Showing 1 to 100 of 301 matching Articles
Results per page:
Export (CSV)
By
Wu, Dekai
9 Citations
We offer a perspective on EBMT from a statistical MT standpoint, by developing a threedimensional MT model space based on three pairs of definitions: (1) logical versus statistical MT, (2) schemabased versus examplebased MT, and (3) lexical versus compositional MT. Within this space we consider the interplay of three key ideas in the evolution of transfer, examplebased, and statistical approaches to MT. We depict how all translation models face these issues in one way or another, regardless of the school of thought, and suggest where the real questions for the future may lie.
more …
By
Lambert, Patrik; Gispert, Adrià; Banchs, Rafael; Mariño, José B.
Show all (4)
9 Citations
The purpose of this paper is to provide guidelines for building a word alignment evaluation scheme. The notion of word alignment quality depends on the application: here we review standard scoring metrics for full text alignment and give explanations on how to use them better. We discuss strategies to build a reference corpus, and show that the ratio between ambiguous and unambiguous links in the reference has a great impact on scores measured with these metrics. In particular, automatically computed alignments with higher precision or higher recall can be favoured depending on the value of this ratio. Finally, we suggest a strategy to build a reference corpus particularly adapted to applications where recall plays a significant role, like in machine translation. The manually aligned corpus we built for the SpanishEnglish European Parliament corpus is also described. This corpus is freely available.
more …
By
Carl, Michael
2 Citations
According to the system theory of von Bertalanffy (1968), Bertalanffy, a “system” is an entity that can be distinguished from its environment and that consists of several parts. System theory investigates the role of the parts, their interaction and the relation of the whole with its environment. System theory of the second order examines how an observer relates to the system. This paper traces some of the recent discussion of examplebased machine translation (EBMT) and compares a number of EBMT and statistical MT systems. It is found that translation examples are linguistic systems themselves that consist of words, phrases and other constituents. Two properties of Luhmann’s (2002) system theory are discussed in this context: EBMT has focussed on the properties of structures suited for translation and the design of their reentry points, and SMT develops connectivity operators which select the most likely continuations of structures. While technically the SMT and EBMT approaches complement each other, the principal distinguishing characteristic results from different sets of values which SMT and EBMT followers prefer.
more …
By
Groves, Declan; Way, Andy
4 Citations
This paper presents an extended, harmonised account of our previous work on combining subsentential alignments from phrasebased statistical machine translation (SMT) and examplebased MT (EBMT) systems to create novel hybrid datadriven systems capable of outperforming the baseline SMT and EBMT systems from which they were derived. In previous work, we demonstrated that while an EBMT system is capable of outperforming a phrasebased SMT (PBSMT) system constructed from freely available resources, a hybrid ‘examplebased’ SMT system incorporating marker chunks and SMT subsentential alignments is capable of outperforming both baseline translation models for French–English translation. In this paper, we show that similar gains are to be had from constructing a hybrid ‘statistical’ EBMT system. Unlike the previous research, here we use the Europarl training and test sets, which are fast becoming the standard data in the field. On these data sets, while all hybrid ‘statistical’ EBMT variants still fall short of the quality achieved by the baseline PBSMT system, we show that adding the marker chunks to create a hybrid ‘examplebased’ SMT system outperforms the two baseline systems from which it is derived. Furthermore, we provide further evidence in favour of hybrid systems by adding an SMT targetlanguage model to the EBMT system, and demonstrate that this too has a positive effect on translation quality. We also show that many of the subsentential alignments derived from the Europarl corpus are created by either the PBSMT or the EBMT system, but not by both. In sum, therefore, despite the obvious convergence of the two paradigms, the crucial differences between SMT and EBMT contribute positively to the overall translation quality. The central thesis of this paper is that any researcher who continues to develop an MT system using either of these approaches will benefit further from integrating the advantages of the other model; dogged adherence to one approach will lead to inferior systems being developed.
more …
By
Lepage, Yves; Denoual, Etienne
16 Citations
We have designed, implemented and assessed an EBMT system that can be dubbed the “purest ever built”: it strictly does not make any use of variables, templates or patterns, does not have any explicit transfer component, and does not require any preprocessing or training of the aligned examples. It uses only a specific operation, proportional analogy, that implicitly neutralizes divergences between languages and captures lexical and syntactic variations along the paradigmatic and syntagmatic axes without explicitly decomposing sentences into fragments. Exactly the same genuine implementation of such a core engine was evaluated on different tasks and language pairs. To begin with, we compared our system on two tasks of a previous MT evaluation campaign to rank it among other current stateoftheart systems. Then, we illustrated the “universality” of our system by participating in a recent MT evaluation campaign, with exactly the same core engine, for a wide variety of language pairs. Finally, we studied the influence of extra data like dictionaries and paraphrases on the system performance.
more …
By
Cicekli, Ilyas
1 Citations
This paper presents a generalization technique that induces translation templates from a given set of translation examples by replacing differing parts in the examples with typed variables. Since the type of each variable is inferred during the learning process, each induced template is also associated with a set of type constraints. The type constraints that are associated with a translation template restrict the usage of the translation template in certain contexts in order to avoid some of the wrong translations. The types of variables are induced using type lattices designed for both the source and target languages. The proposed generalization technique has been implemented as a part of an examplebased machine translation system.
more …
By
Zwarts, Joost
55 Citations
The semantics of directional prepositions is investigated from the perspective of aspect. What distinguishes telic PPs (like to the house) from atelic PPs (like towards the house), taken as denoting sets of paths, is their algebraic structure: atelic PPs are cumulative, closed under the operation of concatenation, telic PPs are not. Not only does this allow for a natural and compositional account of how PPs contribute to the aspect of a sentence, but it also guides our understanding of the lexical semantics of prepositions in important ways. Semantically, prepositions turn out to be quite similar to nouns and verbs. Nominal distinctions (like singular and plural, mass and count) and verbal classes (like semelfactives and degree achievements) have their prepositional counterparts.
more …
By
Carletta, Jean; Evert, Stefan; Heid, Ulrich; Kilgour, Jonathan
Show all (4)
22 Citations
The NITE XML Toolkit (NXT) is open source software for working with language corpora, with particular strengths for multimodal and heavily crossannotated data sets. In NXT, annotations are described by types and attribute value pairs, and can relate to signal via start and end times, to representations of the external environment, and to each other via either an arbitrary graph structure or a multirooted tree structure characterized by both temporal and structural orderings. Simple queries in NXT express variable bindings for ntuples of objects, optionally constrained by type, and give a set of conditions on the ntuples combined with boolean operators. The defined operators for the condition tests allow full access to the timing and structural properties of the data model. A complex query facility passes variable bindings from one query to another for filtering, returning a tree structure. In addition to describing NXTȁ9s core data handling and search capabilities, we explain the standoff XML data storage format that it employs and illustrate its use with examples from an early adopter of the technology.
