Phrase linguistic classification and generalization for improving statistical machine translation

In this paper a method to incorporate linguistic information regarding single-word and compound verbs is proposed, as a first step towards an SMT model based on linguistically-classified phrases. By substituting these verb structures by the base form of the head verb, we achieve a better statistical word alignment performance, and are able to better estimate the translation model and generalize to unseen verb forms during translation. Preliminary experiments for the English - Spanish language pair are performed, and future research lines are detailed. © 2005 Association for Computational Linguistics.

A Structured Prediction Approach for Statistical Machine Translation

We propose a new formally syntax-based method for statistical machine translation. Transductions between parsing trees are transformed into a problem of sequence tagging, which is then tackled by a search- based structured prediction method. This allows us to automatically acquire transla- tion knowledge from a parallel corpus without the need of complex linguistic parsing. This method can achieve compa- rable results with phrase-based method (like Pharaoh), however, only about ten percent number of translation table is used. Experiments show that the structured pre- diction approach for SMT is promising for its strong ability at combining words.

A Computational Model for the Acquisition and Use of Phonological Knowledge

Does knowledge of language consist of symbolic rules? How do children learn and use their linguistic knowledge? To elucidate these questions, we present a computational model that acquires phonological knowledge from a corpus of common English nouns and verbs. In our model the phonological knowledge is encapsulated as boolean constraints operating on classical linguistic representations of speech sounds in term of distinctive features. The learning algorithm compiles a corpus of words into increasingly sophisticated constraints. The algorithm is incremental, greedy, and fast. It yields one-shot learning of phonological constraints from a few examples. Our system exhibits behavior similar to that of young children learning phonological knowledge. As a bonus the constraints can be interpreted as classical linguistic rules. The computational model can be implemented by a surprisingly simple hardware mechanism. Our mechanism also sheds light on a fundamental AI question: How are signals related to symbols?

Sparse Representations for Fast, One-Shot Learning

Humans rapidly and reliably learn many kinds of regularities and generalizations. We propose a novel model of fast learning that exploits the properties of sparse representations and the constraints imposed by a plausible hardware mechanism. To demonstrate our approach we describe a computational model of acquisition in the domain of morphophonology. We encapsulate phonological information as bidirectional boolean constraint relations operating on the classical linguistic representations of speech sounds in term of distinctive features. The performance model is described as a hardware mechanism that incrementally enforces the constraints. Phonological behavior arises from the action of this mechanism. Constraints are induced from a corpus of common English nouns and verbs. The induction algorithm compiles the corpus into increasingly sophisticated constraints. The algorithm yields one-shot learning from a few examples. Our model has been implemented as a computer program. The program exhibits phonological behavior similar to that of young children. As a bonus the constraints that are acquired can be interpreted as classical linguistic rules.

Linguistic hedges for ant-generated rules

M. Galea and Q. Shen. Linguistic hedges for ant-generated rules. Proceedings of the 15th International Conference on Fuzzy Systems, pages 9105-9112, 2006.

Simultaneous ant colony optimisation algorithms for learning linguistic fuzzy rules

M. Galea and Q. Shen. Simultaneous ant colony optimisation algorithms for learning linguistic fuzzy rules. A. Abraham, C. Grosan and V. Ramos (Eds.), Swarm Intelligence in Data Mining, pages 75-99.

A Corpus of Latin Inscriptions of the Roman Empire containing Celtic personal names

Raybould, M. and Sims-Williams, P. (2007). A Corpus of Latin Inscriptions of the Roman Empire containing Celtic personal names. Aberystwyth: CMCS publications. RAE2008

A role for LH in the regulation of expression of mRNAs encoding components of the insulin-like growth factor (IGF) system in the ovine corpus luteum

P.M. Hastie and W. Haresign (2006). A role for LH in the regulation of expression of mRNAs encoding components of the insulin-like growth factor (IGF) system in the ovine corpus luteum. Animal Reproduction Science, 96(1-2), 196-209. Sponsorship: DEFRA RAE2008

An ambiguity-based theory of the linguistic verbal joke in English

Instytut Filologii Angielskiej

Exploitation of linguistic ambiguity in Polish and English jokes

English & Polish jokes based on linguistic ambiguity are constrasted. Linguistic ambiguity results from a multiplicity of semantic interpretations motivated by structural pattern. The meanings can be "translated" either by variations of the corresponding minimal strings or by specifying the type & extent of modification needed between the two interpretations. C. F. Hockett's (1972) translatability notion that a joke is linguistic if it cannot readily be translated into other languages without losing its humor is used to interpret some cross-linguistic jokes. It is claimed that additional intralinguistic criteria are needed to classify jokes. By using a syntactic representation, the humor can be explained & compared cross-linguistically. Since the mapping of semantic values onto lexical units is highly language specific, translatability is much less frequent with lexical ambiguity. Similarly, phonological jokes are not usually translatable. Pragmatic ambiguity can be translated on the basis of H. P. Grice's (1975) cooperative principle of conversation that calls for discourse interpretations. If the distinction between linguistic & nonlinguistic jokes is based on translatability, pragmatic jokes must be excluded from the classification. Because of their universality, pragmatic jokes should be included into the linguistic classification by going beyond the translatability criteria & using intralinguistic features to describe them.

Ambiguity-generating devices in linguistic verbal jokes

Unpublished paper, written in 1996.

