Suchergebnisse

Microevolution due to pollution can occur mainly through genetic drift bottlenecks, especially of small sized populations facing intense lethal pulses of contaminants, through mutations, increasing allelic diversity, and through natural selection, with the disappearance of the most sensitive genotypes. This loss of genotypes can lead to serious effects if coupled to specific hypothetical scenarios. These may be categorized as leading, first, to the loss of alleles—the recessive tolerance inheritance hypothesis. Second, leading to a reduction of the population growth rate—the mutational load and fitness costs hypotheses. Third, leading to an increased susceptibility of further genetic erosion both at future inputs of the same contaminant—differential physiological recovery, endpoints (dis)association, and differential phenotypic plasticity hypotheses—and at sequential or simultaneous inputs of other contaminants—the multiple stressors differential tolerance hypothesis. Species in narrowly fluctuating environments (tropics and deep sea) may have a particularly high susceptibility to genetic erosion—the Plus c¸a change (plus c'est la meme chose) hypothesis. A discussion on the consequences of these hypotheses is what this essay aimed at. ; This study was partially funded by FSE and POPH (Ciência 2007), by national funds (OE) through Foundation for Science and Technology (FCT) and Ministério da Ciência, Tecnologia e Ensino Superior (MCTES) (http://alfa.fct. mctes.pt) and co-funded by the European Union (project ref. PTDC/ AAC-AMB/104532/2008).

All over the world, many initiatives have been taken to fight fake news. Governments (e.g., France, Germany, United Kingdom and Spain), on their own way, started to take action regarding legal accountability for those who manufacture or propagate fake news. Different media outlets have also taken a multitude of initiatives to deal with this phenomenon, such as the increase of discipline, accuracy and transparency of publications made internally. Some structural changes have lately been made in said companies and entities in order to better evaluate news in general. As such, many teams were built entirely to fight fake news - the so-called "fact-checkers". These have been adopting different techniques in order to do so: From the typical use of journalists to find out the true behind a controversial statement, to data-scientists that apply forefront techniques such as text mining and machine learning to support the journalist's decisions. Many of these entities, which aim to maintain or improve their reputation, started to focus on high standards for quality and reliable information, which led to the creation of official and dedicated departments for fact-checking. In this revision paper, not only will we highlight relevant contributions and efforts across the fake news identification and classification status quo, but we will also contextualize the Portuguese language state of affairs in the current state-of-the-art. ; info:eu-repo/semantics/publishedVersion

Abstract Brazil is signatory of the Global Strategy for Plant Conservation (GSPC), which provides guidelines and directions to existing national policies. This strategy aims to halt the continuing loss of plant diversity through the achievement of 16 outcome-based targets set for 2020. One of these targets (target 7) states that at least 75% of known threatened plant species should be preserved in situ. Here, we assessed the effectiveness of the Brazilian current network of protected areas (PAs) and indigenous lands (ILs) in representing all known threatened plant species. We found that the number of species represented inside PAs and ILs varied according to data type. When using occurrence records, we found that 699 (33%) threatened plant species lie completely outside PAs (and/or ILs) and that 1,405 species (67%) have at least one record inside at least one PA (and/or IL). The number of species unrepresented decreased when we considered polygons of distribution. In this case, only 219 (10%) are supposedly unprotected. Although Brazil is almost reaching GSPC Target 7 in terms of absolute numbers, the government still needs to allocate resources for properly managing and improving the conservation status of its imperiled flora and expand the network of PAs.

The process of protecting sensitive data is continually growing and becoming increasingly important, especially as a result of the directives and laws imposed by the European Union. The effort to create automatic systems is continuous, but, in most cases, the processes behind them are still manual or semi-automatic. In this work, we have developed a component that can extract and classify sensitive data, from unstructured text information in European Portuguese. The objective was to create a system that allows organizations to understand their data and comply with legal and security purposes. We studied a hybrid approach to the problem of Named Entity Recognition for the Portuguese language. This approach combines several techniques such as rule-based/lexical-based models, machine learning algorithms, and neural networks. The rule-based and lexical-based approaches were used only for a set of specific classes. For the remaining classes of entities, two statistical models were tested—Conditional Random Fields and Random Forest and, finally, a Bidirectional-LSTM approach as experimented. Regarding the statistical models, we realized that Conditional Random Fields is the one that can obtain the best results, with a f1-score of 65.50%. With the Bi-LSTM approach, we have achieved a result of 83.01%. The corpora used for training and testing were HAREM Golden Collection, SIGARRA News Corpus, and DataSense NER Corpus. ; info:eu-repo/semantics/publishedVersion

The process of sensitive data preservation is a manual and a semi-automatic procedure. Sensitive data preservation suffers various problems, in particular, affect the handling of confidential, sensitive and personal information, such as the identification of sensitive data in documents requiring human intervention that is costly and propense to generate error, and the identification of sensitive data in large-scale documents does not allow an approach that depends on human expertise for their identification and relationship. DataSense will be highly exportable software that will enable organizations to identify and understand the sensitive data in their possession in unstructured textual information (digital documents) in order to comply with legal, compliance and security purposes. The goal is to identify and classify sensitive data (Personal Data) present in large-scale structured and non-structured information in a way that allows entities and/or organizations to understand it without calling into question security or confidentiality issues. The DataSense project will be based on European-Portuguese text documents with different approaches of NLP (Natural Language Processing) technologies and the advances in machine learning, such as Named Entity Recognition, Disambiguation, Co-referencing (ARE) and Automatic Learning and Human Feedback. It will also be characterized by the ability to assist organizations in complying with standards such as the GDPR (General Data Protection Regulation), which regulate data protection in the European Union. ; info:eu-repo/semantics/acceptedVersion

Functionally graded aluminium matrix composites reinforced with SiC particles are attractive materials for a broad range of engineering applications in the automotive, aircraft, sports, military and aerospace industries, whenever a superior combination of surface and bulk mechanical properties are required. In general, these materials are developed for the production of high wear resistant components. Also, often this kind of mechanical part operates in the presence of aggressive environments, such as marine atmospheres. In this work, aluminium composites with functionally graded properties, obtained by centrifugal cast, are characterised by reciprocating pin-on-plate sliding wear tests against nodular cast iron. Three different volume fractions of SiC reinforcing particles in each functionally graded material were considered. Sliding experiments were performed with and without the presence of a lubricant (3% NaCl aqueous solution). All tests were carried out at room temperature, under a normal load of 10N and constant frequency (1 Hz) and stroke (6 mm). In the case of the lubricated tests, electrochemical parameters (corrosion potential) were monitored during sliding. The worn surfaces as well as the wear debris were characterised by SEM/EDS. Friction values were in the order of 0.42 for unlubricated conditions, but varied between 0.22 and 0.37 when the aqueous solution was present. Relatively high wear rates (over 110-6 gm-1) were obtained in both unlubricated and lubricated sliding. The volume fraction of SiC particles exerted a net effect on the tribological response of the composites, although conditioned by the presence or absence of the aqueous solution. A deleterious effect of the chloride aqueous solution on the degradation of the matrix, leading to the pull-out of SiC particles induced by the dissolution of the matrix/particle interface was evidenced. Observation of the worn surface morphology indicated that the presence of the lubricant modifies the protective action promoted by the combined effect ...