more …
By
Hutchins, John
11 Citations
In the last decade the dominant models of MT have been datadriven or corpusbased. Of the two main trends, statistical machine translation and examplebased machine translation (EBMT), the latter is much less clearly defined. In a review of the recently published collection edited by Michael Carl and Andy Way, this essay surveys the basic processes, methods, main problems and tasks of EBMT, and attempts to provide a definition of the essence of EBMT in comparison with statistical MT and traditional rulebased MT.
more …
By
Lasersohn, Peter
246 Citations
This paper argues that truth values of sentences containing predicates of “personal taste” such as fun or tasty must be relativized to individuals. This relativization is of truth value only, and does not involve a relativization of semantic content: If you say roller coasters are fun, and I say they are not, I am negating the same content which you assert, and directly contradicting you. Nonetheless, both our utterances can be true (relative to their separate contexts). A formal semantic theory is presented which gives this result by introducing an individual index, analogous to the world and time indices commonly used, and by treating the pragmatic context as supplying a particular value for this index. The context supplies this value in the derivation of truth values from content, not in the derivation of content from character. Predicates of personal taste therefore display a kind of contextual variation in interpretation which is unlike the familiar variation exhibited by pronouns and other indexicals.
more …
By
Naumov, Pavel
2 Citations
The classical propositional logic is known to be sound and complete with respect to the set semantics that interprets connectives as set operations. The paper extends propositional language by a new binary modality that corresponds to partial recursive function type constructor under the above interpretation. The cases of deterministic and nondeterministic functions are considered and for both of them semantically complete modal logics are described and decidability of these logics is established.
more …
By
Gispert, Joan; Torrens, Antoni
3 Citations
In this paper we characterize, classify and axiomatize all axiomatic extensions of the IMT3 logic. This logic is the axiomatic extension of the involutive monoidal tnorm logic given by ¬φ^{3} ∨ φ. For our purpose we study the lattice of all subvarieties of the class IMT3, which is the variety of IMTLalgebras given by the equation ¬(x^{3}) ∨ x ≈ ⊤, and it is the algebraic counterpart of IMT3 logic. Since every subvariety of IMT3 is generated by their totally ordered members, we study the structure of all IMT3chains in order to determine the lattice of all subvarieties of IMT3. Given a family of IMT3chains the number of elements of the largest odd finite subalgebra in the family and the number of elements of the largest even finite subalgebra in the family turns out to be a complete classifier of the variety generated. We obtain a canonical set of generators and a finite equational axiomatization for each subvariety and, for each corresponding logic, a finite set of characteristic matrices and a finite set of axioms.
more …
By
Bezhanishvili, Guram; Esakia, Leo; Gabelaia, David
30 Citations
We consider two topological interpretations of the modal diamond—as the closure operator (Csemantics) and as the derived set operator (dsemantics). We call the logics arising from these interpretations Clogics and dlogics, respectively. We axiomatize a number of subclasses of the class of nodec spaces with respect to both semantics, and characterize exactly which of these classes are modally definable. It is demonstrated that the dsemantics is more expressive than the Csemantics. In particular, we show that the dlogics of the six classes of spaces considered in the paper are pairwise distinct, while the Clogics of some of them coincide.
more …
By
Belluce, Lawrence P.; Nola, Antonio
We describe a class of MValgebras which is a natural generalization of the class of “algebras of continuous functions”. More specifically, we're interested in the algebra of frame maps Hom
$${_{\cal F}}$$
(Ω(A), K) in the category T of frames, where A is a topological MValgebra, Ω(A) the lattice of open sets of A, and K an arbitrary frame.
Given a topological space X and a topological MValgebra A, we have the algebra C (X, A) of continuous functions from X to A. We can look at this from a frame point of view. Among others we have the result: if K is spatial, then C(pt(K), A), pt(K) the points of K, embeds into Hom
$${_{\cal F}}$$
(Ω(A), K) analogous to the case of C (X, A) embedding into Hom
$${_{\cal F}}$$
(Ω(A), Ω (X)).
more …
By
Pambuccian, Victor
6 Citations
We show that the firstorder theory of a large class of plane geometries and the firstorder theory of their groups of motions, understood both as groups with a unary predicate singling out linereflections, and as groups acting on sets, are mutually interpretable.
more …
By
Bimbó, Katalin
8 Citations
The fixed point combinator (Y) is an important nonproper combinator, which is defhable from a combinatorially complete base. This combinator guarantees that recursive equations have a solution. Structurally free logics (LC) turn combinators into formulas and replace structural rules by combinatory ones. This paper introduces the fixed point and the dual fixed point combinator into structurally free logics. The admissibility of (multiple) cut in the resulting calculus is not provable by a simple adaptation of the similar proof for LC with proper combinators. The novelty of our proof—beyond proving the cut for a newly extended calculus–is that we add a fourth induction to the byandlarge Gentzenstyle proof.
more …
By
MirandaGarcía, A.; CalleMartÍn, J.
6 Citations
The measure of lexical repetition constitutes one of the variables used to determine the lexical richness of literary texts, a value further employed in authorship attribution studies. Although most of the constants for lexical richness actually depend on text length, Yule’s characteristic is considered to be highly reliable for being text length independent. It is not the aim of this paper questioning the validity of K to measure the lexical repeatrate, nor to evaluate its usefulness in authorship studies, but to review the most accurate procedure to calculate its value in the light of the lack of standardization found in the specific literature. At the same time, the peculiar calculation of Yule’s K by TACT is explained. Our study suggests that standardization will certainly help improve the studies where K is employed.
more …
By
Labbé, Cyril; Labbé, Dominique
2 Citations
We present a new method to describe the contextual meaning of a key word in a corpus. The vocabulary of the sentences containing this word is compared to that of the entire corpus in order to highlight the words which are significantly overutilized in the neighbourhood of this key word (they are associated in the author’s mind) and the ones which are significantly underutilized (they are mutually exclusive). This method provides an interesting tool for lexicography and literary studies as is shown by applying it to the word amour (love) in the work of Pierre Corneille, the most famous French playwright of the 17th century.
more …
By
Skvortsov, Dmitrij
4 Citations
We propose a new, rather simple and short proof of Kripkecompleteness for the predicate variant of Dummett's logic. Also a family of Kripkeincomplete extensions of this logic that are complete w.r.t. Kripke frames with equality (or equivalently, w.r.t. Kripke sheaves [8]), is described.
