Extraction automatique d´associations lexicales à partir de corpora

The automatic acquisition of lexical associations from corpora is a crucial issue for Natural Language Processing. A lexical association is a recurrent combination of words that co-occur together more often than expected by chance in a given domain. In fact, lexical associations define linguistic phenomena such as idiomes, collocations or compound words. Due to the fact that the sense of a lexical association is not compositionnal, their identification is fundamental for the realization of analysis and synthesis that take into account all the subtilities of the language. In this report, we introduce a new statistically-based architecture that extracts from naturally occurring texts contiguous and non contiguous. For that purpose, three new concepts have been defined : the positional N-gram models, the Mutual Expectation and the GenLocalMaxs algorithm. Thus, the initial text is fisrtly transformed in a set of positionnal N-grams i.e ordered vectors of simple lexical units. Then, an association measure, the Mutual Expectation, evaluates the degree of cohesion of each positional N-grams based on the identification of local maximum values of Mutual Expectation. Great efforts have also been carried out to evaluate our metodology. For that purpose, we have proposed the normalisation of five well-known association measures and shown that both the Mutual Expectation and the GenLocalMaxs algorithm evidence significant improvements comparing to existent metodologies.

Model of a specialized bidirectional dictionary for German and Portuguese

Dissertação apresentada para cumprimento dos requisitos necessários à obtenção do grau de Mestre em Terminologia e Gestão da Informação de Especialidade

Colocaciones léxicas en dicconarios monolingües generales del español

El objetivo de la presente investigación es analizar el tratamiento que algunos de los diccionarios generales monolingües del español aparecidos en los últimos diez años han dado al fenómeno de las colocaciones léxicas, a saber, aquellas combinaciones de palabras que, desde el punto de vista de la norma, presentan ciertas restricciones combinatorias, esencialmente de carácter semántico, impuestas por el uso (Corpas: 1996). Los repertorios objeto de análisis han sido: el "Diccionario Salamanca de la lengua española", dirigido por Juan Gutiérrez (1996); el "Diccionario del español actual", de Manuel Seco, Olimpia Andrés y Gabino Ramos (1999); la vigésima segunda edición del diccionario de la RAE (2001); y el "Gran diccionario de uso del español actual. Basado en el corpus Cumbre", dirigido por Aquilino Sánchez (2001). Nuestro estudio se ha fundamentado en un corpus de 52 colocaciones léxicas confeccionado a partir del análisis de las subentradas contenidas en la letra "b" de cada uno de los diccionarios seleccionados. Posteriormente, hemos examinado las entradas correspondientes a cada uno de los elementos que constituyen la colocación (base y colocativo) con el fin de observar si los diccionarios estudiados dan cuenta de estas mismas combinaciones en otras partes del artículo lexicográfico, como son las definiciones o los ejemplos. A la hora de analizar la información lexicográfica hemos centrado nuestra atención en cuatro aspectos: a) la información contenida en las páginas preliminares de cada una de las obras; b) la ubicación de las colocaciones en el artículo lexicográfico; c) la asignación de la colocación a un artículo determinado; y d) la marcación gramatical.

Aprender verbos semánticamente similares: materiales didácticos electrónicos para estudiantes de ELE

El trabajo consiste en un análisis y la evaluación de materiales electrónicos existentes sobre parejas de verbos con significados parecidos, además de la ampliación de estos materiales y la creación de un blog para poner-los a disposición de los estudiantes de ELE

Approche cognitive de la traduction dans les langues de spécialité : vers une systématisation de la description de la conceptualisation métaphorique

Département de linguistique et de traduction

Modèle de structuration des relations lexicales fondé sur le formalisme des fonctions lexicales

Thèse réalisée en cotutelle avec l'Université Paris Diderot (Paris 7)

Évaluation de deux méthodes d’identification des liens lexicaux : méthode manuelle et méthode statistique

Ce mémoire présente une évaluation des différentes méthodes utilisées en lexicographie afin d’identifier les liens lexicaux dans les dictionnaires où sont répertoriées des collocations. Nous avons ici comparé le contenu de fiches du DiCo, un dictionnaire de dérivés sémantiques et de collocations créé selon les principes de la lexicologie explicative et combinatoire, avec les listes de cooccurrents générées automatiquement à partir du corpus Le Monde 2002. Notre objectif est ici de proposer des améliorations méthodologiques à la création de fiches de dictionnaire du type du DiCo, c’est-à-dire, des dictionnaires d’approche qualitative, où la collocation est définie comme une association récurrente et arbitraire entre deux items lexicaux et où les principaux outils méthodologiques utilisés sont la compétence linguistique de ses lexicographes et la consultation manuelle de corpus de textes. La consultation de listes de cooccurrents est une pratique associée habituellement à une approche lexicographique quantitative, qui définit la collocation comme une association entre deux items lexicaux qui est plus fréquente, dans un corpus, que ce qui pourrait être attendu si ces deux items lexicaux y étaient distribués de façon aléatoire. Nous voulons mesurer ici dans quelle mesure les outils utilisés traditionnellement dans une approche quantitative peuvent être utiles à la création de fiches lexicographiques d’approche qualitative, et de quelle façon leur utilisation peut être intégrée à la méthodologie actuelle de création de ces fiches.

