Skip to main content

Gerardo Sierra

Universidad Nacional Autónoma de México, Instituto de Ingeniería, Faculty Member

Followers

62

Following

25

Co-authors

20

Public Views

Gerardo Sierra Martínez is a National Reseacher of Mexico. He leads the Language Engineering Group at the Instituto de Ingeniería of the Universidad Nacional Autónoma de México (UNAM). He holds a Ph.D. in Computational Linguistics from UMIST, UK. His research interest is focused on language engineering and includes computational lexicography, concept extraction, corpus linguistics, text mining and forensic linguistics. He is author of the book “Introducción a corpus lingüísticos” and more than a hundred or papers on journals, chapter of books and peer reviewed proceedings.

less

University of Minnesota - Twin Cities

Szent Pál Akadémia

Fernando Rodriguez Mediano

CSIC (Consejo Superior de Investigaciones Científicas-Spanish National Research Council)

Harvey (Chaim) Hames

Ben Gurion University of the Negev

Nicoline van der Sijs

Radboud University Nijmegen

University of Michigan

University of Pretoria

Alberdina Houtman

Protestant Theological University, The Netherlands

Interests

Uploads

Papers by Gerardo Sierra

Proceedings of the 1st Workshop on Definition Extraction

by Gerardo Sierra and maría pozzi

@Book{wDE:2009, editor = {Gerardo Sierra and Mara Pozzi and Juan-Manuel Torres}, title = {Proceed... more

An aligned corpus of Spanish bibles

by Gerardo Sierra and Gemma Bel-Enguix

https://scholar.google.com.mx/citations?view_op=view_citation&hl=es&user=0Z-crG4AAAAJ&sortby=pubdate&citation_for_view=0Z-crG4AAAAJ:UarirCmVI0EC#:~:text=Revista-,Language%20Resources%20and%20Evaluation,-P%C3%A1ginas, 2024

We present a comprehensive and valuable resource in the form of an aligned parallel corpus compri... more We present a comprehensive and valuable resource in the form of an aligned parallel corpus comprising translations of the Bible in Spanish. Our collection encompasses a total of eleven Bibles, originating from diverse centuries (XVI, XIX, XX), various religious denominations (Protestant, Catholic), and geographical regions (Spain, Latin America). The process of aligning the verses across these translations has been meticulously carried out, ensuring that the content is organized in a coherent manner. As a result, this corpus serves as a useful convenient resource for various linguistic analyses, including paraphrase detection, semantic clustering, and the exploration of biases present within the texts. To illustrate the utility of this resource, we provide several examples that demonstrate how it can be effectively employed in these applications.

Juan Carlos Sager: home polifacètic, avançat al seu temps

Terminalia, 2023

J uan Carlos Sager deixa un gran llegat no només de publicacions, sinó també de molts altres elem... more J uan Carlos Sager deixa un gran llegat no només de publicacions, sinó també de molts altres elements, igual d'importants. I els records d'aquests perduren en la memòria de tothom que el va conèixer i admirar. La majoria de persones que han abordat professionalment els llenguatges d'especialitat, la traducció, la lexicografia, la traducció automàtica, l'enginyeria lingüística, la tecnologia de la informació i, sobretot, la terminologia, han conegut en Juan Carlos a través de les seves paraules. Però els que vam tenir la fortuna de conèixer-lo personalment-perquè va fer-nos de professor o n'érem col•legues, companys o amics-el recordem, a més, com algú que, tot i que inicialment imposava i podia ser una mica distant, sempre va compartir desinteressadament els seus coneixements i les seves experiències acadèmiques, i amb qui vam establir una relació de confiança i respecte mutus. El professor Sager va néixer a l'Argentina el 3 d'agost de 1929. Va iniciar els estudis universitaris a Tucumán. Tot i que era molt jove, ja tenia una curiositat i un interès científic molt grans pels temes més capdavanters en lingüística aplicada. Després, va anar als Estats Units, a la Universitat de Notre Dame, i més tard a la Gran Bretanya, a Manchester, on va romandre la resta de la seva vida i on es van gestar les seves aportacions més destacades.

Estructura semántica del léxico en un diccionario onomasiológico práctico

Onomasiological dictionary is a practical tool that can help the user to search for related or si... more Onomasiological dictionary is a practical tool that can help the user to search for related or similar terms of certain concepts he does not know how to denominate in a correct way. Computational Linguistics allows to develop this tool through a software. It eases the search for the desired term from an “idea” using any word in any order. An important aspect is foresee all the possible words to be used by user. The lexical analysis of pertinent words for onomasiological search is presented. Application for one selected terminology is given.