more …
By
Litak, Tadeusz; Wolter, Frank
6 Citations
We prove that all finitely axiomatizable tense logics with temporal operators for ‘always in the future’ and ‘always in the past’ and determined by linear fows time are coNPcomplete. It follows, for example, that all tense logics containing a density axiom of the form ■^{n+1}_{F}p → ^{n}_{F}p, for some n ≥ 0, are coNPcomplete. Additionally, we prove coNPcompleteness of all ∩irreducible tense logics. As these classes of tense logics contain many Kripke incomplete bimodal logics, we obtain many natural examples of Kripke incomplete normal bimodal logics which are nevertheless coNPcomplete.
more …
By
Popescu, Andrei
We introduce ŁukasiewiczMoisil relation algebras, obtained by considering a relational dimension over ŁukasiewiczMoisil algebras. We prove some arithmetical properties, provide a characterization in terms of complex algebras, study the connection with relational Post algebras and characterize the simple structures and the matrix relation algebras.
more …
By
Braüner, Torben
1 Citations
A hybrid logic is obtained by adding to an ordinary modal logic further expressive power in the form of a second sort of propositional symbols called nominals and by adding socalled satisfaction operators. In this paper we consider hybridized versions of S5 (“the logic of everywhere”) and the modal logic of inequality (“the logic of elsewhere”). We give natural deduction systems for the logics and we prove functional completeness results.
more …
By
Küsters, Ralf; Molitor, Ralf
The least common subsumer (lcs) of a set of concept descriptions is the most specific concept description that subsumes all of the concept descriptions in the given set. By computing the lcs, commonalities between concept descriptions can be made explicit. This is an important inference task useful in several applications, including, for instance, the bottomup construction of description logic knowledge bases.
Previous work on the lcs has concentrated on description logics that either allow for number restrictions or for existential restrictions. Many applications, however, require to combine these constructors. In this work, we present an algorithm for computing the lcs in the description logic ALEN which comprises both constructors—number and existential restrictions—as well as concept conjunction, primitive negation, and value restrictions. To prove correctness of our lcs algorithm, we develop a structural characterization of subsumption in ALEN.
more …
By
Bulińska, Maria
1 Citations
The Lambek calculus introduced in Lambek [6] is a strengthening of the type reduction calculus of Ajdukiewicz [1]. We study Associative Lambek Calculus L in Gentzen style axiomatization enriched with a finite set Γ of nonlogical axioms, denoted by L(Γ).It is known that finite axiomatic extensions of Associative Lambek Calculus generate all recursively enumerable languages (see Buszkowski [2]). Then we confine nonlogical axioms to sequents of the form p → q, where p and q are atomic types. For calculus L(Γ) we prove interpolation lemma (modifying the Roorda proof for L [10]) and the binary reduction lemma (using the Pentus method [9] with modification from [3]). In consequence we obtain the weak equivalence of the ContextFree Grammars and grammars based on L(Γ).
more …
By
Ahmed, Tarek Sayed
3 Citations
We show that not all epimorphisms are surjective in certain classes of infinite dimensional cylindric algebras, Pinter's substitution algebras and Halmos' quasipolyadic algebras with and without equality. It follows that these classes fail to have the strong amalgamation property. This answers a question in [3] and a question of Pigozzi in his landmark paper on amalgamation [9]. The cylindric case was first proved by Judit Madarasz [7]. The proof presented herein is substantially different. By a result of Németi, our result implies that the Bethdefinability Theorem fails for certain expansions of first order logic
more …
By
Artstein, Ron
6 Citations
Quantificational arguments can take scope outside of temporal adjunct clauses, in an apparent violation of locality restrictions: the sentence few secretaries cried after each executive resigned allows the quantificational NP each executive to take scope above few secretaries. I show how this scope relation is the result of local operations: the adjunct clause is a temporal generalized quantifier which takes scope over the main clause (Pratt and Francez, Linguistic and Philosophy 24(2), 187–222. [2001]), and within the adjunct clause, the quantificational argument takes scope above the implicit determiner which forms the temporal generalized quantifier. The paper explores various relations among quantificational arguments across clause boundaries, including temporal clauses that are modified internally by a temporal adverbial and temporal clauses with embedded sentential complements.
more …
By
Lambek, Joachim; Scott, Philip
We study the monoid of primitive recursive functions and investigate a onestep construction of a kind of exact completion, which resembles that of the familiar category of modest sets, except that the partial equivalence relations which serve as objects are recursively enumerable. As usual, these constructions involve the splitting of symmetric idempotents.
more …
By
Verhoeven, Liza; Horsten, Leon
2 Citations
This paper is a contribution to the program of constructing formal representations
of pragmatic aspects of human reasoning. We propose a formalization within the framework of Adaptive Logics of the exclusivity implicature governing the connective ‘or’.Keywords: exclusivity implicature, Adaptive Logics.
more …
By
Cintula, Petr
2 Citations
Compactness is an important property of classical propositional logic. It can be defined in two equivalent ways. The first one states that simultaneous satisfiability of an infinite set of formulae is equivalent to the satisfiability of all its finite subsets. The second one states that if a set of formulae entails a formula, then there is a finite subset entailing this formula as well.
In propositional manyvalued logic, we have different degrees of satisfiability and different possible definitions of entailment, hence the questions of compactness is more complex. In this paper we will deal with compactness of Gödel, Gödel_{Δ}, and Gödel_{∼} logics.
There are several results (all for the countable set of propositional variables) concerning the compactness (based on satisfiability) of these logic by Cintula and Navara, and the question of compactness (based on entailment) for Gödel logic was fully answered by Baaz and Zach (see papers [3] and [2]).
