Pollard, C. and I. Sag, Information-based syntax and semantics, Technical Report 13, Center for the Study of Language and Information (1987).  Home/Search   Document Not in Database   Summary   Related Articles   Check

....Language and Information, vol. 2, 1993, pp. 1 18. Previous version as DFKI Research Report RR 91 01, January 1991. 1 Introduction Feature constraint grammars, also known as unification grammars, have become the predominant family of declarative grammar formalisms in Computational Linguistics [11, 7, 21, 20, 16]. The common assumption of these formalisms is that linguistic objects can be described by means of their features, which are functional attributes. Figure 1, for instance, shows the description of a linguistic object that may represent the sentence John sings a song . The features appear as ....

....constraint grammars, namely, sort equations. A sort equation is a pair S consisting of two feature terms. A feature algebra is a model of a set of sort equations if for every equation both sides denote the same set. Grammar rules in Functional Unification Grammar [11] and the more recent HPSG [16] are stated by means of sort equations. Figure 2 shows a simple grammar in this style (sorts start with capital letters) which generates 3 tense present Gamma Gamma Gamma Gamma Gamma Gamma Gamma Psi subj R obj ffl verb patient agent Gamma ....

C. Pollard and I. A. Sag. Information-Based Syntax and Semantics, volume 13 of CSLI Lecture Notes. Center for the Study of Language and Information, Stanford University, Cal., 1987.

.... in software configuration management, where feature logic is used to denote software versions and to deduce their mutual consistency [50, 51] The first mathematical treatment of record descriptions was the formalisms of y terms [1] In other approaches, y terms were called feature structures [40] or feature terms [46] In contrast to earlier work, the notion feature structure was mostly used for designating a record structure itself [14, 39, 42] rather than a record description. Logical descriptions of record structures lead to the notion of feature logic [25, 23, 46] When we call these ....

Carl J. Pollard and Ivan A. Sag. Information-based Syntax and Semantics, Vol. 1. Number 13 in CSLI Lecture Notes. Center for the Study of Language and Information, Stanford University, 1987. Distributed by University of Chicago Press.

.... [46, 43] It is also closely related to Bresnan and Kaplan s Lexical Functional Grammar formalism (lfg) 17] Other unification grammar formalisms such as Kay s Functional Unification Grammar (fug) 24, 25, 26, 40] Uszkoreit s Categorial Unification Grammar [49] or Pollard and Sag s hpsg [37] employ different phrase structure rules and different feature constraints. Shieber s [44] introduction to unification based approaches to grammar is an excellent survey of existing formalisms and provides the linguistic motivations our presentation is lacking. Other state of the art guides into ....

....and different feature constraints. Shieber s [44] introduction to unification based approaches to grammar is an excellent survey of existing formalisms and provides the linguistic motivations our presentation is lacking. Other state of the art guides into this fascinating area of research are [37] and [35] Johnson s thesis [16] gives a formal account of an lfg like formalism and investigates a feature constraint language with disjunctions and negations. Shieber s thesis [45] gives a rigorous formalization of the patr formalism. For instance, the word drink can be used as transitive ....

C. Pollard and I. Sag. Information-Based Syntax and Semantics, volume 13 of CSLI Lecture Notes. Center for the Study of Language and Information, Stanford University, CA, 1987.

.... uni cation [ A tKaci et al. 1993 ] A t Kaci et al. 1997 ] of feature structures (OSF terms) is achieving head corner chart based parsing [ Kay, 1989 ] Sikkel, 1997 ] Shieber et al. 1995 ] The system was tested and tuned on LinGO, Copestake et al. 1999 ] the large scale HPSG [ Pollard and Sag, 1994 ] grammar for English developed at CSLI, University of Stanford. In the Light framework, the LinGO grammar is viewed as a Light logic grammar. The aim of this paper is to give the reader the main insights on how we proceeded to prepare the LinGO grammar for the parsing with the ABC Light system, ....

....stated by the grammar writer, but used further on [ Ciortuz, 2000 ] to optimise the grammar s compiled code. Signi cant progress has been achieved during the last couple of years in the area of ecient processing with feature based grammars. Concerning parsing with largescale HPSG grammars [ Pollard and Sag, 1994 ] notably the LinGO grammar [ Copestake et al. 1999 ] for English developed at CSLI, University of Stanford, the [ Oepen et al. 2000 ] work presents some of the most advanced results. 1 In the meantime, while still under development, our ABC Light compiler, designed to do parsing with feature ....