Author Identification in Malayalam using n-grams

Author identification is the problem of identifying the author of an anonymous text or text whose authorship is in doubt from a given set of authors. The works by different authors are strongly distinguished by quantifiable features of the text. This paper deals with the attempts made on identifying the most likely author of a text in Malayalam from a list of authors. Malayalam is a Dravidian language with agglutinative nature and not much successful tools have been developed to extract syntactic & semantic features of texts in this language. We have done a detailed study on the various stylometric features that can be used to form an authors profile and have found that the frequencies of word collocations can be used to clearly distinguish an author in a highly inflectious language such as Malayalam. In our work we try to extract the word level and character level features present in the text for characterizing the style of an author. Our first step was towards creating a profile for each of the candidate authors whose texts were available with us, first from word n-gram frequencies and then by using variable length character n-gram frequencies. Profiles of the set of authors under consideration thus formed, was then compared with the features extracted from anonymous text, to suggest the most likely author.

Collocazioni

Encyclopedia entry on "collocations" in language

Patterns of discourse semantics: A corpus-assisted study of financial crisis in British newspaper discourse in 2009

Corpus-assisted analyses of public discourse often focus on the lexical level. This article argues in favour of corpus-assisted analyses of discourse, but also in favour of conceptualising salient lexical items in public discourse in a more determined way. It draws partly on non-Anglophone academic traditions in order to promote a conceptualisation of discourse keywords, thereby highlighting how their meaning is determined by their use in discourse contexts. It also argues in favour of emphasising the cognitive and epistemic dimensions of discourse-determined semantic structures. These points will be exemplified by means of a corpus-assisted, as well as a frame-based analysis of the discourse keyword financial crisis in British newspaper articles from 2009. Collocations of financial crisis are assigned to a generic matrix frame for ‘event’ which contains slots that specify possible statements about events. By looking at which slots are more, respectively less filled with collocates of financial crisis, we will trace semantic presence as well as absence, and thereby highlight the pragmatic dimensions of lexical semantics in public discourse. The article also advocates the suitability of discourse keyword analyses for systematic contrastive analyses of public/political discourse and for lexicographical projects that could serve to extend the insights drawn from corpus-guided approaches to discourse analysis.

SPARE-ICE: synergistic Ice Water Path from passive operational sensors

This article presents SPARE-ICE, the Synergistic Passive Atmospheric Retrieval Experiment-ICE. SPARE-ICE is the first Ice Water Path (IWP) product combining infrared and microwave radiances. By using only passive operational sensors, the SPARE-ICE retrieval can be used to process data from at least the NOAA 15 to 19 and MetOp satellites, obtaining time series from 1998 onward. The retrieval is developed using collocations between passive operational sensors (solar, terrestrial infrared, microwave), the CloudSat radar, and the CALIPSO lidar. The collocations form a retrieval database matching measurements from passive sensors against the existing active combined radar-lidar product 2C-ICE. With this retrieval database, we train a pair of artificial neural networks to detect clouds and retrieve IWP. When considering solar, terrestrial infrared, and microwave-based measurements, we show that any combination of two techniques performs better than either single-technique retrieval. We choose not to include solar reflectances in SPARE-ICE, because the improvement is small, and so that SPARE-ICE can be retrieved both daytime and nighttime. The median fractional error between SPARE-ICE and 2C-ICE is around a factor 2, a figure similar to the random error between 2C-ICE ice water content (IWC) and in situ measurements. A comparison of SPARE-ICE with Moderate Resolution Imaging Spectroradiometer (MODIS), Pathfinder Atmospheric Extended (PATMOS-X), and Microwave Surface and Precipitation Products System (MSPPS) indicates that SPARE-ICE appears to perform well even in difficult conditions. SPARE-ICE is available for public use.

Understanding inter-satellite biases of microwave humidity sounders using global simultaneous nadir overpasses

Simultaneous nadir overpasses (SNOs) of polar-orbiting satellites are most frequent in polar areas but can occur at any latitude when the equatorial crossing times of the satellites become close owing to orbital drift. We use global SNOs of polar orbiting satellites to evaluate the intercalibration of microwave humidity sounders from the more frequent high-latitude SNOs. We have found based on sensitivity analyses that optimal distance and time thresholds for defining collocations are pixel centers less than 5 km apart and time differences less than 300 s. These stringent collocation criteria reduce the impact of highly variable surface or atmospheric conditions on the estimated biases. Uncertainties in the estimated biases are dominated by the combined radiometric noise of the instrument pair. The effects of frequency changes between different versions of the humidity sounders depend on the amount of water vapor in the atmosphere. There are significant scene radiance and thus latitude dependencies in the estimated biases and this has to taken into account while intercalibrating microwave humidity sounders. Therefore the results obtained using polar SNOs will not be representative for moist regions, necessitating the use of global collocations for reliable intercalibration.