Comparative analysis of the computational characteristics in modern sentiment analysis systems for Spanish

Existen múltiples sistemas para análisis de sentimiento con diseños heterogéneos y niveles variad... more Existen múltiples sistemas para análisis de sentimiento con diseños heterogéneos y niveles variados de desempeño. En este artículo se presenta un modelo de generación de especificaciones computacionales de los sistemas para identificación de polaridad tendientes a facilitar comparaciones más profundas y detalladas de las técnicas que se utilizan. Buscamos crear conciencia entre los investigadores de la información necesaria para la construcción de especificaciones que permitan replicar sistemas. A su vez, discutimos las dificultades que se tiene al evaluar y al hacer comparaciones congruentes entre sistemas e ir más allá del resultado que se puede obtener para una tarea específica sobre un conjunto de datos particular. Estamos convencidos de que una estructuración completa y clara de todos los procesos y de los ajustes a que son sometidos los trabajos presentados en las competencias es crucial para enriquecer el conocimiento del uso de estrategias hacia la mejora general de los sist...

Authorship Verification by combining SVMs with Kernels Optimized for Different Feature Categories

We present our approach to the PAN-2015 authorship verification task. We combine one-class SVM cl... more We present our approach to the PAN-2015 authorship verification task. We combine one-class SVM classifiers under the hypothesis that different categories of features a) are better suited for different authors and b) have different underlying topologies. Thus, we have each classifier operate in a different feature subset with a different kernel function, and the output is used to train a logistic regression model which assigns a different weight to each category of features. Results show that further improvement of the method is needed, and we discuss its shortcomings.

An Improved Automated Definition Extraction Method Based on Lexicographic and Lexico-Semantic Features

2016 Fifteenth Mexican International Conference on Artificial Intelligence (MICAI), 2016

This paper shows that results obtained from extracting definitions in computational lexicography,... more This paper shows that results obtained from extracting definitions in computational lexicography, often have a high recall but a low precision. Herein, we present an improved, automated, rule-based analytical definitions extraction method that uses hypernym identification. This kind of definitions allow us to improve the stateof-the-art precision reported in definitions extracting. Furthermore, this method incorporates a hypernyms extraction module, which has proven to be a necessary first step for generating automated definitions.

Detection of Aggressive Tweets in Mexican Spanish Using Multiple Features with Parameter Optimization

This paper explains our approach to Aggressiveness Identification in the MEX-A3T shared task, who... more This paper explains our approach to Aggressiveness Identification in the MEX-A3T shared task, whose aim is the detection of aggressive tweets. The task proposes a binary classification for every tweet: aggressive and non-aggressive. We approached the problem using linguistically motivated features and several types of n-grams (words, characters, functional words, punctuation symbols, among others). We trained a Support Vector Machine using a combinatorial framework that optimizes the results of the classifier. Our best run achieved a F1-score of 0,4549, which is the 5 best among the twenty-six runs.

Recognition of definitions associated to predicative phrases in definitional contexts

En este trabajo se exponen los resultados de un experimento para el reconocimiento semiautomático... more En este trabajo se exponen los resultados de un experimento para el reconocimiento semiautomático de definiciones, considerando su relación con frases predicativas. Tal experimento considera el uso combinado de la llamada teoría de la predicación para delimitar patrones sintácticos, junto con el cálculo de probabilidades condicionales con miras a determinar la frecuencia de asociación entre definiciones y predicaciones.This paper exposes the results obtained from an experiment to recognise semi-automatically definitions, taking in mind its association with predicative phrases. This experiment considers the combined use of the called theory of predication, with the calculus of conditional probabilities for determining the frequency of these associations between definitions and predicative phrases

Costa Rica Central Valley gastronomic lexicon structuring from an ontological arrangement

by HAZEL BARAHONA GAMBOA and Gerardo Sierra

Fórum Linguístico, 2020

This paper describes how terminology and ontologies interact in a coherent and exhaustive manner,... more This paper describes how terminology and ontologies interact in a coherent and exhaustive manner, and how the neological appearance is obtained from the derivation of the specificity and functionality of a specific corpus. This research shows categories that do not exist either in the Glossary of Costa Rican popular cuisine (SEDÓ MASÍS, 2008), or in the Dictionary of Costarriquenismos (QUESADA PACHECO, 2012). The omission of some of these categories clearly shows that the terms have been historically organized taking into account only the lexical factor, but not the conceptual organization. Therefore, this study intends to play with the "disagreement" between speakers and specialists to show that a reorganization of the terms of an area is possible. For this, a corpus of 596 recipes and 4652 ingredients for ontological representation was compiled. Each ontological class was defined by means of its functionality in the recipes. The ontological architecture is formalized thr...