C. Pollard and I. Sag. Headdriven Phrase Structure Grammar. Center for the Study of Language and Information, Stanford, 1994.

....whose syntactic part was developed by K. Netter (with support by J. Nerbonne) and which has an integrated semantic representation developed by J. Nerbonne and W. Kasper [Ner92] Kas93] The style of the grammar follows very much the spirit of Head Driven Phrase Structure Grammar (HPSG) PS87] PS93] However, it also incorporates insights from other grammar frameworks (e.g. categorial grammar) and extensions to the theory which are not yet part of standard HPSG. The grammar is implemented in a formalism called Type Description Language (TDL) which was developed by H. U. Krieger and U. ....

Pollard, C. / Sag, I.: Information-Based Syntax and Semantics. Vol. II: Agreement, Binding and Control. CSLI Lecture Notes. Center for the Study of Language and Information, Stanford, 1993

....grammar whose syntactic part was developed by K. Netter (with support by J. Nerbonne) and which has an integrated semantic representation developed by J. Nerbonne and W. Kasper [Ner92] Kas93] The style of the grammar follows very much the spirit of Head Driven Phrase Structure Grammar (HPSG) PS87] PS93] However, it also incorporates insights from other grammar frameworks (e.g. categorial grammar) and extensions to the theory which are not yet part of standard HPSG. The grammar is implemented in a formalism called Type Description Language (TDL) which was developed by H. U. Krieger and ....

Pollard, C. / Sag, I.: Information-Based Syntax and Semantics. Vol. I: Fundamentals. CSLI Lecture Notes, Number 13. Center for the Study of Language and Information, Stanford, 1987

....description have been more and more relocated into the lexicon of those grammars, where the basic SRL definitions are attached to single words. Two prominent variants of those grammar formalisms are Lexical Functional Grammar (LFG) BK88] and Head Driven Phrase Structure Grammar (HPSG) PS88] Parsing algorithms for those grammars are often based on chart parsing techniques ( Gor88] generalizations of the Earley and CKY algorithms for context free grammars ( Ear70, You67] The overall goal of the approaches mentioned, is to supply grammars which generalize over certain ....

C. Pollard and I. Sag. Information-based Syntax and Semantics. CLSI Lecture Notes. Center for the Study of Language and Information, Stanford, 1988.

.... : 19 2 1 Introduction Pollard Sag s seminal work on Head Driven Phrase Structure Grammar has shown that a great deal of syntax and semantics can be neatly encoded within typed feature structures, thus leading for the first time to a highly lexicalized theory of language [20, 21]. Moreover, the formalisms underlying these structures can be given a precise set theoretical semantics along the lines of Smolka and others. 1 However, there are certain areas within computational linguistics, for which, until recently, no satisfactory formulation in a uniform, constraint based ....

....a limited method of logical form simplification. This is achieved by enriching the feature logic underlying HPSG however, without sticking to external relational constraints. 11 3. 1 Encoding Logical Form Simplification In the following, we refer to Pollard and Sag s first volume of HPSG [20]. Even though the examples given throughout this section are simplified in that the structure of SEM is flat , i.e. only consists of top level attributes like OP (operator) SC (scope) CONN (connective) etc. the idea developed here can be easily adapted to more complex forms of HPSG and other ....

Carl Pollard and Ivan A. Sag. Information-Based Syntax and Semantics. Vol. I: Fundamentals. CSLI Lecture Notes, Number 13. Center for the Study of Language and Information, Stanford, 1987.

....form to the consistency problem of feature terms. This reduction shows in particular that the consistency problem remains NP hard even if only feature terms without complements and agreements are considered. Several variants of feature descriptions are being used in computational linguistics (see [22, 19, 20] for introductory expositions) In unification grammars [10, 14] unification of feature descriptions is the basic operational mechanism for parsing and generating natural language. A unification method for feature descriptions consists of a normal form that exhibits inconsistency and an algorithm ....

C. Pollard and I. Sag. Information-Based Syntax and Semantics, volume 13 of CSLI Lecture Notes. Center for the Study of Language and Information, Stanford University, CA, 1987.

.... In order to handle movement phenomena, i.e. syntactic structures which deviate from the presumed basic word order of a given natural language, in many computational approaches, the syntactic categories are annotated with additional information to allow for gap threading (see e.g. 4] [26]) Extended versions of categorial grammars, however, do not extend the definition of syntactic category but the set of inference rules, which manipulate the categories. 1.2.1 Combinatory Categorial Grammars The common denominator of the various versions of Combinatory Categorial Grammars is a ....

