After trying Festival Speech Synthesis System, we decided to develop our own TTS framework, conducive to perform the necessary research experiments for developing good quality TTS for Indian languages. In most of the attempts on Indian language TTS, there is no prosody model, provision for handling foreign language words and no phrase break prediction leading to the possibility of introducing appropriate pauses in the synthesized speech. Further, in the Indian context, there is a real felt need for a bilingual TTS, involving English, along with the Indian language. In fact, it may be desirable to also have a trilingual TTS, which can also take care of the language of the neighboring state or Hindi, in addition. Thus, there is a felt need for a full-fledged TTS development framework, which lends itself for experimentation involving all the above issues and more. This thesis work is therefore such a serious attempt to develop a modular, unit selection based TTS framework. The developed system has been tested for its effectiveness to create intelligible speech in Tamil and Kannada. The created system has also been used to carry out two research experiments on TTS. The first part of the work is the design and development of corpus-based concatenative Tamil speech synthesizer in Matlab and C. A synthesis database has been created with 1027 phonetically rich, pre-recorded sentences, segmented at the phone level. From the sentence to be synthesized, specifications of the required target units are predicted. During synthesis, database units are selected that best match the target specification according to a distance metric and a concatenation quality metric. To accelerate matching, the features of the end frames of the database units have been precomputed and stored. The selected units are concatenated to produce synthetic speech. The high values of the obtained mean opinion scores for the TTS output reveal that speech synthesized using our TTS is intelligible and acceptably natural and can possibly be put to commercial use with some additional features. Experiments carried out by others using my TTS framework have shown that, whenever the required phonetic context is not available in the synthesis database., similar phones that are perceptually indistinguishable may be substituted. The second part of the work deals with the design and modification of the developed TTS framework to be embedded in mobile phones. Commercial GSM FR, EFR and AMR speech codecs are used for compressing our synthesis database. Perception experiments reveal that speech synthesized using a highly compressed database is reasonably natural. This holds promise in the future to read SMSs and emails on mobile phones in Indian languages. Finally, we observe that incorporating prosody and pause models for Indian language TTS would further enhance the quality of the synthetic speech. These are some of the potential, unexplored areas ahead, for research in speech synthesis in Indian languages.

In this dissertation, we introduce a novel text representation method mainly used for text classification purpose. The presented representation method is initially based on a variety of closeness relationships between pairs of words in text passages within the entire corpus. This representation is then used as the basis for our multi-level lightweight ontological representation method (TOR-FUSE), in which documents are represented based on their contexts and the goal of the learning task. The method is unlike the traditional representation methods, in which all the documents are represented solely based on the constituent words of the documents, and are totally isolated from the goal that they are represented for. We believe choosing the correct granularity of representation features is an important aspect of text classification. Interpreting data in a more general dimensional space, with fewer dimensions, can convey more discriminative knowledge and decrease the level of learning perplexity. The multi-level model allows data interpretation in a more conceptual space, rather than only containing scattered words occurring in texts. It aims to perform the extraction of the knowledge tailored for the classification task by automatic creation of a lightweight ontological hierarchy of representations. In the last step, we will train a tailored ensemble learner over a stack of representations at different conceptual granularities. The final result is a mapping and a weighting of the targeted concept of the original learning task, over a stack of representations and granular conceptual elements of its different levels (hierarchical mapping instead of linear mapping over a vector). Finally the entire algorithm is applied to a variety of general text classification tasks, and the performance is evaluated in comparison with well-known algorithms.