In this paper we give a nearly complete answer to the problem of compactness based on both concepts for all three logics and for an arbitrary cardinality of the set of propositional variables. Finally, we show a tight correspondence between these two concepts
more …
By
Belluce, L. Peter; Grigolia, Revaz; Lettieri, Ada
10 Citations
Representations of monadic MV algebra, the characterization of locally finite monadic MV algebras, with axiomatization of them, definability of nontrivial monadic operators on finitely generated free MV algebras are given. Moreover, it is shown that finitely generated mrelatively complete subalgebra of finitely generated free MV algebra is projective.
more …
By
Aloni, Maria
6 Citations
This article discusses pragmatic aspects of our interpretation of intensional constructions like questions and prepositional attitude reports. In the first part, it argues that our evaluation of these constructions may vary relative to the identification methods operative in the context of use. This insight is then given a precise formalization in a possible world semantics. In the second part, an account of actual evaluations of questions and attitudes is proposed in the framework of bidirectional optimality theory. Pragmatic meaning selections are explained as the result of specific rankings of potentially conflicting generation and interpretation constraints.
more …
By
Muskens, Reinhard
25 Citations
The paper shows how ideas that explain the sense of an expression as a method or algorithm for finding its reference, preshadowed in Frege’s dictum that sense is the way in which a referent is given, can be formalized on the basis of the ideas in Thomason (1980). To this end, the function that sends propositions to truth values or sets of possible worlds in Thomason (1980) must be replaced by a relation and the meaning postulates governing the behaviour of this relation must be given in the form of a logic program. The resulting system does not only throw light on the properties of sense and their relation to computation, but also shows circular behaviour if some ingredients of the Liar Paradox are added. The connection is natural, as algorithms can be inherently circular and the Liar is explained as expressing one of those. Many ideas in the present paper are closely related to those in Moschovakis (1994), but receive a considerably lighter formalization.
more …
By
Avron, Arnon
20 Citations
We investigate two large families of logics, differing from each other by the treatment of negation. The logics in one of them are obtained from the positive fragment of classical logic (with or without a propositional constant ff for “the false”) by adding various standard Gentzentype rules for negation. The logics in the other family are similarly obtained from LJ^{+}, the positive fragment of intuitionistic logic (again, with or without ff). For all the systems, we provide simple semantics which is based on nondeterministic fourvalued or threevalued structures, and prove soundness and completeness for all of them. We show that the role of each rule is to reduce the degree of nondeterminism in the corresponding systems. We also show that all the systems considered are decidable, and our semantics can be used for the corresponding decision procedures. Most of the extensions of LJ^{+} (with or without ff) are shown to be conservative over the underlying logic, and it is determined which of them are not.
more …
By
Beeson, Michael; Veroff, Robert; Wos, Larry
This article answers two questions (posed in the literature), each concerning the guaranteed existence of proofs free of double negation. A proof is free of double negation if none of its deduced steps contains a term of the formn(n(t)) for some term t, where n denotes negation. The first question asks for conditions on the hypotheses that, if satisfied, guarantee the existence of a doublenegationfree proof when the conclusion is free of double negation. The second question asks about the existence of an axiom system for classical propositional calculus whose use, for theorems with a conclusion free of double negation, guarantees the existence of a doublenegationfree proof. After giving conditions that answer the first question, we answer the second question by focusing on the Lukasiewicz threeaxiom system. We then extend our studies to infinitevalued sentential calculus and to intuitionistic logic and generalize the notion of being doublenegation free. The doublenegation proofs of interest rely exclusively on the inference rule condensed detachment, a rule that combines modus ponens with an appropriately general rule of substitution. The automated reasoning program Otter played an indispensable role in this study.
more …
By
Dunn, J. Michael; Zhou, Chunlai
16 Citations
We study an application of gaggle theory to unary negative modal operators. First we treat negation as impossibility and get a minimal logic system Ki that has a perp semantics. Dunn's kite of different negations can be dealt with in the extensions of this basic logic K_{i}. Next we treat negation as “unnecessity” and use a characteristic semantics for different negations in a kite which is dual to Dunn's original one. K_{u} is the minimal logic that has a characteristic semantics. We also show that Shramko's falsification logic FL can be incorporated into some extension of this basic logic K_{u}. Finally, we unite the two basic logics K_{i} and K_{u} together to get a negative modal logic K_{}, which is dual to the positive modal logic K_{+} in [7]. Shramko has suggested an extension of Dunn's kite and also a dual version in [12]. He also suggested combining them into a “united” kite. We give a united semantics for this united kite of negations.
more …
By
Kamide, Norihiro
5 Citations
A general Gentzenstyle framework for handling both bilattice (or strong) negation and usual negation is introduced based on the characterization of negation by a modallike operator. This framework is regarded as an extension, generalization or re finement of not only bilattice logics and logics with strong negation, but also traditional logics including classical logic LK, classical modal logic S4 and classical linear logic CL. Cutelimination theorems are proved for a variety of proposed sequent calculi including CLS (a conservative extension of CL) and CLS_{cw} (a conservative extension of some bilattice logics, LK and S4). Completeness theorems are given for these calculi with respect to phase semantics, for SLK (a conservative extension and fragment of LK and CLS_{cw}, respectively) with respect to a classicallike semantics, and for SS4 (a conservative extension and fragment of S4 and CLScw, respectively) with respect to a Kripketype semantics. The proposed framework allows for an embedding of the proposed calculi into LK, S4 and CL.
more …
By
Odintsov, Sergei P.
17 Citations
The article is devoted to the systematic study of the lattice εN4_{⊥} consisting of logics extending N4_{⊥}. The logic N4_{⊥} is obtained from paraconsistent Nelson logic N4 by adding the new constant ⊥ and axioms ⊥ → p, p → ∼ ⊥. We study interrelations between εN4_{⊥} and the lattice of superintuitionistic logics. Distinguish in εN4_{⊥} basic subclasses of explosive logics, normal logics, logics of general form and study how they are relate.
more …
By
Pearce, David; Valverde, Agustín
12 Citations
Certain extensions of Nelson's constructive logic N with strong negation have recently become important in arti.cial intelligence and nonmonotonic reasoning, since they yield a logical foundation for answer set programming (ASP). In this paper we look at some extensions of Nelson's .rstorder logic as a basis for de.ning nonmonotonic inference relations that underlie the answer set programming semantics. The extensions we consider are those based on 2element, hereandthere Kripke frames. In particular, we prove completeness for .rstorder hereandthere logics, and their minimal strong negation extensions, for both constant and varying domains. We choose the constant domain version, which we denote by QN^{c}_{5}, as a basis for de.ning a .rstorder nonmonotonic extension called equilibrium logic. We establish several metatheoretic properties of QN^{c}_{5}, including Skolem forms and Herbrand theorems and Interpolation, and show that the .rstoder version of equilibrium logic can be used as a foundation for answer set inference.
more …
By
Shramko, Yaroslav
20 Citations
We consider a logic which is semantically dual (in some precise sense of the term) to intuitionistic. This logic can be labeled as “falsification logic”: it embodies the Popperian methodology of scientific discovery. Whereas intuitionistic logic deals with constructive truth and nonconstructive falsity, and Nelson's logic takes both truth and falsity as constructive notions, in the falsification logic truth is essentially nonconstructive as opposed to falsity that is conceived constructively. We also briefly clarify the relationships of our falsification logic to some other logical systems.
