Una propuesta y un etiquetador de codificación morfosintáctica para corpus de referencia en lengua española

Este trabajo presenta una propuesta de codificación morfosintáctica para corpus de referencia en lengua española basada en los estándares de la Text Encoding Initiative (TEI), The Network of European Reference Corpora (NERC) y The Expert Advisory Group on Language Engineering Standards (EAGLES) tal y como se presenta en (Martín de Santa Olalla, 1994). Presentamos también el trabajo de creación de etiquetador morfosintáctico que utiliza el conjunto de etiquetas que ésta contiene.

The secret to legal foretelling: generic and inter-generic aspects of vagueness in contracts, patents and regulations

In this genre analysis research paper, we compare U.S. patents, contracts, and regulations on technical matters with a focus upon the relation between vagueness and communicative purposes and subpurposes of these three genres. Our main interest is the investigation of intergeneric conventions across the three genres, based on the software analysis of three corpora (one for each genre, 1 million words per corpus). The result of the investigation is that intergeneric conventions are found at the level of types of expressed linguistic vagueness, but that intergeneric conventions at the level of actual formulations are rare. The conclusion is that at this latter level the influence from the situation type underlying the individual genre is more important than the overarching legal character of the genres, when we talk about introducing explicit vagueness in the text.

UPM system for WMT 2012

This paper describes the UPM system for the Spanish-English translation task at the NAACL 2012 workshop on statistical machine translation. This system is based on Moses. We have used all available free corpora, cleaning and deleting some repetitions. In this paper, we also propose a technique for selecting the sentences for tuning the system. This technique is based on the similarity with the sentences to translate. With our approach, we improve the BLEU score from 28.37% to 28.57%. And as a result of the WMT12 challenge we have obtained a 31.80% BLEU with the 2012 test set. Finally, we explain different experiments that we have carried out after the competition.

Corpus de tesis de ingeniería : paralelismo de las conclusiones entre inglés y español