Joint Sentiment Topic Model for objective text clustering1

Journal of Intelligent & Fuzzy Systems, 2019

This article introduces a different method for text representation in order to perform clustering... more This article introduces a different method for text representation in order to perform clustering over different articles which, arguably, has no subjective information with similar topic-sentiment use of language. Using the joint sentiment/topic model, the text is vectorized in a low dimensional space. These vectors were then used as distance measurement for clustering texts. While comparing this unusual method with a traditional bag of words representation an improvement in the performance of the algorithms was observed. The authors think this method of representation might have implications for future studies of the computational interpretation of texts.

Rhetorical Relations in the Speech of Alzheimer’s Patients and Healthy Elderly Subjects: An Approach from the RST

Computación y Sistemas, 2018

The study is aimed to extract discourse relations patterns in conversational speech of subjects w... more The study is aimed to extract discourse relations patterns in conversational speech of subjects with Alzheimer's Disease (AD) and adults with healthy aging processes using the Rhetorical Structure Theory (RST). By means of the RST, we analyzed semi-structured interviews of native Spanish speakers. Seven subjects were in the mild, moderate or advanced stages of AD, and 6 were cognitively intact individuals. The procedure involved the segmentation of each conversational discourse into Semantic Dialog Units (SDUs), the labeling of their rhetorical relations and the construction of tree diagrams. We perform a correlation analysis to determine the significance of the use of rhetorical relations for each group. We found a significantly (p-value < .05) lower rhetorical relations production density in subjects with AD. We also observed that most rhetorical relations used by healthy older subjects were Elaboration, Concession, Interpretation, Non-Volitional Cause, Solutionhood and Volitional Result.

Terminótica y sexualidad: un proyecto integral

Extraction of Definitional Contexts from Biomedical Corpora

En este proyecto se formula una metodología para extraer contextos definitorios desde corpus de b... more En este proyecto se formula una metodología para extraer contextos definitorios desde corpus de biomedicina en español, con el fin de generar los siguientes productos: (i) un listado de candidatos a términos, (ii) un listado de candidatos a definiciones, y (iii) una taxonomía de términos biomédicos basada en relaciones de hiponimia/hiperonimia. Nuestro método permite crear un sistema capaz de extraer tales contextos, el cual puede verse como un módulo que cubriría las primeras etapas a seguir para construir una ontología basada en información textual.In this project we formulate a methodology for extracting definitional contexts from corpus of biomedicine in Spanish, in order to generate the following products: (i) a list of candidate terms, (ii) a list of candidates for definitions, and (iii) a taxonomy of biomedical terms relationships based on hyponym/hyperonym. Our methodology allows the creation of a system capable of extracting such contexts, which can be seen as a module that...

A Mixtec-Spanish Parallel Corpus

Computational technologies have a key role in Computational Linguistics. Thanks to the capability... more Computational technologies have a key role in Computational Linguistics. Thanks to the capability of compiling and analyzing large collections of texts with computers many resources and applications have been designed that have caused a fast development in Natural Language Processing and Artificial Intelligence. Corpora and parallel corpora are basic instruments for approaching natural language, making it possible the implementation of models for machine translation, automatic summarization, information extraction and other methods for language understanding and analysis. All these advances in language technologies need large amounts of data. The most spread and bestrepresented languages in media and internet generate every day Giga Bytes of information that can easily be processed and studied. However, most of the languages in the world are under-represented in social life, the media and, the internet. These are low-resourced languages. An example of this is indigenous languages in...

Development of the Parallel Corpus of Mexican Languages (CPLM)

Mexico has a great language diversity. In addition to Spanish, there are 68 language groups and 3... more Mexico has a great language diversity. In addition to Spanish, there are 68 language groups and 364 variants (INALI, 2008), divided into 11 families. However, this wealth has been threatened due to discrimination against speakers. Indeed, Spanish has been imposed from the legislative, political and economic point of view, which has interrupted the intergenerational transmission of originary languages and, with it, caused the gradual loss of use spaces and communicative functions. Likewise, few technologies have been developed for these languages, because there are few texts written on the internet. The CPLM is a collaborative parallel corpus that contains texts aligned in Spanish and in six indigenous languages: Mayan, Ch'ol, Mazatec, Mixtec, Otomi and Nahuatl. This article describes the development of the CPLM, as well as the difficulties presented throughout the process.