more …
By
Tennant, Neil
3 Citations
I reformulate the AGMaccount of contraction (which would yield an account also of revision). The reformulation involves using introduction and elimination rules for relational notions. Then I investigate the extent to which the two main methods of partial meet contraction and safe contraction can be employed for theories closed under intuitionistic consequence.
more …
By
Vakarelov, Dimiter
7 Citations
Constructive logic with Nelson negation is an extension of the intuitionistic logic with a special type of negation expressing some features of constructive falsity and refutation by counterexample. In this paper we generalize this logic weakening maximally the underlying intuitionistic negation. The resulting system, called subminimal logic with Nelson negation, is studied by means of a kind of algebras called generalized Nlattices. We show that generalized Nlattices admit representation formalizing the intuitive idea of refutation by means of counterexamples giving in this way a counterexample semantics of the logic in question and some of its natural extensions. Among the extensions which are near to the intuitionistic logic are the minimal logic with Nelson negation which is an extension of the Johansson's minimal logic with Nelson negation and its in a sense dual version — the cominimal logic with Nelson negation. Among the extensions near to the classical logic are the well known 3valued logic of Lukasiewicz, two 12valued logics and one 48valued logic. Standard questions for all these logics — decidability, Kripkestyle semantics, complete axiomatizability, conservativeness are studied. At the end of the paper extensions based on a new connective of selfdual conjunction and an analog of the Lukasiewicz middle value ½ have also been considered.
more …
By
Wiśniewski, Andrzej; Vanackere, Guido; Leszczyńska, Dorota
4 Citations
This paper develops a new proof method for two propositional paraconsistent logics: the propositional part of Batens' weak paraconsistent logic CLuN and Schütte's maximally paraconsistent logic Φ_{v}. Proofs are de.ned as certain sequences of questions. The method is grounded in Inferential Erotetic Logic.
more …
By
Asudeh, Ash
8 Citations
This paper presents a variablefree analysis of relational nouns in Glue Semantics, within a Lexical Functional Grammar (LFG) architecture. Relational nouns and resumptive pronouns are bound using the usual binding mechanisms of LFG. Special attention is paid to the bound readings of relational nouns, how these interact with genitives and obliques, and their behaviour with respect to scope, crossover and reconstruction. I consider a puzzle that arises regarding relational nouns and resumptive pronouns, given that relational nouns can have bound readings and resumptive pronouns are just a specific instance of bound pronouns. The puzzle is why is it impossible for bound implicit arguments of relational nouns to be resumptive? The puzzle is highlighted by a wellknown variety of variablefree semantics, where pronouns and relational noun phrases are identical both in category and (base) type. I show that the puzzle also arises for an established variablebased theory. I present an analysis of resumptive pronouns that crucially treats resumptives in terms of the resource logic linear logic that underlies Glue Semantics: a resumptive pronoun is a perfectly ordinary pronoun that constitutes a surplus resource; this surplus resource requires the presence of a resumptivelicensing resource consumer, a manager resource. Manager resources properly distinguish between resumptive pronouns and bound relational nouns based on differences between them at the level of semantic structure. The resumptive puzzle is thus solved. The paper closes by considering the solution in light of the hypothesis of direct compositionality. It is argued that a directly compositional version of the theory is possible, although perhaps not desirable. The implications for direct compositionality are considered.
more …
By
Cahill, Aoife; Burke, Michael; Forst, Martin; O’donovan, Ruth; Rohrer, Christian; Genabith, Josef; Way, Andy
Show all (7)
2 Citations
Deep unification (constraint)based grammars are usually handcrafted. Scaling such grammars from fragments to unrestricted text is timeconsuming and expensive. This problem can be exacerbated in multilingual broadcoverage grammar development scenarios. Cahill et al. (2002, 2004) and O’Donovan et al. (2004) present an automatic fstructure annotationbased methodology to acquire broadcoverage, deep, LexicalFunctional Grammar (LFG) resources for English from the PennII Treebank. In this paper we show how this model can be adapted to a multilingual grammar development scenario to induce robust, widecoverage, PCFGbased LFG approximations for German from the TIGER Treebank. We show how the architecture of LFG, in particular the distinction between cstructure and fstructure representations, facilitates multilingual, treebankbased unification grammar induction, allowing us to crosslinguistically reuse the lexical extraction and parsing modules from O’Donovan et al. (2004) and Cahill et al. (2004), respectively. We evaluate our grammars against the PARC 700 Dependency Bank (King et al., 2003), against dependency structures for 2000 heldout sentences from the TIGER Corpus as well as against a handcrafted dependency gold standard for 100 TIGER trees. Currently, our resources achieve 81.79% fscore against the PARC 700, a 2.19% improvement over the best result reported for a handcrafted grammar in Kaplan et al. (2004), 74.6% against the 2000 heldout TIGER dependency structures and 71.08% against the 100sentence TIGER gold standard, with substantially improved coverage compared to handcrafted resources. We have since applied our methodology to induce widecoverage LFG resources for Chinese (Burke et al., 2004b) from the Penn Chinese Treebank (Xue et al., 2002) and for Spanish from the CAST3LB Treebank (Civit, 2003).
more …
By
Copestake, Ann; Flickinger, Dan; Pollard, Carl; Sag, Ivan A.
Show all (4)
123 Citations
Minimal recursion semantics (MRS) is a framework for computational semantics that is suitable for parsing and generation and that can be implemented in typed feature structure formalisms. We discuss why, in general, a semantic representation with minimal structure is desirable and illustrate how a descriptively adequate representation with a nonrecursive structure may be achieved. MRS enables a simple formulation of the grammatical constraints on lexical and phrasal semantics, including the principles of semantic composition. We have integrated MRS with a broadcoverage HPSG grammar.
more …
By
Smrž, Pavel
This paper presents our work on parallel grammars for Czech and Russian. As both languages belong to the Slavic language group it has been possible to employ the same syntactic analyzer based on the notion of metagrammar. Our metagrammar, first developed for Czech, has been generalized with the aim of covering the same language phenomena in Czech and Russian by a unified set of metagrammar constructs. Preliminary parsing results based on the developed metagrammar are given at the end of this article.
more …
By
Bateman, John A.; KruijffKorbayová, Ivana; Kruijff, GeertJan
5 Citations
This article reports on our experience with developing multilingual grammar resources for natural language generation (NLG). We employ a strong notion of multilinguality: (i) Grammars for different languages share their overall organization, as well as those descriptions that reflect similarities between languages and (ii) a single realization engine is used to generate with these grammars. This strong notion arises from the functionalist approach we adopt: we hypothesize that languages are likely to share communicative functions, despite possibly differing in how these functions are realized. We discuss the advantages of this view in the development of largecoverage generation grammars for a broad variety of languages.
more …
By
Kobele, Gregory M.