News companies have a need to automate and make the editors process of writing about hot and new events more effective. Current technologies involve robotic programs that fills in values in templates and website listeners that notifies the editors when changes are made so that the editor can read up on the source change at the actual website. Editors can provide news faster and better if directly provided with abstracts of the external sources. This study applies deep learning algorithms to automatically formulate abstracts and tag sources with appropriate tags based on the context. The study is a full stack solution, which manages both the editors need for speed and the training, testing and validation of the algorithms. Decision Tree, Random Forest, Multi Layer Perceptron and phrase document vectors are used to evaluate the categorisation and Recurrent Neural Networks is used to paraphrase unformatted texts. In the evaluation a comparison between different models trained by the algorithms with a variation of parameters are done based on the F-score. The results shows that the F-scores are increasing the more document the training has and decreasing the more categories the algorithm needs to consider. The Multi-Layer Perceptron perform best followed by Random Forest and finally Decision Tree. The document length matters, when larger documents are considered during training the score is increasing considerably. A user survey about the paraphrase algorithms shows the paraphrase result is insufficient to satisfy editors need. It confirms a need for more memory to conduct longer experiments.

Cette thèse s'inscrit dans les domaines des systèmes Question Réponse en domaine restreint, la recherche d'information ainsi que TALN. Les systèmes de Question Réponse (QR) ont pour objectif de retrouver un fragment pertinent d'un document qui pourrait être considéré comme la meilleure réponse concise possible à une question de l'utilisateur. Le but de cette thèse est de proposer une approche de localisation de réponses dans des masses de données complexes et évolutives décrites ci-dessous.. De nos jours, dans de nombreux domaines d'application, les systèmes informatiques sont instrumentés pour produire des rapports d'événements survenant, dans un format de données textuelles généralement appelé fichiers log. Les fichiers logs représentent la source principale d'informations sur l'état des systèmes, des produits, ou encore les causes de problèmes qui peuvent survenir. Les fichiers logs peuvent également inclure des données sur les paramètres critiques, les sorties de capteurs, ou une combinaison de ceux-ci. Ces fichiers sont également utilisés lors des différentes étapes du développement de logiciels, principalement dans l'objectif de débogage et le profilage. Les fichiers logs sont devenus un élément standard et essentiel de toutes les grandes applications. Bien que le processus de génération de fichiers logs est assez simple et direct, l'analyse de fichiers logs pourrait être une tâche difficile qui exige d'énormes ressources de calcul, de temps et de procédures sophistiquées. En effet, il existe de nombreux types de fichiers logs générés dans certains domaines d'application qui ne sont pas systématiquement exploités d'une manière efficace en raison de leurs caractéristiques particulières. Dans cette thèse, nous nous concentrerons sur un type des fichiers logs générés par des systèmes EDA (Electronic Design Automation). Ces fichiers logs contiennent des informations sur la configuration et la conception des Circuits Intégrés (CI) ainsi que les tests de vérification effectués sur eux. Ces informations, très peu exploitées actuellement, sont particulièrement attractives et intéressantes pour la gestion de conception, la surveillance et surtout la vérification de la qualité de conception. Cependant, la complexité de ces données textuelles complexes, c.-à-d. des fichiers logs générés par des outils de conception de CI, rend difficile l'exploitation de ces connaissances. Plusieurs aspects de ces fichiers logs ont été moins soulignés dans les méthodes de TALN et Extraction d'Information (EI). Le grand volume de données et leurs caractéristiques particulières limitent la pertinence des méthodes classiques de TALN et EI. Dans ce projet de recherche nous cherchons à proposer une approche qui permet de répondre à répondre automatiquement aux questionnaires de vérification de qualité des CI selon les informations se trouvant dans les fichiers logs générés par les outils de conception. Au sein de cette thèse, nous étudions principalement "comment les spécificités de fichiers logs peuvent influencer l'extraction de l'information et les méthodes de TALN?". Le problème est accentué lorsque nous devons également prendre leurs structures évolutives et leur vocabulaire spécifique en compte. Dans ce contexte, un défi clé est de fournir des approches qui prennent les spécificités des fichiers logs en compte tout en considérant les enjeux qui sont spécifiques aux systèmes QR dans des domaines restreints. Ainsi, les contributions de cette thèse consistent brièvement en :〉Proposer une méthode d'identification et de reconnaissance automatique des unités logiques dans les fichiers logs afin d'effectuer une segmentation textuelle selon la structure des fichiers. Au sein de cette approche, nous proposons un type original de descripteur qui permet de modéliser la structure textuelle et le layout des documents textuels.〉Proposer une approche de la localisation de réponse (recherche de passages) dans les fichiers logs. Afin d'améliorer la performance de recherche de passage ainsi que surmonter certains problématiques dûs aux caractéristiques des fichiers logs, nous proposons une approches d'enrichissement de requêtes. Cette approches, fondée sur la notion de relevance feedback, consiste en un processus d'apprentissage et une méthode de pondération des mots pertinents du contexte qui sont susceptibles d'exister dans les passage