A Parallel Corpus Mixtec-Spanish

This work is about the compilation process of parallel documents Spanish-Mixtec. There are not ma... more This work is about the compilation process of parallel documents Spanish-Mixtec. There are not many Spanish-Mixec parallel texts and most of the sources are non-digital books. Due to this, we need to face the errors when digitizing the sources and difficulties in sentence alignment, as well as the fact that does not exist a standard orthography. Our parallel corpus consists of sixty texts coming from books and digital repositories. These documents belong to different domains: history, traditional stories, didactic material, recipes, ethnographical descriptions of each town and instruction manuals for disease prevention. We have classified this material in five major categories: didactic (6 texts), educative (6 texts), interpretative (7 texts), narrative (39 texts), and poetic (2 texts). The final total of tokens is 49,814 Spanish words and 47,774 Mixtec words. The texts belong to the states of Oaxaca (48 texts), Guerrero (9 texts) and Puebla (3 texts). According to this data, we see...

A Machine Learning Approach for Detecting Aggressive Tweets in Spanish

This paper presents our approach to the aggressive detection track at MEX-A3T 2018. The track con... more This paper presents our approach to the aggressive detection track at MEX-A3T 2018. The track consists in identifying whether a tweet is aggressive or not. To solve this task we follow a machine learning approach, we trained the logistic regression algorithm on linguistically motivated features, and several types of n-grams. We applied several pre-processing steps for standardizing tweets in order to capture relevant information. Our best run achieved 42.85% of f-score on the aggressiveness class, which is between 30% and 40% less than our best cross-validation result on the training set.

Spanish Word Embeddings Learned on Word Association Norms

Word embeddings are vector representations of words in an n-dimensional space used for many natur... more Word embeddings are vector representations of words in an n-dimensional space used for many natural language processing tasks. A large training corpus is needed for learning good quality word embeddings. In this work, we present a method based on the node2vec algorithm for learning embeddings based on paths in a graph. We used a collection of Word Association Norms in Spanish to build a graph of word connections. The nodes of the network correspond to the words in the corpus, whereas the edges correspond to a pair of words given in a free association test. We evaluated our word vectors in human annotated benchmarks, achieving better results than those trained on a billion-word corpus such as, word2vec, fasttext, and glove.

Automatic Summarization Using Terminological and Semantic Resources

In this paper we present a new algorithm for automatic summarization of specialized texts combini... more In this paper we present a new algorithm for automatic summarization of specialized texts combining terminological and semantic resources: a term extractor and an ontology. The term extractor provides the list of the terms that are present in the text together their corresponding termhood. The ontology is used to calculate the semantic similarity among the terms found in the main body and those present in the document title. The phrases with the highest score are chosen to take part of the final summary. We evaluate the algorithm with ROUGE, comparing the resulting summaries with the summaries of other summarizers. The sentence selection algorithm was also tested as part of a standalone summarizer. It obtains good results, but there is a space for improvement.

Proceedings of the 1st Workshop on Definition Extraction

by Gerardo Sierra and maría pozzi

@Book{wDE:2009, editor = {Gerardo Sierra and Mara Pozzi and Juan-Manuel Torres}, title = {Proceed... more

An aligned corpus of Spanish bibles

by Gerardo Sierra and Gemma Bel-Enguix

https://scholar.google.com.mx/citations?view_op=view_citation&hl=es&user=0Z-crG4AAAAJ&sortby=pubdate&citation_for_view=0Z-crG4AAAAJ:UarirCmVI0EC#:~:text=Revista-,Language%20Resources%20and%20Evaluation,-P%C3%A1ginas, 2024

We present a comprehensive and valuable resource in the form of an aligned parallel corpus compri... more We present a comprehensive and valuable resource in the form of an aligned parallel corpus comprising translations of the Bible in Spanish. Our collection encompasses a total of eleven Bibles, originating from diverse centuries (XVI, XIX, XX), various religious denominations (Protestant, Catholic), and geographical regions (Spain, Latin America). The process of aligning the verses across these translations has been meticulously carried out, ensuring that the content is organized in a coherent manner. As a result, this corpus serves as a useful convenient resource for various linguistic analyses, including paraphrase detection, semantic clustering, and the exploration of biases present within the texts. To illustrate the utility of this resource, we provide several examples that demonstrate how it can be effectively employed in these applications.