Carl Pollard and Ivan Sag. An Information-Based Syntax and Semantics. I. Fundamentals, volume 13 of Lecture Notes. Center for Study of Language and Information, Stanford, Ca., 1987.

....unavoidable. It is shown that the calculus is sound and complete with respect to the so called feature structure interpretations of feature terms. 1 Introduction Both in general knowledge representation (e.g. Ait Kaci Nasr s LOGIN [2] 1] and in computational linguistics (e.g. Pollard Sag [10]) the data structures for representing the terminological knowledge are usually more complex than first order terms. The main differences are the following 1. the argument positions are labeled with feature or attribute names 2. the number of arguments of a term is not fixed (open arity) 3. the ....

Carl Pollard and Ivan Sag. An Information-Based Syntax and Semantics. I. Fundamentals, volume 13 of Lecture Notes. Center for Study of Language and Information, Stanford, Ca., 1987.

.... which would cover as many as possible of the extensions to simple unification formalisms that had been proposed in the linguistic literature, especially in the fields of Head Driven Phrase 1 ESPRIT Basic Research Action 3175: Dynamic Interpretation of Natural Language Structure Grammar (HPSG, PS87] and Lexical Functional Grammar (LFG, KB82] Hence, the name CUF. These extensions included formal devices like disjunction, negation, typing feature structures, operators for set or list manipulation, constraints to express functional uncertainty, subsumption relations, or even arbitrary ....

Carl Pollard and Ivan A. Sag. Information-Based Syntax and Semantics. CSLI Lecture Notes 13. Center for the Study of Language and Information, Stanford University, 1987.

....semilattice. Types are a necessary requirement for a grammar development environment because they serve as abbreviations for lexicon entries, immediate dominance rule schemata, and universal as well as language specific principles, as is familiar from Head Driven Phrase Structure Grammar (HPSG) [16]. Types in TDL not 1 For instance, ALE employs a bottom up chart parser, whereas TFS relies entirely on type deduction. Note that recursive types can be substituted by definite clauses, as is the case for CUF, such that parsing generation roughly corresponds to SLD resolution. only serve as a ....

Carl Pollard and Ivan A. Sag. Information-Based Syntax and Semantics. Vol. I: Fundamentals. CSLI Lecture Notes, Number 13. Center for the Study of Language and Information, Stanford, 1987.

.... like TFS [ Zajac 1992 ] CUF [ Dorre and Dorna 1993 ] or TDL [ Krieger and Schafer 1994 ] Types are ordered hierarchically as is known from object oriented programming languages, a feature heavily employed in lexicalized grammar theories like Head Driven Phrase Structure Grammar (HPSG) Pollard and Sag 1987 ] This leads to multiple inheritance in the description of linguistic entities. In general, not only is a type related to other types through the inheritance hierarchy, but is also provided with feature constraints that are idiosyncratic to this type. Hence, a type symbol can serve as an ....

Carl Pollard and Ivan A. Sag. Information-Based Syntax and Semantics. Vol. I: Fundamentals. CSLI Lecture Notes, Number 13. Center for the Study of Language and Information, Stanford, 1987.

....semilattice. Types are a necessary requirement for a grammar development environment because they serve as abbreviations for lexicon entries, immediate dominance rule schemata, and universal as well as language specific principles, as is familiar from Head Driven Phrase Structure Grammar (HPSG) [16]. Types in TDL not only serve as a shorthand, like templates, but also have other advantages over templates: Structuring Linguistic Knowledge Hierarchically ordered types allow for a modular way to adequately represent linguistic knowledge. Moreover, generalizations can be made at the ....

Carl Pollard and Ivan A. Sag. Information-Based Syntax and Semantics. Vol. I: Fundamentals. CSLI Lecture Notes, Number 13. Center for the Study of Language and Information, Stanford, 1987.

....Language and Information, vol. 2, 1993, pp. 1 18. Previous version as DFKI Research Report RR 91 01, January 1991. 1 Introduction Feature constraint grammars, also known as unification grammars, have become the predominant family of declarative grammar formalisms in Computational Linguistics [11, 7, 21, 20, 16]. The common assumption of these formalisms is that linguistic objects can be described by means of their features, which are functional attributes. Figure 1, for instance, shows the description of a linguistic object that may represent the sentence John sings a song . The features appear as ....