Collocating satellite-based radar and radiometer measurements — methodology and usage examples

Collocations between two satellite sensors are occasions where both sensors observe the same place at roughly the same time. We study collocations between the Microwave Humidity Sounder (MHS) on-board NOAA-18 and the Cloud Profiling Radar (CPR) on-board CloudSat. First, a simple method is presented to obtain those collocations and this method is compared with a more complicated approach found in literature. We present the statistical properties of the collocations, with particular attention to the effects of the differences in footprint size. For 2007, we find approximately two and a half million MHS measurements with CPR pixels close to their centrepoints. Most of those collocations contain at least ten CloudSat pixels and image relatively homogeneous scenes. In the second part, we present three possible applications for the collocations. Firstly, we use the collocations to validate an operational Ice Water Path (IWP) product from MHS measurements, produced by the National Environment Satellite, Data and Information System (NESDIS) in the Microwave Surface and Precipitation Products System (MSPPS). IWP values from the CloudSat CPR are found to be significantly larger than those from the MSPPS. Secondly, we compare the relation between IWP and MHS channel 5 (190.311 GHz) brightness temperature for two datasets: the collocated dataset, and an artificial dataset. We find a larger variability in the collocated dataset. Finally, we use the collocations to train an Artificial Neural Network and describe how we can use it to develop a new MHS-based IWP product. We also study the effect of adding measurements from the High Resolution Infrared Radiation Sounder (HIRS), channels 8 (11.11 μm) and 11 (8.33 μm). This shows a small improvement in the retrieval quality. The collocations described in the article are available for public use.

Methane cross-validation between three Fourier Transform Spectrometers: SCISAT ACE-FTS, GOSAT TANSO-FTS, and ground-based FTS measurements in the Canadian high Arctic

We present cross-validation of remote sensing measurements of methane profiles in the Canadian high Arctic. Accurate and precise measurements of methane are essential to understand quantitatively its role in the climate system and in global change. Here, we show a cross-validation between three datasets: two from spaceborne instruments and one from a ground-based instrument. All are Fourier Transform Spectrometers (FTSs). We consider the Canadian SCISAT Atmospheric Chemistry Experiment (ACE)-FTS, a solar occultation infrared spectrometer operating since 2004, and the thermal infrared band of the Japanese Greenhouse Gases Observing Satellite (GOSAT) Thermal And Near infrared Sensor for carbon Observation (TANSO)-FTS, a nadir/off-nadir scanning FTS instrument operating at solar and terrestrial infrared wavelengths, since 2009. The ground-based instrument is a Bruker 125HR Fourier Transform Infrared (FTIR) spectrometer, measuring mid-infrared solar absorption spectra at the Polar Environment Atmospheric Research Laboratory (PEARL) Ridge Lab at Eureka, Nunavut (80° N, 86° W) since 2006. For each pair of instruments, measurements are collocated within 500 km and 24 h. An additional criterion based on potential vorticity values was found not to significantly affect differences between measurements. Profiles are regridded to a common vertical grid for each comparison set. To account for differing vertical resolutions, ACE-FTS measurements are smoothed to the resolution of either PEARL-FTS or TANSO-FTS, and PEARL-FTS measurements are smoothed to the TANSO-FTS resolution. Differences for each pair are examined in terms of profile and partial columns. During the period considered, the number of collocations for each pair is large enough to obtain a good sample size (from several hundred to tens of thousands depending on pair and configuration). Considering full profiles, the degrees of freedom for signal (DOFS) are between 0.2 and 0.7 for TANSO-FTS and between 1.5 and 3 for PEARL-FTS, while ACE-FTS has considerably more information (roughly 1° of freedom per altitude level). We take partial columns between roughly 5 and 30 km for the ACE-FTS–PEARL-FTS comparison, and between 5 and 10 km for the other pairs. The DOFS for the partial columns are between 1.2 and 2 for PEARL-FTS collocated with ACE-FTS, between 0.1 and 0.5 for PEARL-FTS collocated with TANSO-FTS or for TANSO-FTS collocated with either other instrument, while ACE-FTS has much higher information content. For all pairs, the partial column differences are within ± 3 × 1022 molecules cm−2. Expressed as median ± median absolute deviation (expressed in absolute or relative terms), these differences are 0.11 ± 9.60 × 10^20 molecules cm−2 (0.012 ± 1.018 %) for TANSO-FTS–PEARL-FTS, −2.6 ± 2.6 × 10^21 molecules cm−2 (−1.6 ± 1.6 %) for ACE-FTS–PEARL-FTS, and 7.4 ± 6.0 × 10^20 molecules cm−2 (0.78 ± 0.64 %) for TANSO-FTS–ACE-FTS. The differences for ACE-FTS–PEARL-FTS and TANSO-FTS–PEARL-FTS partial columns decrease significantly as a function of PEARL partial columns, whereas the range of partial column values for TANSO-FTS–ACE-FTS collocations is too small to draw any conclusion on its dependence on ACE-FTS partial columns.