Juan Carlos Sager: home polifacètic, avançat al seu temps

Terminalia, 2023

J uan Carlos Sager deixa un gran llegat no només de publicacions, sinó també de molts altres elem... more J uan Carlos Sager deixa un gran llegat no només de publicacions, sinó també de molts altres elements, igual d'importants. I els records d'aquests perduren en la memòria de tothom que el va conèixer i admirar. La majoria de persones que han abordat professionalment els llenguatges d'especialitat, la traducció, la lexicografia, la traducció automàtica, l'enginyeria lingüística, la tecnologia de la informació i, sobretot, la terminologia, han conegut en Juan Carlos a través de les seves paraules. Però els que vam tenir la fortuna de conèixer-lo personalment-perquè va fer-nos de professor o n'érem col•legues, companys o amics-el recordem, a més, com algú que, tot i que inicialment imposava i podia ser una mica distant, sempre va compartir desinteressadament els seus coneixements i les seves experiències acadèmiques, i amb qui vam establir una relació de confiança i respecte mutus. El professor Sager va néixer a l'Argentina el 3 d'agost de 1929. Va iniciar els estudis universitaris a Tucumán. Tot i que era molt jove, ja tenia una curiositat i un interès científic molt grans pels temes més capdavanters en lingüística aplicada. Després, va anar als Estats Units, a la Universitat de Notre Dame, i més tard a la Gran Bretanya, a Manchester, on va romandre la resta de la seva vida i on es van gestar les seves aportacions més destacades.

Estructura semántica del léxico en un diccionario onomasiológico práctico

Onomasiological dictionary is a practical tool that can help the user to search for related or si... more Onomasiological dictionary is a practical tool that can help the user to search for related or similar terms of certain concepts he does not know how to denominate in a correct way. Computational Linguistics allows to develop this tool through a software. It eases the search for the desired term from an “idea” using any word in any order. An important aspect is foresee all the possible words to be used by user. The lexical analysis of pertinent words for onomasiological search is presented. Application for one selected terminology is given.

Comparative analysis of the computational characteristics in modern sentiment analysis systems for Spanish

Existen múltiples sistemas para análisis de sentimiento con diseños heterogéneos y niveles variad... more Existen múltiples sistemas para análisis de sentimiento con diseños heterogéneos y niveles variados de desempeño. En este artículo se presenta un modelo de generación de especificaciones computacionales de los sistemas para identificación de polaridad tendientes a facilitar comparaciones más profundas y detalladas de las técnicas que se utilizan. Buscamos crear conciencia entre los investigadores de la información necesaria para la construcción de especificaciones que permitan replicar sistemas. A su vez, discutimos las dificultades que se tiene al evaluar y al hacer comparaciones congruentes entre sistemas e ir más allá del resultado que se puede obtener para una tarea específica sobre un conjunto de datos particular. Estamos convencidos de que una estructuración completa y clara de todos los procesos y de los ajustes a que son sometidos los trabajos presentados en las competencias es crucial para enriquecer el conocimiento del uso de estrategias hacia la mejora general de los sist...

Authorship Verification by combining SVMs with Kernels Optimized for Different Feature Categories

We present our approach to the PAN-2015 authorship verification task. We combine one-class SVM cl... more We present our approach to the PAN-2015 authorship verification task. We combine one-class SVM classifiers under the hypothesis that different categories of features a) are better suited for different authors and b) have different underlying topologies. Thus, we have each classifier operate in a different feature subset with a different kernel function, and the output is used to train a logistic regression model which assigns a different weight to each category of features. Results show that further improvement of the method is needed, and we discuss its shortcomings.

An Improved Automated Definition Extraction Method Based on Lexicographic and Lexico-Semantic Features

2016 Fifteenth Mexican International Conference on Artificial Intelligence (MICAI), 2016

This paper shows that results obtained from extracting definitions in computational lexicography,... more This paper shows that results obtained from extracting definitions in computational lexicography, often have a high recall but a low precision. Herein, we present an improved, automated, rule-based analytical definitions extraction method that uses hypernym identification. This kind of definitions allow us to improve the stateof-the-art precision reported in definitions extracting. Furthermore, this method incorporates a hypernyms extraction module, which has proven to be a necessary first step for generating automated definitions.