El objetivo de este Proyecto Fin de Carrera es abordar el análisis del capítulo de conclusiones de tesis de ingeniería de telecomunicación a partir de un corpus comparable en inglés y español. A través del léxico podrán conocerse las expresiones típicas y la estructura de capítulo de conclusiones, tanto en inglés como en español. Para empezar este Proyecto, se ha compilado los corpus que se quieren comparar, en total se ha digitalizado tres corpus, uno con 24 conclusiones de tesis doctorales en español, otro con el mismo número de capítulos de conclusiones de tesis doctorales en inglés (PhD) y por último un corpus de conclusiones de tesis de fin de máster y de grado. El primer análisis que se ha realizado es el de la estructura de las conclusiones a partir de los títulos y subtítulos del capítulo. Se han comparado los títulos más utilizados y se han comentado las coincidencias y diferencias entre los corpus. La estructura vista a través de los subtítulos, se ha comparado con la propuesta por la autora Glasman-Deal (2011) en trabajos académicos de investigación, principalmente en artículos de investigación. La siguiente parte del Proyecto se ha centrado en el estudio del léxico, para ello nos hemos ayudado de la herramienta informática Wordsmith tools de la que se han explicado sus herramientas y funciones más útiles para este trabajo entre ellas el plot, que informa número de archivos en la que aparece cada palabra en el corpus. Las palabras con mayor plot son las más usadas por todos los doctorandos cuando escriben el capítulo de conclusiones .Se han elaborado unas pirámides donde se han colocado las palabras propias del género académico de las tesis por orden de uso. Las más usadas, con mayor plot, en la base y según se asciende aparecen las que tienen menor plot, con el fin de ver de una forma gráfica el peso que tiene cada palabra en el corpus. El siguiente paso del análisis del léxico ha tenido el objetivo de diferenciar los contextos de uso de las palabras incluidas en las pirámides. Se ha diferenciado entre los usos de las palabras dependiendo de su denotación académica o técnica. Esta comparación ha permitido comprobar que dentro del mismo corpus un sustantivo como contribuciones tiene connotación positiva o negativa dependiendo del contexto. Con los ejemplos aportados por los corpus se proporciona una base para el análisis lingüístico, centrado en los sustantivos, en este trabajo. Para finalizar el Proyecto, se ha implementado una base de datos con los resultados obtenidos del análisis de los sustantivos en la que se pueden ver las palabras que corresponden a cada nivel de la pirámide y ejemplos del uso de estas palabras. The aim of this Project is to analyze the concluding chapter of PhD thesis in the field of telecommunication engineering by means of a comparable corpus in English and Spanish. Through the lexis we will be able to capture useful expressions and the typical structure of the chapter in these specialized thesis, either in English and Spanish. To start with, three corpora have been compiled. The first one consists of 24 concluding chapters of PhD thesis in Spanish; the second, is made up of the same number of chapters of PhD thesis in the English language; and finally, 24 further chapters of Master and Degree thesis in English were digitalized and prepared for lexis analysis. Second, the study of the structure of the chapter of conclusions has been carried out. In this part the most common titles in the chapter of conclusions have been analysed and compared so as to find differences and similarities between the two languages compared. Moreover, the structure found through the subtitles in the conclusions of the thesis has been compared with the structure proposed by Glasman-Deal (2011) in her book Science Research Writing. Third, the study has been focused on the lexis of each corpus. These corpora have been treated with a lexis analyser called Wordsmith tools. The variables of frequency and plot have been applied to withdraw the most widely used nouns from the list of all the words found in any of the corpus. A pyramidal structure has been designed in order to show the academic or gender nouns - the ones usually found in the concluding chapter of thesis – nouns with a higher plot in the corpus. Two different types of context have been found for these nouns: technical and academic denotation. To show the difference in use of these nouns, arranged examples of contexts are given for each of the words studied. Finally, a database has been implemented to arrange the results of the lexis study. In this database the most significant examples of each noun are shown.

The unified sentiment lexicon using GPUs

This approach aims at aligning, unifying and expanding the set of sentiment lexicons which are available on the web in order to increase their robustness of coverage. A sentiment lexicon is a critical and essential resource for tagging subjective corpora on the web or elsewhere. In many situations, the multilingual property of the sentiment lexicon is important because the writer is using two languages alternately in the same text, message or post. Our USL approach computes the unified strength of polarity of each lexical entry based on the Pearson correlation coefficient which measures how correlated lexical entries are with a value between 1 and -1, where 1 indicates that the lexical entries are perfectly correlated, 0 indicates no correlation, and -1 means they are perfectly inversely correlated and the UnifiedMetrics procedure for CPU and GPU, respectively.

FragFlow: automated fragment detection in scientific workflows

Scientific workflows provide the means to define, execute and reproduce computational experiments. However, reusing existing workflows still poses challenges for workflow designers. Workflows are often too large and too specific to reuse in their entirety, so reuse is more likely to happen for fragments of workflows. These fragments may be identified manually by users as sub-workflows, or detected automatically. In this paper we present the FragFlow approach, which detects workflow fragments automatically by analyzing existing workflow corpora with graph mining algorithms. FragFlow detects the most common workflow fragments, links them to the original workflows and visualizes them. We evaluate our approach by comparing FragFlow results against user-defined sub-workflows from three different corpora of the LONI Pipeline system. Based on this evaluation, we discuss how automated workflow fragment detection could facilitate workflow reuse.