adaptés. Cela dit, nous proposons également une nouvelle fonction originale de pondération (scoring), appelée TRQ (Term Relatedness to Query) qui a pour objectif de donner un poids élevé aux termes qui ont une probabilité importante de faire partie des passages pertinents. Cette approche est également adaptée et évaluée dans les domaines généraux.〉Etudier l'utilisation des connaissances morpho-syntaxiques au sein de nos approches. A cette fin, nous nous sommes intéressés à l'extraction de la terminologie dans les fichiers logs. Ainsi, nous proposons la méthode Exterlog, adaptée aux spécificités des logs, qui permet d'extraire des termes selon des patrons syntaxiques. Afin d'évaluer les termes extraits et en choisir les plus pertinents, nous proposons un protocole de validation automatique des termes qui utilise une mesure fondée sur le Web associée à des mesures statistiques, tout en prenant en compte le contexte spécialisé des logs. / In this thesis, we present contributions to the challenging issues which are encounteredin question answering and locating information in complex textual data, like log files. Question answering systems (QAS) aim to find a relevant fragment of a document which could be regarded as the best possible concise answer for a question given by a user. In this work, we are looking to propose a complete solution to locate information in a special kind of textual data, i.e., log files generated by EDA design tools.Nowadays, in many application areas, modern computing systems are instrumented to generate huge reports about occurring events in the format of log files. Log files are generated in every computing field to report the status of systems, products, or even causes of problems that can occur. Log files may also include data about critical parameters, sensor outputs, or a combination of those. Analyzing log files, as an attractive approach for automatic system management and monitoring, has been enjoying a growing amount of attention [Li et al., 2005]. Although the process of generating log files is quite simple and straightforward, log file analysis could be a tremendous task that requires enormous computational resources, long time and sophisticated procedures [Valdman, 2004]. Indeed, there are many kinds of log files generated in some application domains which are not systematically exploited in an efficient way because of their special characteristics. In this thesis, we are mainly interested in log files generated by Electronic Design Automation (EDA) systems. Electronic design automation is a category of software tools for designing electronic systems such as printed circuit boards and Integrated Circuits (IC). In this domain, to ensure the design quality, there are some quality check rules which should be verified. Verification of these rules is principally performed by analyzing the generated log files. In the case of large designs that the design tools may generate megabytes or gigabytes of log files each day, the problem is to wade through all of this data to locate the critical information we need to verify the quality check rules. These log files typically include a substantial amount of data. Accordingly, manually locating information is a tedious and cumbersome process. Furthermore, the particular characteristics of log files, specially those generated by EDA design tools, rise significant challenges in retrieval of information from the log files. The specific features of log files limit the usefulness of manual analysis techniques and static methods. Automated analysis of such logs is complex due to their heterogeneous and evolving structures and the large non-fixed vocabulary.In this thesis, by each contribution, we answer to questions raised in this work due to the data specificities or domain requirements. We investigate throughout this work the main concern "how the specificities of log files can influence the information extraction and natural language processing methods?". In this context, a key challenge is to provide approaches that take the log file specificities into account while considering the issues which are specific to QA in restricted domains. We present different contributions as below:> Proposing a novel method to recognize and identify the logical units in the log files to perform a segmentation according to their structure. We thus propose a method to characterize complex logicalunits found in log files according to their syntactic characteristics. Within this approach, we propose an original type of descriptor to model the textual structure and layout of text documents.> Proposing an approach to locate the requested information in the log files based on passage retrieval. To improve the performance of passage retrieval, we propose a novel query expansion approach to adapt an initial query to all types of corresponding log files and overcome the difficulties like mismatch vocabularies. Our query expansion approach relies on two relevance feedback steps. In the first one, we determine the explicit relevance feedback by identifying the context of questions. The second phase consists of a novel type of pseudo relevance feedback. Our method is based on a new term weighting function, called TRQ (Term Relatedness to Query), introduced in this work, which gives a score to terms of corpus according to their relatedness to the query. We also investigate how to apply our query expansion approach to documents from general domains.> Studying the use of morpho-syntactic knowledge in our approaches. For this purpose, we are interested in the extraction of terminology in the log files. Thus, we here introduce our approach, named Exterlog (EXtraction of TERminology from LOGs), to extract the terminology of log files. To evaluate the extracted terms and choose the most relevant ones, we propose a candidate term evaluation method using a measure, based on the Web and combined with statistical measures, taking into account the context of log files.