Detection of Aggressive Tweets in Mexican Spanish Using Multiple Features with Parameter Optimization

This paper explains our approach to Aggressiveness Identification in the MEX-A3T shared task, who... more This paper explains our approach to Aggressiveness Identification in the MEX-A3T shared task, whose aim is the detection of aggressive tweets. The task proposes a binary classification for every tweet: aggressive and non-aggressive. We approached the problem using linguistically motivated features and several types of n-grams (words, characters, functional words, punctuation symbols, among others). We trained a Support Vector Machine using a combinatorial framework that optimizes the results of the classifier. Our best run achieved a F1-score of 0,4549, which is the 5 best among the twenty-six runs.

Recognition of definitions associated to predicative phrases in definitional contexts

En este trabajo se exponen los resultados de un experimento para el reconocimiento semiautomático... more En este trabajo se exponen los resultados de un experimento para el reconocimiento semiautomático de definiciones, considerando su relación con frases predicativas. Tal experimento considera el uso combinado de la llamada teoría de la predicación para delimitar patrones sintácticos, junto con el cálculo de probabilidades condicionales con miras a determinar la frecuencia de asociación entre definiciones y predicaciones.This paper exposes the results obtained from an experiment to recognise semi-automatically definitions, taking in mind its association with predicative phrases. This experiment considers the combined use of the called theory of predication, with the calculus of conditional probabilities for determining the frequency of these associations between definitions and predicative phrases

Costa Rica Central Valley gastronomic lexicon structuring from an ontological arrangement

by HAZEL BARAHONA GAMBOA and Gerardo Sierra

Fórum Linguístico, 2020

This paper describes how terminology and ontologies interact in a coherent and exhaustive manner,... more This paper describes how terminology and ontologies interact in a coherent and exhaustive manner, and how the neological appearance is obtained from the derivation of the specificity and functionality of a specific corpus. This research shows categories that do not exist either in the Glossary of Costa Rican popular cuisine (SEDÓ MASÍS, 2008), or in the Dictionary of Costarriquenismos (QUESADA PACHECO, 2012). The omission of some of these categories clearly shows that the terms have been historically organized taking into account only the lexical factor, but not the conceptual organization. Therefore, this study intends to play with the "disagreement" between speakers and specialists to show that a reorganization of the terms of an area is possible. For this, a corpus of 596 recipes and 4652 ingredients for ontological representation was compiled. Each ontological class was defined by means of its functionality in the recipes. The ontological architecture is formalized thr...

Joint Sentiment Topic Model for objective text clustering1

Journal of Intelligent & Fuzzy Systems, 2019

This article introduces a different method for text representation in order to perform clustering... more This article introduces a different method for text representation in order to perform clustering over different articles which, arguably, has no subjective information with similar topic-sentiment use of language. Using the joint sentiment/topic model, the text is vectorized in a low dimensional space. These vectors were then used as distance measurement for clustering texts. While comparing this unusual method with a traditional bag of words representation an improvement in the performance of the algorithms was observed. The authors think this method of representation might have implications for future studies of the computational interpretation of texts.

Rhetorical Relations in the Speech of Alzheimer’s Patients and Healthy Elderly Subjects: An Approach from the RST

Computación y Sistemas, 2018

The study is aimed to extract discourse relations patterns in conversational speech of subjects w... more The study is aimed to extract discourse relations patterns in conversational speech of subjects with Alzheimer's Disease (AD) and adults with healthy aging processes using the Rhetorical Structure Theory (RST). By means of the RST, we analyzed semi-structured interviews of native Spanish speakers. Seven subjects were in the mild, moderate or advanced stages of AD, and 6 were cognitively intact individuals. The procedure involved the segmentation of each conversational discourse into Semantic Dialog Units (SDUs), the labeling of their rhetorical relations and the construction of tree diagrams. We perform a correlation analysis to determine the significance of the use of rhetorical relations for each group. We found a significantly (p-value < .05) lower rhetorical relations production density in subjects with AD. We also observed that most rhetorical relations used by healthy older subjects were Elaboration, Concession, Interpretation, Non-Volitional Cause, Solutionhood and Volitional Result.