Language resources and linked data: a practical perspective

Recently, experts and practitioners in language resources have started recognizing the benefits of the linked data (LD) paradigm for the representation and exploitation of linguistic data on the Web. The adoption of the LD principles is leading to an emerging ecosystem of multilingual open resources that conform to the Linguistic Linked Open Data Cloud, in which datasets of linguistic data are interconnected and represented following common vocabularies, which facilitates linguistic information discovery, integration and access. In order to contribute to this initiative, this paper summarizes several key aspects of the representation of linguistic information as linked data from a practical perspective. The main goal of this document is to provide the basic ideas and tools for migrating language resources (lexicons, corpora, etc.) as LD on the Web and to develop some useful NLP tasks with them (e.g., word sense disambiguation). Such material was the basis of a tutorial imparted at the EKAW’14 conference, which is also reported in the paper.

Extended phone log-likelihood ratio features and acoustic-based I-vectors for language recognition

This paper presents new techniques with relevant improvements added to the primary system presented by our group to the Albayzin 2012 LRE competition, where the use of any additional corpora for training or optimizing the models was forbidden. In this work, we present the incorporation of an additional phonotactic subsystem based on the use of phone log-likelihood ratio features (PLLR) extracted from different phonotactic recognizers that contributes to improve the accuracy of the system in a 21.4% in terms of Cavg (we also present results for the official metric during the evaluation, Fact). We will present how using these features at the phone state level provides significant improvements, when used together with dimensionality reduction techniques, especially PCA. We have also experimented with applying alternative SDC-like configurations on these PLLR features with additional improvements. Also, we will describe some modifications to the MFCC-based acoustic i-vector system which have also contributed to additional improvements. The final fused system outperformed the baseline in 27.4% in Cavg.

Un estudio del adjetivo en un corpus de textos escritos en inglés y español pertenecientes al discurso de especialidad de las Tecnologías de la Información y las Comunicaciones