....grammars, namely, sort equations. A sort equation is a pair S : S 0 consisting of two feature terms. A feature algebra is a model of a set of sort equations if for every equation both sides denote the same set. Grammar rules in Functional Unification Grammar [11] and the more recent HPSG [16] are stated by means of sort equations. Figure 2 shows a simple grammar in this style (sorts start with capital letters) which generates ffl tense present pred Gamma Gamma Gamma Gamma Gamma Gamma Gamma Gamma Psi subj R obj ffl verb sing patient oe ....

C. Pollard and I. A. Sag. Information-Based Syntax and Semantics, volume 13 of CSLI Lecture Notes. Center for the Study of Language and Information, Stanford University, Cal., 1987.

.... TFS [ Zajac, 1992 ] CUF [ Dorre and Dorna, 1993 ] or TDL [ Krieger and Schafer, 1994 ] Types are ordered hierarchically as it is known from object oriented programming languages, a feature heavily employed in lexicalized grammar theories like Head Driven Phrase Structure Grammar (HPSG) Pollard and Sag, 1987 ] This leads to multiple inheritance in the description of linguistic entities. In general, not only is a type related to other types through the inheritance hierarchy, but is also provided with feature constraints that are idiosyncratic to this type. Hence, a type symbol can serve as an ....

Carl Pollard and Ivan A. Sag. Information-Based Syntax and Semantics. Vol. I: Fundamentals. CSLI Lecture Notes, Number 13. Center for the Study of Language and Information, Stanford, 1987.

.... to which a lot of attention has been devoted (e.g. 15, 38, 144] Partial orders underlie central aspects of many domains, such as mathematical logic [128] sorted logic [27, 28, 93] and logic programming [3, 4, 93] type systems [106] natural language processing (e.g. typed feature structures [23, 71, 118], systemic networks [80, 101] object orientation (e.g. databases [1] languages [24] knowledge representation (e.g. conceptual structures [42, 136] knowledge bases [45] description logics [17, 18, 159] default inheritance and non monotonic reasoning [22, 85, 143, 151] machine learning ....

....93, 97, 101, 102, 104, 114] In order to empower logical terms for encoding, we developed sparse terms [51] based on an analogy to sparse matrices. There are many similarities, but also some important differences, between sparse terms and terms in LIFE [4] as well as sorted feature structures [23, 118]. Although mathematically clean, partial orders limit the representation of taxonomic knowledge to subsortsupersort (or isa) relationships. We cannot, for example, directly state that two sorts are incompatible or define one sort as the intersection of a set of other sorts. This poses problems for ....

[Article contains additional citation context not shown here]

C. Pollard and I. Sag. Information-Based Syntax and Semantics. CSLI Lecture Notes No. 13. Center for the Study of Language and Information, Stanford University, Stanford, CA, 1987.

....The parse trees for the two cases are given in Figure 1. The ambiguity is resolved by computing the consistency of the semantic attributes of the modifier total with its two possible arguments. This can be done using techniques closely related to those of contemporary unification based grammars [3, 4]. In those approaches, the syntactic consistency of two constituents is computed by unification of the feature structures (the collections of attributes and values) of the children of a node. For semantics, a more complex computation than unification may be required. In Fig. 1a, this computation ....

Pollard, C. and Sag, I.A.: Information-based Syntax and Semantics. Vol. 1. Center for the Study of Language and Information (1987).

....introduced. Some of these, like Definite Clause Grammars [PW80] Functional Unification Grammar [Kay85] and PATR II [Shi86] have been introduced primarily to offer powerful formalisms for grammar description; others like Lexical Functional Grammar [KB82] and Head Driven Phrase Structure Grammar [PS87, PS94] with the aim to provide a theory of linguistic phenomena in natural language. Unification grammars are related to attribute grammars [Knu68, Knu71] which are typically used in the field of compiler construction. There are some fundamental differences in the underlying logic, but these cannot be ....

Pollard C., Sag I.A. (1987): An Information-Based Syntax and Semantics, Vol. 1: Fundamentals. CSLI Lecture Notes 13, Center for the Study of Language and Information, Stanford University, Stanford, Ca. 39

No context found.

Pollard, C. and I. Sag, Information-based syntax and semantics, Technical Report 13, Center for the Study of Language and Information (1987).

No context found.

Pollard, C.J. and Sag, I.A. : Information-Based Syntax and Semantics, Vol. 1: Fundamentals, CSLI Lecture Notes No. 12, Center for the Study of Language and Information, Stanford University, 1988.

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