The main archive of life sciences literature currently contains more than 18,000,000 references, and it is virtually impossible for any human to stay up-to-date with this large number of papers, even in a specific sub-domain. Not every fact that is reported in the literature is novel and distinct. Scientists report repeat experiments, or refer to previous findings. Given the large number of publications, it is not surprising that information on certain topics is repeated over a number of publications. From consensus to contradiction, there are all shades of agreement between the claimed facts in the literature, and considering the volume of the corpus, conflicting findings are not unlikely. Finding such claims is particularly interesting for scientists, as they can present opportunities for knowledge consolidation and future investigations. In this thesis we present a method to extract and contextualise statements about molecular events as expressed in the biomedical literature, and to find those that potentially conflict each other. The approach uses a system that detects event negations and speculation, and combines those with contextual features (e.g. type of event, species, and anatomical location) to build a representational model for establishing relations between different biological events, including relations concerning conflicts. In the detection of negations and speculations, rich lexical, syntactic, and semantic features have been exploited, including the syntactic command relation. Different parts of the proposed method have been evaluated in a context of the BioNLP 09 challenge. The average F-measures for event negation and speculation detection were 63% (with precision of 88%) and 48% (with precision of 64%) respectively. An analysis of a set of 50 extracted event pairs identified as potentially conflicting revealed that 32 of them showed some degree of conflict (64%); 10 event pairs (20%) needed a more complex biological interpretation to decide whether there was a conflict. We also provide an open source integrated text mining framework for extracting events and their context on a large-scale basis using a pipeline of tools that are available or have been developed as part of this research, along with 72,314 potentially conflicting molecular event pairs that have been generated by mining the entire body of accessible biomedical literature. We conclude that, whilst automated conflict mining would need more comprehensive context extraction, it is feasible to provide a support environment for biologists to browse potential conflicting statements and facilitate data and knowledge consolidation.

Les tables du Lexique-Grammaire, dont le développement a été initié par Gross (1975), constituent un lexique syntaxique très riche pour le français. Elles couvrent diverses catégories lexicales telles que les verbes, les noms, les adjectifs et les adverbes. Cette base de données linguistiques n'est cependant pas directement exploitable informatiquement car elle est incomplète et manque de cohérence. Chaque table regroupe un certain nombre d'entrées jugées similaires car elles acceptent des propriétés communes. Ces propriétés ont pour particularité de ne pas être codées dans les tables même mais uniquement décrites dans la littérature. Pour rendre ces tables exploitables, il faut expliciter les propriétés intervenant dans chacune d'entre elles. De plus, un grand nombre de ces propriétés doivent être renommées dans un souci de cohérence. Notre objectif est d'adapter les tables pour les rendre utilisables dans diverses applications de Traitement Automatique des Langues (TAL), notamment l'analyse syntaxique. Nous expliquons les problèmes rencontrés et les méthodes adoptées pour permettre leur intégration dans un analyseur syntaxique. Nous proposons LGExtract, un outil générique pour générer un lexique syntaxique pour le TAL à partir des tables du Lexique-Grammaire. Il est relié à une table globale dans laquelle nous avons ajouté les propriétés manquantes et un unique script d'extraction incluant toutes les opérations liées à chaque propriété devant être effectuées pour toutes les tables. Nous présentons également LGLex, le nouveau lexique syntaxique généré des verbes, des noms prédicatifs, des expressions figées et des adverbes. Ensuite, nous montrons comment nous avons converti les verbes et les noms prédicatifs de ce lexique au format Alexina, qui est celui du lexique Lefff (Lexique des Formes Fléchies du Français) (Sagot, 2010), un lexique morphologique et syntaxique à large couverture et librement