Esta investigación se enmarca dentro de los denominados lenguajes de especialidad que para esta tesis será el de las Tecnologías de la Información y la Comunicación (TIC). De todos los aspectos relacionados con el estudio de estos lenguajes que pudieran tener interés lingüístico ha primado el análisis del componente terminológico. Tradicionalmente la conceptualización de un campo del saber se representaba mayoritariamente a través del elemento nominal, así lo defiende la Teoría General de la Terminología (Wüster, 1968). Tanto la lexicología como la lexicografía han aportado importantes contribuciones a los estudios terminológicos para la identificación del componente léxico a través del cual se transmite la información especializada. No obstante esos primeros estudios terminológicos que apuntaban al sustantivo como elmentos denominativo-conceptual, otras teorías más recientes, entre las que destacamos la Teoría Comunicativa de la Terminología (Cabré, 1999) identifican otras estructuras morfosintácticas integradas por otros elementos no nominales portadores igualmente de esa carga conceptual. A partir de esta consideración, hemos seleccionado para este estudio el adjetivo relacional en tanto que representa otra categoría gramatical distinta al sustantivo y mantiene un vínculo con éste debido a su procedencia. Todo lo cual puede suscitar cierto interés terminológico. A través de esta investigación, nos hemos propuesto demostrar las siguientes hipótesis: 1. El adjetivo relacional aporta contenido especializado en su asociación con el componente nominal. 2. El adjetivo relacional es portador de un valor semántico que hace posible identificar con más precisión la relación conceptual de los elementos -adjetivo y sustantivo - de la combinación léxica resultante, especialmente en algunas formaciones ambiguas. 3. El adjetivo relacional, como modificador natural del sustantivo al que acompaña, podría imponer cierta restricción en sus combinaciones y, por tanto, hacer una selección discriminada de los integrantes de la combinación léxica especializada. Teniendo en cuenta las anteriores hipótesis, esta investigación ha delimitado y caracterizado el segmento léxico objeto de estudio: la ‘combinación léxica especializada (CLE)’ formalmente representada por la estructura sintáctica [adjR+n], en donde adjR es el adjetivo y n el sustantivo al que acompaña. De igual forma hemos descrito el marco teórico desde el que abordar nuestro análisis. Se trata de la teoría del Lexicón Generatvio (LG) y de la representación semántica (Pustojovsky, 1995) que propone como explicación de la generación de significados. Hemos analizado las distintas estructuras de representación léxica y en especial la estructura qualia a través de la cual hemos identificado la relación semántica que mantienen los dos ítems léxicos [adjR+n] de la estructura sintáctica de nuestro estudio. El estudio semántico de las dos piezas léxicas ha permitido, además, comprobar el valor denominativo del adjetivo en la combinación. Ha sido necesario elaborar un corpus de textos escritos en inglés y español pertenecientes al discurso de especialidad de las TIC. Este material ha sido procesado para nuestros fines utilizando distintas herramientas electrónicas. Se ha hecho uso de lexicones electrónicos, diccionarios online generales y de especialidad y corpus de referencia online, estos últimos para poder eventualmente validad nuetros datos. Asimismo se han utilizado motores de búsqueda, entre ellos WordNet Search 3.1, para obtener la información semántica de nuestros elementos léxicos. Nuestras conclusiones han corroborado las hipótesis que se planteaban en esta tesis, en especial la referente al valor denominativo-conceptual del adjetivo relacional el cual, junto con el sustantivo al que acompaña, forma parte de la representación cognitiva del lenguaje de especialidad de las TIC. Como continuación a este estudio se proponen sugerencias sobre líneas futuras de investigación así como el diseño de herramientas informáticas que pudieran incorporar estos datos semánticos como complemento de los ítems léxicos dotados de valor denominativo-conceptual. ABSTRACT This research falls within the field of the so-called Specialized Languages which for the purpose of this study is the Information and Communication Technology (ICT) discourse. Considering their several distinguishing features terminology concentrates our interest from the point of view of linguistics. It is broadly assumed that terms represent concepts of a subject field. For the classical view of terminology (Wüster, 1968) these terms are formally represented by nouns. Both lexicology and terminology have made significant contributions to the study of terms. Later research as well as other theories on Terminology such as the Communicative Theory of Terminology (Cabré, 1993) have shown that other lexical units can also represent knowledge organization. On these bases, we have focused our research on the relational adjective which represents a functional unit different from a noun while still connected to the noun by means of its nominal root. This may have a potential terminological interest. Therefore the present research is based on the next hypotheses: 1. The relational adjective conveys specialized information when combined with the noun. 2. The relational adjective has a semantic meaning which helps understand the conceptual relationship between the adjective and the noun being modified and disambiguate certain senses of the resulting lexical combination. 3. The relational adjective may impose some restrictions when choosing the nouns it modifies. Considering the above hypotheses, this study has identified and described a multi-word lexical unit pattern [Radj+n] referred to as a Specialized Lexical Combination (SLC) linguistically realized by a relational adjective, Radj, and a noun, n. The analysis of such a syntactic pattern is addressed from the framework of the Generative Lexicon (Pustojovsky, 1995). Such theory provides several levels of semantic description which help lexical decomposition performed generatively. These levels of semantic representation are connected through generative operations or generative devices which account for the compositional interpretation of any linguistic utterance in a given context. This study analyses these different levels and focuses on one of them, i.e. the qualia structure since it may encode the conceptual meaning of the syntactic pattern [Radj+n]. The semantic study of these two lexical items has ultimately confirmed the conceptual meaning of the relational adjective. A corpus made of online ICT articles from magazines written in English and Spanish – some being their translations - has been used for the word extraction. For this purpose some word processing software packages have been employed. Moreover online general language and specialized language dictionaries have been consulted. Search engines, namely WordNet Search 3.1, have been also exploited to find the semantic information of our lexical units. Online reference corpora in English and Spanish have been used for a contrastive analysis of our data. Finally our conclusions have confirmed our initial hypotheses, i.e. relational adjectives are specialized lexical units which together with the nouns are part of the knowledge representation of the ICT subject field. Proposals for new research have been made together with some other suggestions for the design of computer applications to visually show the conceptual meaning of certain lexical units.