4 Citations
I show that adding a mechanism of feature percolation (via specifier head agreement) to Minimalist Grammars (MGs) [Stabler (1997) In Retore C. (ed.) Logical Aspects of Computational Linguistics, Springer Verlag (Lecture Notes in Computer Science 1328), NY, pp. 68–95] takes them out of the class of contextsensitive grammar formalisms. The main theorem of the paper is that adding a mechanism of feature percolation to MGs allows them to implement infinite abaci [Lambek (1961), Canadian Mathematical bulletin 4, pp. 259–302], which can simulate any Turing Machine computation. As a simple corollary, I show that, for any computable
$$f\!:\mathbb{N} \rightarrow \mathbb{N}$$
, MGs thus enhanced can generate the language
$$L_{a^{f(n)}} = \{ a^{f(n)} : n \in \mathbb{N}\}$$
.
more …
By
King, Tracy Holloway; Forst, Martin; Kuhn, Jonas; Butt, Miriam
Show all (4)
3 Citations
This paper discusses the methodology and tools applied in the Parallel Grammar project (ParGram) to support consistency and parallelism of linguistic representations across multilingual Lexical Functional Grammar (lfg) grammars. A particular issue is that the grammars in the ParGram project are developed at different international sites. The approach that was established over several years relies on (i) a technical tool for checking adherence to the bestpractice feature declaration for linguistic representations, (ii) the coordinated, systematic use of templates for expressing generalizations across lexicon entries and grammar rules, and (iii) a grammar code reviewing committee in which extensions to the existing representations are critically discussed.
more …
By
Ginzburg, Jonathan
3 Citations
One of the important challenges facing NL semantics in the early 21st century is to theoretically underpin analysis and generation of conversational interaction. I start by considering certain requirements a semantic framework needs in order to be viable for this task, with reference to a benchmark example. One fundamental requirement is the provision of an ontology which incorporates propositions, questions, and similar abstract entities. The main theme of this paper concerns the construction of such an ontology. I argue that Barwise and Perry’s approach to ontology – including its nonstandard trichotomy distinguishing between situations/events, situation types, and propositions – provides useful building blocks. I implement the construction using the type theoretic framework developed by Cooper [this Journal].
more …
By
Cooper, Robin
20 Citations
This paper is part of a broader project whose aim is to present a coherent unified approach to natural language dialogue semantics using tools from type theory. Here we explore aspects of our approach which relate to situation theory and situation semantics. We first point out a relationship between type theory and the Austinian notion of truth. We then consider how records in type theory might be used to represent situations and how dependent record types can be used to model constraints on situations. We then sketch treatments of attitude phenomena for which Barwise and Perry proposed situation semantic analyses (perception complements, belief, the Pierre puzzle) as well as two other intensional phenomena (intensional verbs and intentional identity). Finally we give a characterisation of the type theory used and a small illustrative fragment of English.
more …
By
Borthen, Kaja; Haugereid, Petter
2 Citations
This paper concerns grammatical phenomena sensitive to certain classes of nominal forms, i.e., those that encode different kinds of referential properties of the nominal. We propose a grammar component for defining and picking out such semantic classes of nominal forms within typed feature structure formalisms such as the one used in HPSG, thus aiming at standardizing the representation of such phenomena. The grammar component includes four semantic features associated with the discourse referent of a nominal, i.e., cognitive status, specificity, partitivity, and whether the nominal has a universal interpretation or not. The proposed grammar component reduces to an assumed minimum a relatively large set of features that have already been proposed in analyses of the kind of phenomena at focus here, and it is hypothesized that parts of the structure are likely to be shared among grammars for different languages.
more …
By
Nam, Seungho
2 Citations
This paper, extending Keenan’s (1987, 1988) Caseextension of generalized quantifiers, proposes a natural algebraic semantics of DPcoordination and DPcomposition. DPs in subject and nonsubject positions are uniformly identified as a caseextension of a usual generalized quantifier, and DPs with different semantic cases combine with each other to yield polyadic quantifiers. The paper proves that each set of caseextensions forms a complete atomic Boolean algebra consisting of a full set of unary quantifiers, and it further shows that natural language requires the full power of binary quantification, i.e., the full set of (Fregean) reducible and unreducible binary quantifiers. This result, Type <2> Effability, is derived from the fact that the set of all binary quantifiers can be constructed by taking the meet/join closure of the set of (composed) reducible type <2> quantifiers. This fact is illustrated with nonconstituent coordination constructions (e.g., gapping in English and Korean), whose interpretation requires arbitrary meets and joins of reducible binary quantifiers.
more …
By
Zong, Chengqing; Seligman, Mark
4 Citations
This paper argues that the time is now right to field practical Spoken Language Translation (SLT) systems. Several sorts of practical systems can be built over the next few years if system builders recognize that, at the present state of the art, users must cooperate and compromise with the programs. Further, SLT systems can be arranged on a scale, in terms of the degree of cooperation or compromise they require from users. In general, the broader the intended linguistic or topical coverage of a system, the more user cooperation or compromise it will presently require. The paper briefly discusses the component technologies of SLT systems as they relate to user cooperation and accommodation (“human factors engineering”), with examples from the authors’ work. It describes three classes of “cooperative” SLT systems which could be put into practical use during the next few years.
more …
By
McShane, Marjorie; Nirenburg, Sergei; Beale, Stephen
6 Citations
This paper describes salient aspects of the OntoSem lexicon of English, a lexicon whose semantic descriptions can either be grounded in a languageindependent ontology, rely on extraontological expressive means, or exploit a combination of the two. The variety of descriptive means, as well as the conceptual complexity of semantic description to begin with, necessitates that OntoSem lexicons be compiled primarily manually. However, once a semantic description is created for a lexeme in one language, it can be reused in others, often with little or no modification. Said differently, the challenge in building a semantic lexicon is describing semantics; once the semantics are described, it is relatively straightforward to connect given meanings to the appropriate head words in other languages. In this paper we provide a brief overview of the OntoSem lexicon and processing environment, orient our approach to lexical semantics among others in the field, and describe in more detail what we mean by the largely languageindependent lexicon. Finally, we suggest reasons why our resources might be of interest to the larger community.
more …
By
Belnap, Nuel
4 Citations
“Flat presemantics” lets each parameter of truth (etc.) be considered separately and equally, and without worrying about grammatical complications. This allows one to become a little clearer on a variety of philosophicallogical points, such as the use fulness of Carnapian tolerance and the deep relativity of truth. A more definite result of thinking in terms of flat presemantics lies in the articulation of some instructive ways of categorizing operations on meanings in purely logical terms in relation to various parame ters of truth (etc.); namely, closing vs. leaving open, local vs. translocal, and anchored vs. unanchored. Basic relations among these categories are established.