disponible pour le français. Ceci permet son intégration dans l'analyseur syntaxique FRMG (French MetaGrammar) (Thomasset et de La Clergerie, 2005), un analyseur profond à large couverture pour le français, basé sur les grammaires d'arbres adjoints (TAG), reposant habituellement sur le Lefff. Cette étape de conversion consiste à extraire l'information syntaxique codée dans les tables du Lexique-Grammaire. Nous présentons les fondements linguistiques de ce processus de conversion et le lexique obtenu. Nous évaluons l'analyseur syntaxique FRMG sur le corpus de référence de la campagne d'évaluation d'analyseurs du français Passage (Produire des Annotations Syntaxiques à Grande Échelle) (Hamon et al., 2008), en comparant sa version basée sur le Lefff avec notre version reposant sur les tables du Lexique-Grammaire converties / Lexicon-Grammar tables, whose development was initiated by Gross (1975), are a very rich syntactic lexicon for the French language. They cover various lexical categories such as verbs, nouns, adjectives and adverbs. This linguistic database is nevertheless not directly usable by computer programs, as it is incomplete and lacks consistency. Tables are defined on the basis of features which are not explicitly recorded in the lexicon. These features are only described in literature. To use these tables, we must make explicit the essential features appearing in each one of them. In addition, many features must be renamed for consistency sake. Our aim is to adapt the tables, so as to make them usable in various Natural Language Processing (NLP) applications, in particular parsing.We describe the problems we encountered and the approaches we followed to enable their integration into a parser. We propose LGExtract, a generic tool for generating a syntactic lexicon for NLP from the Lexicon-Grammar tables. It relies on a global table in which we added the missing features and on a single extraction script including all operations related to each property to be performed for all tables. We also present LGLex, the new generated lexicon of French verbs, predicative nouns, frozen expressions and adverbs.Then, we describe how we converted the verbs and predicatives nouns of this lexicon into the Alexina framework, that is the one of the Lefff lexicon (Lexique des Formes Fléchies du Français) (Sagot, 2010), a freely available and large-coverage morphological and syntactic lexicon for French. This enables its integration in the FRMG parser (French MetaGrammar) (Thomasset et de La Clergerie, 2005), a large-coverage deep parser for French, based on Tree-Adjoining Grammars (TAG), that usually relies on the Lefff. This conversion step consists in extracting the syntactic information encoded in Lexicon-Grammar tables. We describe the linguistic basis of this conversion process, and the resulting lexicon. We evaluate the FRMG parser on the reference corpus of the evaluation campaign for French parsersPassage (Produire des Annotations Syntaxiques à Grande Échelle) (Hamon et al., 2008), by comparing its Lefff-based version to our version relying on the converted Lexicon-Grammar tables

A sumarização multidocumento consiste na produção de um sumário/resumo a partir de uma coleção de textos sobre um mesmo assunto. Devido à grande quantidade de informação disponível na Web, esta tarefa é de grande relevância já que pode facilitar a leitura dos usuários. Os aspectos informativos representam as unidades básicas de informação presentes nos textos. Por exemplo, em textos jornalísticos em que se relata um fato/acontecimento, os aspectos podem representar a seguintes informações: o que aconteceu, onde aconteceu, quando aconteceu, como aconteceu, e por que aconteceu. Conhecendo-se esses aspectos e as estratégias de produção e organização de sumários, é possível automatizar a tarefa de sumarização. No entanto, para o Português do Brasil, não há pesquisa feita sobre sumarização com base em aspectos. Portanto, neste trabalho de mestrado, investigaram-se métodos de sumarização multidocumento com base em aspectos informativos, pertencente à abordagem profunda para a sumarização, em que se busca interpretar o texto para se produzir sumários mais informativos. Em particular, implementaram-se duas etapas relacionadas: (i) identificação automática de aspectos os aspectos informativos e (ii) desenvolvimento e avaliação de dois métodos de sumarização com base em padrões de aspectos (ou templates) em sumários. Na etapa (i), criaram-se classificadores de aspectos com base em anotador de papéis semânticos, reconhecedor de entidades