Terminótica y sexualidad: un proyecto integral

Extraction of Definitional Contexts from Biomedical Corpora

En este proyecto se formula una metodología para extraer contextos definitorios desde corpus de b... more En este proyecto se formula una metodología para extraer contextos definitorios desde corpus de biomedicina en español, con el fin de generar los siguientes productos: (i) un listado de candidatos a términos, (ii) un listado de candidatos a definiciones, y (iii) una taxonomía de términos biomédicos basada en relaciones de hiponimia/hiperonimia. Nuestro método permite crear un sistema capaz de extraer tales contextos, el cual puede verse como un módulo que cubriría las primeras etapas a seguir para construir una ontología basada en información textual.In this project we formulate a methodology for extracting definitional contexts from corpus of biomedicine in Spanish, in order to generate the following products: (i) a list of candidate terms, (ii) a list of candidates for definitions, and (iii) a taxonomy of biomedical terms relationships based on hyponym/hyperonym. Our methodology allows the creation of a system capable of extracting such contexts, which can be seen as a module that...

A Mixtec-Spanish Parallel Corpus

Computational technologies have a key role in Computational Linguistics. Thanks to the capability... more Computational technologies have a key role in Computational Linguistics. Thanks to the capability of compiling and analyzing large collections of texts with computers many resources and applications have been designed that have caused a fast development in Natural Language Processing and Artificial Intelligence. Corpora and parallel corpora are basic instruments for approaching natural language, making it possible the implementation of models for machine translation, automatic summarization, information extraction and other methods for language understanding and analysis. All these advances in language technologies need large amounts of data. The most spread and bestrepresented languages in media and internet generate every day Giga Bytes of information that can easily be processed and studied. However, most of the languages in the world are under-represented in social life, the media and, the internet. These are low-resourced languages. An example of this is indigenous languages in...

Development of the Parallel Corpus of Mexican Languages (CPLM)

Mexico has a great language diversity. In addition to Spanish, there are 68 language groups and 3... more Mexico has a great language diversity. In addition to Spanish, there are 68 language groups and 364 variants (INALI, 2008), divided into 11 families. However, this wealth has been threatened due to discrimination against speakers. Indeed, Spanish has been imposed from the legislative, political and economic point of view, which has interrupted the intergenerational transmission of originary languages and, with it, caused the gradual loss of use spaces and communicative functions. Likewise, few technologies have been developed for these languages, because there are few texts written on the internet. The CPLM is a collaborative parallel corpus that contains texts aligned in Spanish and in six indigenous languages: Mayan, Ch'ol, Mazatec, Mixtec, Otomi and Nahuatl. This article describes the development of the CPLM, as well as the difficulties presented throughout the process.

A Parallel Corpus Mixtec-Spanish

This work is about the compilation process of parallel documents Spanish-Mixtec. There are not ma... more This work is about the compilation process of parallel documents Spanish-Mixtec. There are not many Spanish-Mixec parallel texts and most of the sources are non-digital books. Due to this, we need to face the errors when digitizing the sources and difficulties in sentence alignment, as well as the fact that does not exist a standard orthography. Our parallel corpus consists of sixty texts coming from books and digital repositories. These documents belong to different domains: history, traditional stories, didactic material, recipes, ethnographical descriptions of each town and instruction manuals for disease prevention. We have classified this material in five major categories: didactic (6 texts), educative (6 texts), interpretative (7 texts), narrative (39 texts), and poetic (2 texts). The final total of tokens is 49,814 Spanish words and 47,774 Mixtec words. The texts belong to the states of Oaxaca (48 texts), Guerrero (9 texts) and Puebla (3 texts). According to this data, we see...

A Machine Learning Approach for Detecting Aggressive Tweets in Spanish

This paper presents our approach to the aggressive detection track at MEX-A3T 2018. The track con... more This paper presents our approach to the aggressive detection track at MEX-A3T 2018. The track consists in identifying whether a tweet is aggressive or not. To solve this task we follow a machine learning approach, we trained the logistic regression algorithm on linguistically motivated features, and several types of n-grams. We applied several pre-processing steps for standardizing tweets in order to capture relevant information. Our best run achieved 42.85% of f-score on the aggressiveness class, which is between 30% and 40% less than our best cross-validation result on the training set.

Spanish Word Embeddings Learned on Word Association Norms

Word embeddings are vector representations of words in an n-dimensional space used for many natur... more Word embeddings are vector representations of words in an n-dimensional space used for many natural language processing tasks. A large training corpus is needed for learning good quality word embeddings. In this work, we present a method based on the node2vec algorithm for learning embeddings based on paths in a graph. We used a collection of Word Association Norms in Spanish to build a graph of word connections. The nodes of the network correspond to the words in the corpus, whereas the edges correspond to a pair of words given in a free association test. We evaluated our word vectors in human annotated benchmarks, achieving better results than those trained on a billion-word corpus such as, word2vec, fasttext, and glove.