Pozyskiwanie danych leksykalnych z tekstówelektronicznych (na materiale czasopisma naukowego)

Praca dotyczy wybranych metod pozyskiwania, czyli ekscerpcji, informacji o charakterze leksykalnym z elektronicznych zbiorów tekstów.Jej celem jest, po pierwsze, sformułowanie nowych, oryginalnych metod, które mogą być użyteczne w pozyskiwaniu materiału do analiz leksykalnych, a następnie zbadanie ich na wybranym zbiorze tekstów.Planowano opracowanie metod niewymagających zaawansowanej znajomości programowania komputerowego, a jednocześnie umożliwiających uzyskanie wartościowych wyników, gdzie za wartościowość metody uznaje się daną wydajność ekscerpcyjną. Trzy sformułowane metody dopracowano i zoptymalizowano.Metoda ekscerpcji jednostek nowych dostarczyła ponad 1000 wyrazów nowych, niezarejestrowanych, metoda ekscerpcji kolokacji w oparciu o akronimy daje ponad 6000 jednostek, zaś metoda ekscerpcji kolokacji wykorzystująca końcówkę liczby mnogiej dała ponad 110 tysięcy wyodrębnionych jednostek.

Exploiting Phonological Constraints for Handshape Inference in ASL Video

Handshape is a key articulatory parameter in sign language, and thus handshape recognition from signing video is essential for sign recognition and retrieval. Handshape transitions within monomorphemic lexical signs (the largest class of signs in signed languages) are governed by phonological rules. For example, such transitions normally involve either closing or opening of the hand (i.e., to exclusively use either folding or unfolding of the palm and one or more fingers). Furthermore, akin to allophonic variations in spoken languages, both inter- and intra- signer variations in the production of specific handshapes are observed. We propose a Bayesian network formulation to exploit handshape co-occurrence constraints, also utilizing information about allophonic variations to aid in handshape recognition. We propose a fast non-rigid image alignment method to gain improved robustness to handshape appearance variations during computation of observation likelihoods in the Bayesian network. We evaluate our handshape recognition approach on a large dataset of monomorphemic lexical signs. We demonstrate that leveraging linguistic constraints on handshapes results in improved handshape recognition accuracy. As part of the overall project, we are collecting and preparing for dissemination a large corpus (three thousand signs from three native signers) of American Sign Language (ASL) video. The video have been annotated using SignStream® [Neidle et al.] with labels for linguistic information such as glosses, morphological properties and variations, and start/end handshapes associated with each ASL sign.

Evaluation of the quality of the management of cancer of the corpus uteri-Selection of relevant quality indicators and implementation in Belgium

Objective Describe the methodology and selection of quality indicators (QI) to be implemented in the EFFECT (EFFectiveness of Endometrial Cancer Treatment) project. EFFECT aims to monitor the variability in Quality of Care (QoC) of uterine cancer in Belgium, to compare the effectiveness of different treatment strategies to improve the QoC and to check the internal validity of the QI to validate the impact of process indicators on outcome. Methods A QI list was retrieved from literature, recent guidelines and QI databases. The Belgian Healthcare Knowledge Center methodology was used for the selection process and involved an expert's panel rating the QI on 4 criteria. The resulting scores and further discussion resulted in a final QI list. An online EFFECT module was developed by the Belgian Cancer Registry including the list of variables required for measuring the QI. Three test phases were performed to evaluate the relevance, feasibility and understanding of the variables and to test the compatibility of the dataset. Results 138 QI were considered for further discussion and 82 QI were eligible for rating. Based on the rating scores and consensus among the expert's panel, 41 QI were considered measurable and relevant. Testing of the data collection enabled optimization of the content and the user-friendliness of the dataset and online module. Conclusions This first Belgian initiative for monitoring the QoC of uterine cancer indicates that the previously used QI selection methodology is reproducible for uterine cancer. The QI list could be applied by other research groups for comparison. © 2013 Elsevier Inc.

Traitement de données issues d'un corpus écrit multilingue. Approche agile pour l'analyse du discours eurorégional

L'article présente quelques éléments de la procédure mise en place pour traiter un corpus écrit comportant 617 textes (près de 500 000 mots) relatifs aux eurorégions. Complexe et hétérogène à plusieurs titres (technique, linguistique, éditorial, générique, énonciatif), le corpus pose la difficulté majeure de l’appréhension de données multilingues (français, italien, espagnol, anglais, allemand, néerlandais). Sa manipulation a nécessité une réflexion adaptée et une démarche de modélisation que nous qualifions d’« agile » en raison de son caractère souple et itératif. La plateforme d’analyse élaborée permet de disposer de résultats utiles à l’analyse qualitative ultérieure du discours eurorégional. Elle articule un logiciel d'analyse morphosyntaxique éprouvé (TreeTagger) à des programmes (Perl) et à une base de données (SQLite) développés pour optimiser les requêtes multilingues simultanées et l’exportation automatique des résultats. Les fonctionnalités liées à la localisation contextualisée de mots- pivots, au recueil de dénominations et à la détection de segments répétés nous servent ici de guides pour exprimer les besoins de la recherche, les problèmes rencontrés et les solutions proposées. L'analyse d'observables récurrents, à savoir les notions de décision et de responsabilité, illustre le propos.