Graph-induced restricted Boltzmann machines for document modeling

Discovering knowledge from unstructured texts is a central theme in data mining and machine learning. We focus on fast discovery of thematic structures from a corpus. Our approach is based on a versatile probabilistic formulation – the restricted Boltzmann machine (RBM) –where the underlying graphical model is an undirected bipartite graph. Inference is efficient document representation can be computed with a single matrix projection, making RBMs suitable for massive text corpora available today. Standard RBMs, however, operate on bag-of-words assumption, ignoring the inherent underlying relational structures among words. This results in less coherent word thematic grouping. We introduce graph-based regularization schemes that exploit the linguistic structures, which in turn can be constructed from either corpus statistics or domain knowledge. We demonstrate that the proposed technique improves the group coherence, facilitates visualization, provides means for estimation of intrinsic dimensionality, reduces overfitting, and possibly leads to better classification accuracy.

A estratégia militar brasileira e o Exército: uma análise da política declaratória de defesa

A Estratégia Nacional de Defesa (END) foi o documento de Defesa Nacional que mais repercutiu nas Forças Armadas e, também, na sociedade brasileira. A END formula diretrizes estratégicas para alcançar os objetivos nacionais de defesa, que estão inseridos na Política Nacional de Defesa (PND). A PND é o documento de mais alto nível do Estado brasileiro para o planejamento de defesa. Para atingir os objetivos definidos na END, o comandante do Exército Brasileiro expediu suas diretrizes, seguidas de um planejamento denominado “Estratégia Braço Forte” (EBF). Posteriormente, por meio do Estado-Maior do Exército (EME), foi produzido um documento intitulado “Processo de Transformação do Exército Brasileiro” (PTEB) cujo teor é fruto de resultados de estudos, diagnósticos, formulações e concepções surgidas no EME ao longo dos trabalhos relativos à END e à elaboração da EBF. Em complemento às medidas de gestão para os assuntos de defesa, o Governo Federal disponibilizou, em julho de 2012, o Livro Branco de Defesa Nacional (LBDN), entregando-o ao Congresso Nacional para apreciação. Vários documentos sobre o tema defesa foram formulados a partir da END. Estes documentos estão enquadrados no nível estratégico, definindo quem é o responsável por seu planejamento e por sua condução. Assim, este trabalho tem como objetivo realizar a análise da Estratégia Nacional de Defesa e apresentar os documentos decorrentes afetos à Defesa Nacional, atinentes ao Exército Brasileiro que podem ser relacionados ao nível estratégico. Para tanto, fez-se uma discussão do conceito de estratégia e níveis estratégicos e realizou-se uma análise da END, considerando críticas existentes sobre este documento. Em seguida, o Livro Branco de Defesa Nacional foi apresentado e uma comparação com a END foi realizada. Em continuação, apresenta-se os dois documentos do Exército elaborados como conseqüência da END: a “Estratégia Braço Forte” e o “Processo de Transformação do Exército Brasileiro”. Por fim, apresenta-se os níveis estratégicos dos documentos mencionados e a coerência dos níveis estratégicos propostos por Luttwak.

As desventuras do duty to mitigate the loss no Brasil : nascimento (e morte) de um brocardo

Primeiramente, o trabalho descreve sinteticamente a tradição common law, inserindo o o duty to mitigate the loss em contexto próprio. Então, traça as linhas gerais que, naquele cenário, conformam o instituto. Atenção é dada aos fundamentos do duty to mitigate the loss e às funções por ele desempenhadas, inclusive nos casos subordinados à United Nations Convention on Contracts for the International Sale of Goods (CISG). No contexto próprio, o duty to mitigate the loss tem por fundamento primeiro a causation e funciona como um limitador do quantum indenizatório; não trata de um “dever”; cabe ao demandante; tem na razoabilidade das medidas mitigadoras e no reembolso das despesas incorridas características essenciais. Em segundo lugar, a dissertação investiga o duty to mitigate the loss no Brasil, a partir da doutrina selecionada e de decisões do Superior Tribunal de Justiça e de diversos tribunais estaduais. As pesquisas demonstram que o duty to mitigate the loss, sob o nome de mitigation doctrine, ingressou no Brasil pela doutrina, mas desta forma repercutiu pouco nos tribunais. Desprovido desta influência doutrinária anterior, o duty to mitigate the loss retornou ao Brasil por meio de enunciado do Conselho da Justiça Federal, proposto em documento cujo conteúdo é objeto de reflexões neste trabalho, notadamente na questão da íntima relação entre o duty to mitigate the loss e a boa-fé objetiva. A partir do leading case no Superior Tribunal de Justiça, que adotou o enunciado e o documento que lhe serviu de proposta, o duty to mitigate the loss expandiu rapidamente no Brasil. Em São Paulo, foi alçado à princípio e brocardo. No Superior Tribunal e Justiça, foi considerado sub-princípio da boa-fé e aplicado em questões adjetivas e substantivas criminais. Comparativamente, o duty to mitigate the loss no Brasil guarda remotas semelhanças ao instituto homônimo da common law. Este trabalho aponta as diferenças entre os intitutos e concluí com a sistematização das funções que o duty to mitigate the loss desempenha no Brasil, formulando proposta para a reedição de enunciado do Conselho da Justiça Federal.

Game Design Document (GDD)

Documento componente do jogo “Musikinésia (http://www.loa.sead.ufscar.br/musikinesia.php)” desenvolvido pela equipe do Laboratório de Objetos de Aprendizagem da Universidade Federal de São Carlos (LOA/UFSCar).