mencionadas, regras manuais e técnicas de aprendizado de máquina. Avaliaram-se os classificadores sobre o córpus CSTNews (Rassi et al., 2013; Felippo et al., 2014). Os resultados foram satisfatórios, demostrando que alguns aspectos podem ser identificados automaticamente em textos jornalísticos com um desempenho razoável. Já na etapa (ii), elaboraram-se dois métodos inéditos de sumarização multidocumento com base em aspectos. Os resultados obtidos mostram que os métodos propostos neste trabalho são competitivos com os métodos da literatura. Salienta-se que esta abordagem para sumarização tem recebido grande destaque ultimamente. Além disso, é inédita nos trabalhos desenvolvidos no Brasil, podendo trazer contribuições importantes para a área. / Multi-document summarization is the task of automatically producing a unique summary from a group of texts on the same topic. With the huge amount of available information in the web, this task is very relevant because it can facilitate the reading of the users. Informative aspects, in particular, represent the basic information units in texts and summaries, e.g., in news texts there should be the following information: what happened, when it happened, where it happened, how it happened and why it happened. Knowing these aspects and the strategies to produce and organize summaries, it is possible to automate the aspect-based summarization. However, there is no research about aspect-based multi-document summarization for Brazilian Portuguese. This research work investigates multi-document summarization methods based on informative aspects, which follows the deep approach for summarization, in which it aims at interpreting the texts to produce more informative summaries. In particular, two main stages are developed: (i) the automatic identification of informative aspects and (ii) and the development and evaluation of two summarization methods based on aspects patterns (or templates). In the step (i) classifiers were created based on semantic role labeling, named entity recognition, handcrafted rules and machine learning techniques. Classifiers were evaluated on the CSTNews annotated corpus (Rassi et al., 2013; Felippo et al., 2014). The results were satisfactory, demonstrating that some aspects can be automatically identified in the news with a reasonable performance. In the step (ii) two novels aspect-based multi-document summarization methods are elaborated. The results show that the proposed methods in this work are competitive with the classical methods. It should be noted that this approach has lately received a lot of attention. Furthermore, it is unprecedented in the summarization task developed in Brazil, with the potential to bring important contributions to the area.

Apesar dos processos de recuperação e filtragem de informação sempre terem usado técnicas básicas de Processamento de Linguagem Natural (PLN) no suporte à estruturação de documentos, ainda são poucas as indicações sobre os avanços relacionados à utilização de técnicas mais sofisticadas de PLN que justifiquem o custo de sua utilização nestes processos, em comparação com as abordagens tradicionais. Este trabalho investiga algumas evidências que fundamentam a hipótese de que a aplicação de métodos que utilizam conhecimento linguístico é viável, demarcando importantes contribuições para o aumento de sua eficiência em adição aos métodos estatásticos tradicionais. É proposto um modelo de representação de texto fundamentado em sintagmas nominais, cuja representatividade de seus descritores é calculada utilizando-se o conceito de evidência, apoiado em métodos estatísticos. Filtros induzidos a partir desse modelo são utilizados para classificar os documentos recuperados analisando-se a relevância implícita no perfil do usuário. O aumento da precisão (e, portanto, da eficácia) em sistemas de Recuperação de Informação, conseqüência da pós-filtragem seletiva de informações, demonstra uma clara evidência de como o uso de técnicas de PLN pode auxiliar a categorização de textos, abrindo reais possibilidades para o aprimoramento do modelo apresentado / Although Information Retrieval and Filtering tasks have always used basic Natural Language Processing (NLP) techniques for supporting document structuring, there is still space for more sophisticated NLP techniques which justify their cost when compared to the traditional approaches. This research aims to investigate some evidences that justify the hypothesis on which the use of linguistic-based methods is feasible and can bring on relevant contributions to this area. In this work noun phrases of a text are used as descriptors whose evidence is calculated by statistical methods. Filters are then induced to classify the retrieved documents by measuring their implicit relevance presupposed by an user profile. The increase of precision (efficacy) in IR systems as a consequence of the use of NLP techniques for text classification in the filtering task is an evidence of how this approach can be further explored

