsing performance but also finds compact yet linguistically rich TIG representations of the data.

Synchronous versions of regulated grammars : generative power and linguistic applications

Abstract: This paper introduces the notion of new synchronous grammars as systems consisting of two context-free grammars with linked rules instead of linked nonterminals Further, synchronous versions of regulated grammars, specifically, matrix grammars and scattered context grammars, are discussed From a theoretical point of view, this paper discusses the power of these synchronous grammars It demonstrates the following main results First, if we synchronize context-free grammars by linking rules, the grammar generates the languages defined by matrix grammars Second, if we synchronize matrix grammars by linking matrices, the generative power remains unchanged Third, synchronous scattered context grammars generate the class of recursively enumerable languages From a more practical viewpoint, this paper presents linguistic application prospects The focus is on natural language translation between Japanese and English