more …
By
Farulewski, Maciej
6 Citations
We study a class of finite models for the Lambek Calculus with additive conjunction and with and without empty antecedents. The class of models enables us to prove the finite model property for each of the above systems, and for some axiomatic extensions of them. This work strengthens the results of [3] where only productfree fragments of these systems are considered. A characteristic feature of this approach is that we do not rely on cut elimination in opposition to e.g. [5], [9].
more …
By
Miyazaki, Yutaka
2 Citations
In this paper, we present three main results on orthologics. Firstly, we give a sufficient condition for an orthologic to have variable separation property and show that the orthomodular logic has this property. Secondly, we show that the class of modular orthologics has an infinite descending chain. Finally we show that there exists a continuum of orthologics.
more …
By
Staruch, Bożena; Staruch, Bogdan
1 Citations
We investigate first order sentences valid in completions of a given partial algebraic structure  a partial model. We give semantic and syntactic description of the set of all sentences valid in every completion of the given partial model  first order theory of this model.
more …
By
Hershfield, Jeffrey
2 Citations
Abstract.
In his work on language John Searle favors an Austinian approach that emphasizes the speech act as the basic unit of meaning and communication, and which sees speaking a language as engaging in a rulegoverned form of behavior. He couples this with a strident opposition to cognitivist approaches that posit unconscious rule following as the causal basis of linguistic competence. In place of unconscious rule following Searle posits what he calls the Background, comprised of nonintentional (nonrepresentational) mental phenomena. I argue that these two aspects of his philosophy of language cannot be reconciled. In order to preserve his view of language as a rulegoverned activity, he must embrace the cognitivist idea of unconscious rule following. Finally, I try to show how such an accommodation would be far less traumatic to Searle’s philosophical system than it might otherwise seem.
more …
By
Humberstone, Lloyd
2 Citations
Geach’s rich paper ‘A Program for Syntax’ introduced many ideas into the arena of categorial grammar, not all of which have been given the attention they warrant in the thirty years since its first publication. Rather surprisingly, one of our findings (Section 3 below) is that the paper not only does not contain a statement of what has widely come to be known as “Geach’s Rule”, but in fact presents considerations which are inimical to the adoption of the rule in question. With regard to at least some amongst the numerous other points extracted here from Geach’s discussion, we shall not be able to reach so definitive a conclusion, and content ourselves with giving the issues an airing.
more …
By
Predelli, Stefano
27 Citations
This essay aims at neutralizing the contextualist challenge against traditional semantics. According to contextualism, utterances of nonelliptical, nonambiguous, and nonindexical sentences may be associated with contrasting truthconditions. In this essay, I grant the contextualist analysis of the sentences in question, and the contextualist assessment of the truthconditions for the corresponding utterances. I then argue that the resulting situation is by no means incompatible with the traditional approach to semantics, and that the evidence put forth by the contextualists may easily be taken into account by the customary treatment of natural languages.
more …
By
Booth, Richard; Richter, Eva
6 Citations
We look at the problem of revising fuzzy belief bases, i.e., belief base revision in which both formulas in the base as well as revisioninput formulas can come attached with varying degrees. Working within a very general framework for fuzzy logic which is able to capture certain types of uncertainty calculi as well as truthfunctional fuzzy logics, we show how the idea of rational change from “crisp” base revision, as embodied by the idea of partial meet (base) revision, can be faithfully extended to revising fuzzy belief bases. We present and axiomatise an operation of partial meet fuzzy base revision and illustrate how the operation works in several important special instances of the framework. We also axiomatise the related operation of partial meet fuzzy base contraction.
more …
By
Tzouvaras, Athanassios
3 Citations
We show that the notion of cardinality of a set is independent from that of wellordering, and that reasonable total notions of cardinality exist in every model of ZF where the axiom of choice fails. Such notions are either definable in a simple and natural way, or nondefinable, produced by forcing. Analogous cardinality notions exist in nonstandard models of arithmetic admitting nontrivial automorphisms. Certain motivating phenomena from quantum mechanics are also discussed in the Appendix.
more …
By
Estival, Dominique
3 Citations
The Language Translation Interface (LTI) is a prototype developed for the Australian Defence Organisation. The aim is to provide a single, simple, interface to a variety of MT tools and utilities for personnel who need to produce translations when they have no easy access to human translators. This paper describes the functionalities of the LTI and reports on our experience with users during development. The LTI has been demonstrated and trialled at several military exercises and the feedback received is now leading to the development of the Language Translation Tools Suite (LTTS)
more …
By
Wiebe, Janyce; Wilson, Theresa; Cardie, Claire
412 Citations
This paper describes a corpus annotation project to study issues in the manual annotation of opinions, emotions, sentiments, speculations, evaluations and other private states in language. The resulting corpus annotation scheme is described, as well as examples of its use. In addition, the manual annotation process and the results of an interannotator agreement study on a 10,000sentence corpus of articles drawn from the world press are presented.
more …
By
Pustejovsky, James; Knippen, Robert; Littman, Jessica; Saurí, Roser
Show all (4)
34 Citations
In this paper, we discuss the role that temporal information plays in natural language text, specifically in the context of question answering systems. We define a descriptive framework with which we can examine the temporally sensitive aspects of natural language queries. We then investigate broadly what properties a general specification language would need, in order to mark up temporal and event information in text. We present a language, TimeML, which attempts to capture the richness of temporal and event related information in language, while demonstrating how it can play an important part in the development of more robust question answering systems.
more …
By
Setzer, Andrea; Gaizauskas, Robert; Hepple, Mark
1 Citations
In this paper we argue for the importance of doing inference over the information expressed by the annotations of temporally annotated corpora. We describe the process of inferential closure which can be applied to determine the full temporal content that follows from an annotation. We illustrate the importance of temporal inference and temporal closure in relation to three tasks, which are: (a) the comparison of different temporal annotations, (b) facilitating the manual annotation process needed to create temporally annotated corpora and (c) empirical investigations done over temporally annotated data.
more …
By
Holmes, M. Randall
2 Citations
Andrzej Kisielewicz has proposed three systems of “double extension set theory” of which we have shown two to be inconsistent in an earlier paper. Kisielewicz presented an argument that the remaining system interprets ZF, which is defective: it actually shows that the surviving possibly consistent system of double extension set theory interprets ZF with Separation and Comprehension restricted to Δ_{0} formulas. We show that this system does interpret ZF, using an analysis of the structure of the ordinals.