O Reconhecimento de Entidades Mencionadas (REM) é uma subtarefa da extração de informações e tem como objetivo localizar e classificar elementos do texto em categorias pré-definidas tais como nome de pessoas, organizações, lugares, datas e outras classes de interesse. Esse conhecimento obtido possibilita a execução de outras tarefas mais avançadas. O REM pode ser considerado um dos primeiros passos para a análise semântica de textos, além de ser uma subtarefa crucial para sistemas de gerenciamento de documentos, mineração de textos, extração da informação, entre outros. Neste trabalho, estudamos alguns métodos de Aprendizado de Máquina aplicados na tarefa de REM que estão relacionados ao atual estado da arte, dentre eles, dois métodos aplicados na tarefa de REM para a língua portuguesa. Apresentamos três diferentes formas de avaliação destes tipos de sistemas presentes na literatura da área. Além disso, desenvolvemos um sistema de REM para língua portuguesa utilizando Aprendizado de Máquina, mais especificamente, o arcabouço de máxima entropia. Os resultados obtidos com o nosso sistema alcançaram resultados equiparáveis aos melhores sistemas de REM para a língua portuguesa desenvolvidos utilizando outras abordagens de aprendizado de máquina. / Named Entity Recognition (NER), a task related to information extraction, aims to classify textual elements according to predefined categories such as names, places, dates etc. This enables the execution of more advanced tasks. NER is a first step towards semantic textual analysis and is also a crucial task for systems of information extraction and other types of systems. In this thesis, I analyze some Machine Learning methods applied to NER tasks, including two methods applied to Portuguese language. I present three ways of evaluating these types of systems found in the literature. I also develop an NER system for the Portuguese language utilizing Machine Learning that entails working with a maximum entropy framework. The results are comparable to the best NER systems for the Portuguese language developed with other Machine Learning alternatives.

In this thesis, the Twitter social media is data mined for information about the covid-19 outbreak during the month of March, starting from the 3’rd and ending on the 31’st. 100,000 tweets were collected from Harvard’s opensource data and recreated using Hydrate. This data is analyzed further using different Natural Language Processing (NLP) methodologies, such as termfrequency inverse document frequency (TF-IDF), lemmatizing, tokenizing, Latent Semantic Analysis (LSA) and Latent Dirichlet Allocation (LDA). Furthermore, the results of the LSA and LDA algorithms is reduced dimensional data that will be clustered using clustering algorithms HDBSCAN and K-Means for later comparison. Different methodologies are used to determine the optimal parameters for the algorithms. This is all done in the python programing language, as there are libraries for supporting this research, the most important being scikit-learn. The frequent words of each cluster will then be displayed and compared with factual data regarding the outbreak to discover if there are any correlations. The factual data is collected by World Health Organization (WHO) and is then visualized in graphs in ourworldindata.org. Correlations with the results are also looked for in news articles to find any significant moments to see if that affected the top words in the clustered data. The news articles with good timelines used for correlating incidents are that of NBC News and New York Times. The results show no direct correlations with the data reported by WHO, however looking into the timelines reported by news sources some correlation can be seen with the clustered data. Also, the combination of LDA and HDBSCAN yielded the most desireable results in comparison to the other combinations of the dimnension reductions and clustering. This was much due to the use of GridSearchCV on LDA to determine the ideal parameters for the LDA models on each dataset as well as how well HDBSCAN clusters its data in comparison to K-Means.