Game Design Document (GDD) - Versão 1

Documento componente do jogo “Digestower (http://www.loa.sead.ufscar.br/digestower.php)” desenvolvido pela equipe do Laboratório de Objetos de Aprendizagem da Universidade Federal de São Carlos (LOA/UFSCar).

O Uso de operações linguístico-discursivas da crítica genética na reescritura de textos

As a professor in Curso de Licenciatura em Letras, from Campus Avançado Profa Maria Eliza Albuquerque Maia (CAMEAM),do Estado do Estado do Rio Grande do Norte (UERN), in the town of Pau do Ferros, in the state of Rio Grande do Norte, we had the chance to carry out several writing activities , as well as guiding re-writing activities for the texts produced. From this experience, we started looking at the need of reflecting upon the writing process in higher education. Thus, we aim at analyzing, in this research, the methodology used in the moment of carrying out the writing practices activities in higher education, investigating, in particular, the rewriting practices, concerning the operations used for carrying out such activities, as well as the sense effects produced from the alterations which were made in the texts. Our theoretical foundation is grounded on a conception of text as a verbal action , what reveals a socio-interactional view of the language (MARCUSCHI, 2008; SAUTCHUK, 2003). As the production of written texts, our research focus, we assume that, for this activity, we deal with distinct figures (active writer and internal reader), so that we can, apart from writing, reflecting upon our writing and, this way, deciding about operations which are carried out to make the alterations which are necessary to the rewriting of our texts (SAUTCHUK, 2003). Still about the theoretical foundations used in this research, we made use of the theories from the Textual Analysis of Discourse (TAD) which discusses the belief on the evidence on the existence of the texts, which is opposite to the fixist view of textuality which believes that the texts exist by themselves. (ADAM, 2008; [2005]2010). Under this perspective, we have also adopted, the concepts which come from genetics criticism which is concerned about the relation between text and genesis, using as objects documents which bring traits of the text in progress, on the ground that the text is the result of work in progress, and the writing practice, on the other hand, as an activity in a continuous movement (HAY, [1975]2002; DE BIASI, [2000]2010; GRÉSILLON, 1989; [1990]2008; [1992]2002; SALLES, 2008a). The methodology in this research is an ethnography-based one, an approach which focuses on the process, as well as is meaning-based. To understand the objectives proposed in our research, we made use of different procedures of collecting data which include an ethnographic study, such as: observation, note-taking, document analysis. The data which were analyzed were collected during the semester of 2008.2, in a first term classroom of Curso de Letras from CAMEAM, when we were able to collect twenty-one written texts and all of them were rewritten based on rewriting activities, what provides a corpus of forty-two texts which will be analyzed based on the linguistics operations identified by Generative Grammar and adopted by Lebrave and Grésillon (2009). From these analyses, we were able to confirm that writing is a process, and rewriting has become an extremely important activity for this process. Still due to these data, we observed that substitution was the most used operation by text authors. We believe that this result is justified by the fact that the substitution, according to what proposes the Genetic Criticism, constitutes the source of all erasure, from which one can easily make a change in writing. Regarding the operations of addition and deletion, we found that they were used in quantitative terms, almost equivalently, which can be explained when we see that the two operations require, by the author of the text, different strategies from those used for the replacement, what includes , respectively, adding or removing a segment. Finally, we found out that the shift operation was the least used, since it works with a segment that will not be replaced, added or deleted, but transferred to another place of text, which requires a greater ability of the author to perform this operation and not compromising the meaning of his/her writing. As a result, we hope to contribute to the reflection on the teaching of writing, considering, in a particular way, those with a Bachelor in Arts. Our analysis will contribute to the teaching of Portuguese language, specifically for activities that guide the production of texts in order to explore with students the ability to rewrite their own text

Solving the conundrum of Job: a probable biblical description of chronic renal failure with neurological symptoms

A doença descrita no livro bíblico de Jó é controversa, e tem interessado a teólogos, psiquiatras e dermatologistas, há tempos. Neste trabalho os autores apontam para evidências do diagnóstico de insuficiência renal crônica com alterações neurológicas.

Digital review 2007 of Latin America and the Caribbean: progress and challenges policies for development with ICT. Abridged document

Includes bibliography

Selecting candidate labels for hierarchical document clusters using association rules

One way to organize knowledge and make its search and retrieval easier is to create a structural representation divided by hierarchically related topics. Once this structure is built, it is necessary to find labels for each of the obtained clusters. In many cases the labels have to be built using only the terms in the documents of the collection. This paper presents the SeCLAR (Selecting Candidate Labels using Association Rules) method, which explores the use of association rules for the selection of good candidates for labels of hierarchical document clusters. The candidates are processed by a classical method to generate the labels. The idea of the proposed method is to process each parent-child relationship of the nodes as an antecedent-consequent relationship of association rules. The experimental results show that the proposed method can improve the precision and recall of labels obtained by classical methods. © 2010 Springer-Verlag.

Improving hierarchical document cluster labels through candidate term selection

One way to organize knowledge and make its search and retrieval easier is to create a structural representation divided by hierarchically related topics. Once this structure is built, it is necessary to find labels for each of the obtained clusters. In many cases the labels must be built using all the terms in the documents of the collection. This paper presents the SeCLAR method, which explores the use of association rules in the selection of good candidates for labels of hierarchical document clusters. The purpose of this method is to select a subset of terms by exploring the relationship among the terms of each document. Thus, these candidates can be processed by a classical method to generate the labels. An experimental study demonstrates the potential of the proposed approach to improve the precision and recall of labels obtained by classical methods only considering the terms which are potentially more discriminative. © 2012 - IOS Press and the authors. All rights reserved.