Automatic Summarization Using Terminological and Semantic Resources

In this paper we present a new algorithm for automatic summarization of specialized texts combini... more In this paper we present a new algorithm for automatic summarization of specialized texts combining terminological and semantic resources: a term extractor and an ontology. The term extractor provides the list of the terms that are present in the text together their corresponding termhood. The ontology is used to calculate the semantic similarity among the terms found in the main body and those present in the document title. The phrases with the highest score are chosen to take part of the final summary. We evaluate the algorithm with ROUGE, comparing the resulting summaries with the summaries of other summarizers. The sentence selection algorithm was also tested as part of a standalone summarizer. It obtains good results, but there is a space for improvement.

Tecnología informática para la confección de diccionarios

Lexicografía hispánica / The Routledge Handbook of Spanish Lexicography, 2023

En el capítulo se presentan las etapas del trabajo lexicográfico en general y del tipo de informa... more En el capítulo se presentan las etapas del trabajo lexicográfico en general y del tipo de información que debe considerarse según el tipo de diccionario que se pretenda realizar. En función del tipo de diccionario (sea monolingüe o bilingüe, especializado, ideológico, fraseológico, dialectal, etcétera) y de la información que vaya a contener, será el diseño del corpus y la selección de las técnicas y herramientas necesarias informáticas.

Description and Evaluation of a Definition Extraction System for Spanish language

by Carme Bach and Gerardo Sierra

dins SIERRA, G.; POZZIi, M.; TORRES-MORENO, J. M. [ed.] Proceedings. 1st International Workshop on Definition Extraction . Shoumen: Incoma Ltd. Pàg. 7-13. ISBN 978-954-452-012-0, 2009

In this paper we present a description of the role of definitional verbal patterns for the extrac... more In this paper we present a description of the role of definitional verbal patterns for the extraction of semantic relations. Several studies show that semantic relations can be extracted from analytic definitions contained in machine-readable dictionaries (MRDs). In addition, definitions found in specialised texts are a good starting point to search for different types of definitions where other semantic relations occur. The extraction of definitional knowledge from specialised corpora represents another interesting approach for the extraction of semantic relations. Here, we present a descriptive analysis of definitional verbal patterns in Spanish and the first steps towards the development of a system for the automatic extraction of definitional knowledge.

Development of the Parallel Corpus of Mexican Languages (CPLM

by Cynthia A Montaño Ramírez and Gerardo Sierra

Mexico has a great language diversity. In addition to Spanish, there are 68 language groups and 3... more Mexico has a great language diversity. In addition to Spanish, there are 68 language groups and 364 variants (INALI, 2008), divided into 11 families. However, this wealth has been threatened due to discrimination against speakers. Indeed, Spanish has been imposed from the legislative, political and economic point of view, which has interrupted the intergenerational transmission of originary languages and, with it, caused the gradual loss of use spaces and communicative functions. Likewise, few technologies have been developed for these languages, because there are few texts written on the internet. The CPLM is a collaborative parallel corpus that contains texts aligned in Spanish and in six indigenous languages: Mayan, Ch'ol, Mazatec, Mixtec, Otomi and Nahuatl. This article describes the development of the CPLM, as well as the difficulties presented throughout the process. Resumen México cuenta con una gran diversidad de lenguas, ya que, aparte del español, existen 68 agrupaciones lingüísticas y 364 variantes (INALI, 2008), repartidas en 11 familias. Sin embargo, esta riqueza se ha visto amenazada debido a la discriminación hacia los hablantes. En efecto español se ha impuesto desde el punto de vista legislativo, político y económico, lo que ha interrumpido la transmisión intergeneracional de las lenguas originarias y, con ello, originado la pérdida paulatina de espacios de uso y funciones comunicativas. Así mismo, pocas tecnologías se han desarrollado para estas lenguas, debido a que existen pocos textos escritos en internet. El CPLM es un corpus paralelo colaborativo que presenta textos alineados en español y en seis lenguas indígenas: maya, ch'ol, mazateco, mixteco, otomí y náhuatl. Este artículo describe el desarrollo del CPLM, así como las dificultades presentadas a lo largo del proceso.