more …
By
KAUFMANN, STEFAN
26 Citations
ABSTRACT
The connection between the probabilities of conditionals and the corresponding conditional probabilities has long been explored in the philosophical literature, but its implementation faces both technical obstacles and objections on empirical grounds. In this paper I ?rst outline the motivation for the probabilistic turn and Lewis’ triviality results, which stand in the way of what would seem to be its most straightforward implementation. I then focus on Richard Jeffrey’s ’randomvariable’ approach, which circumvents these problems by giving up the notion that conditionals denote propositions in the usual sense. Even so, however, the randomvariable approach makes counterintuitive predictions in simple cases of embedded conditionals. I propose to address this problem by enriching the model with an explicit representation of causal dependencies. The addition of such causal information not only remedies the shortcomings of Jeffrey’s conditional, but also opens up the possibility of a uni?ed probabilistic account of indicative and counterfactual conditionals.
more …
By
Szatkowski, Miroslaw
2 Citations
The aim of this paper is to prove strong completeness theorems for several Andersonlike variants of Gödel’s theory wrt. classes of modal structures, in which: (i). 1st order terms order receive only rigid extensions in the constant objectual 1st order domain; (ii). 2nd order terms receive nonrigid extensions in preselected worldrelative objectual domains of 2nd order and rigid intensions in the constant conceptual 2nd order domain.
more …
By
Miller, Joseph S.; Moss, Lawrence S.
28 Citations
In dynamic epistemic logic and other fields, it is natural to consider relativization as an operator taking sentences to sentences. When using the ideas and methods of dynamic logic, one would like to iterate operators. This leads to iterated relativization. We are also concerned with the transitive closure operation, due to its connection to common knowledge. We show that for three fragments of the logic of iterated relativization and transitive closure, the satisfiability problems are fi1 Σ^{1}_{1}–complete. Two of these fragments do not include transitive closure. We also show that the question of whether a sentence in these fragments has a finite (tree) model is fi0 Σ^{0}_{1}–complete. These results go via reduction to problems concerning domino systems.
more …
By
Cocchiarella, Nino B.
4 Citations
ABSTRACT
Bertrand Russell introduced several novel ideas in his 1903 Principles of Mathematics that he later gave up and never went back to in his subsequent work. Two of these are the related notions of denoting concepts and classes as many. In this paper we reconstruct each of these notions in the framework of conceptual realism and connect them through a logic of names that encompasses both proper and common names, and among the latter, complex as well as simple common names. Names, proper or common, and simple or complex, occur as parts of quantifier phrases, which in conceptual realism stand for referential concepts, i.e., cognitive capacities that inform our speech and mental acts with a referential nature and account for the intentionality, or directedness, of those acts. In Russell’s theory, quantifier phrases express denoting concepts (which do not include proper names). In conceptual realism, names, as well as predicates, can be nominalized and allowed to occur as "singular terms", i.e., as arguments of predicates. Occurring as a singular term, a name denotes, if it denotes at all, a class as many, where, as in Russell’s theory, a class as many of one object is identical with that one object, and a class as many of more than one object is a plurality, i.e., a plural object that we call a group. Also, as in Russell’s theory, there is no empty class as many. When nominalized, proper names function as "singular terms" just the way they do in socalled free logic. Leśniewski’s ontology, which is also called a logic of names can be completely interpreted within this conceptualist framework, and the wellknown oddities of Leśniewski’s system are shown not to be odd at all when his system is so interpreted. Finally, we show how the pluralities, or groups, of the logic of classes as many can be used as the semantic basis of plural reference and predication. We explain in this way Russell’s "fundamental doctrine upon which all rests", i.e., "the doctrine that the subject of a proposition may be plural, and that such plural subjects are what is meant by classes [as many] which have more than one term" (Russell 1938, p. 517).
more …
By
Langlais, Philippe; Gandrabur, Simona; Leplus, Thomas; Lapalme, Guy
Show all (4)
2 Citations
Machine Translation (MT) is the focus of extensive scientific investigations driven by regular evaluation campaigns, but which are mostly oriented towards a somewhat particular task: translating news articles into English. In this paper, we investigate how well current MT approaches deal with a realworld task. We have rationally reconstructed one of the only MT systems in daily use which produces highquality translation: the Météo system. We show how a combination of a sentencebased memory approach, a phrasebased statistical engine and a neuralnetwork rescorer can give results comparable to those of the current system. We also explore another possible prospect for MT technology: the translation of weather alerts, which are currently being translated manually by translators at the Canadian Translation Bureau.
more …
By
Chung, Youjin; Lee, JongHyeok
1 Citations
Most previous corpusbased approaches to the resolution of wordsense ambiguity have collected lexical information from the context of the word to be disambiguated. However, they suffer from the problem of data sparseness. To address this problem, this paper proposes a disambiguation method using cooccurring concept codes (CCCs). The use of conceptcode features and conceptcode generalization effectively alleviate the data sparseness problem and also reduce the number of features to a practical size without any loss in system performance. We prove the effectiveness of the CCC features and the conceptcode generalization by experimental evaluations. The proposed disambiguation method is applied to a KoreantoJapanese MT system that experimented with various machinelearning techniques. In a lexical sample evaluation, our CCCbased method achieved a precision of 82.00%, with an 11.83% improvement over the baseline. Also, it achieved a precision of 83.51% in an experiment on real text, which shows that our proposed method is very useful for practical MT systems.
more …
By
Way, Andy; Gough, Nano
5 Citations
This paper presents an extended, harmonised account of our previous work on integrating controlled language data in an Examplebased Machine Translation system. Gough and Way in MT Summit pp. 133–140 (2003) focused on controlling the output text in a novel manner, while Gough and Way (9th Workshop of the EAMT, (2004a), pp. 73–81) sought to constrain the input strings according to controlled language specifications. Our original subsentential alignment algorithm could deal only with 1:1 matches, but subsequent refinements enabled n:m alignments to be captured. A direct consequence was that we were able to populate the system’s databases with more than six times as many potentially useful fragments. Together with two simple novel improvements – correcting a small number of mistranslations in the lexicon, and allowing multiple translations in the lexicon – translation quality improves considerably. We provide detailed automatic and human evaluations of a number of experiments carried out to test the quality of the system. We observe that our system outperforms the rulebased online system Logomedia on a range of automatic evaluation metrics, and that the ‘best’ translation candidate is consistently highly ranked by our system. Finally, we note in a number of tests that the BLEU metric gives objectively different results than other automatic evaluation metrics and a manual evaluation. Despite these conflicting results, we observe a preference for controlling the source data rather than the target translations.
more …
