SEQUENTIAL DECISIONS AND PREDICTIONS IN NATURAL LANGUAGE PROCESSING

Natural language processing has achieved great success in a wide range of ap- plications, producing both commercial language services and open-source language tools. However, most methods take a static or batch approach, assuming that the model has all information it needs and makes a one-time prediction. In this disser- tation, we study dynamic problems where the input comes in a sequence instead of all at once, and the output must be produced while the input is arriving. In these problems, predictions are often made based only on partial information. We see this dynamic setting in many real-time, interactive applications. These problems usually involve a trade-off between the amount of input received (cost) and the quality of the output prediction (accuracy). Therefore, the evaluation considers both objectives (e.g., plotting a Pareto curve). Our goal is to develop a formal understanding of sequential prediction and decision-making problems in natural language processing and to propose efficient solutions. Toward this end, we present meta-algorithms that take an existent batch model and produce a dynamic model to handle sequential inputs and outputs. Webuild our framework upon theories of Markov Decision Process (MDP), which allows learning to trade off competing objectives in a principled way. The main machine learning techniques we use are from imitation learning and reinforcement learning, and we advance current techniques to tackle problems arising in our settings. We evaluate our algorithm on a variety of applications, including dependency parsing, machine translation, and question answering. We show that our approach achieves a better cost-accuracy trade-off than the batch approach and heuristic-based decision- making approaches. We first propose a general framework for cost-sensitive prediction, where dif- ferent parts of the input come at different costs. We formulate a decision-making process that selects pieces of the input sequentially, and the selection is adaptive to each instance. Our approach is evaluated on both standard classification tasks and a structured prediction task (dependency parsing). We show that it achieves similar prediction quality to methods that use all input, while inducing a much smaller cost. Next, we extend the framework to problems where the input is revealed incremen- tally in a fixed order. We study two applications: simultaneous machine translation and quiz bowl (incremental text classification). We discuss challenges in this set- ting and show that adding domain knowledge eases the decision-making problem. A central theme throughout the chapters is an MDP formulation of a challenging problem with sequential input/output and trade-off decisions, accompanied by a learning algorithm that solves the MDP.

Heavy elements in chestnuts

Chestnut fruit (CasChestnuts, Castanea sativa, heavy elements, PIXEtanea sativa Miller) are an important food resource in several countries. Portugal is the third largest European producer, with an average production of 25 thousand tons [1], being mainly produced in the North region of Trás-os-Montes. Earlier studies on chestnuts elemental composition were performed by atomic absorption to detect Ca, Mg, Na, K, Cu, Fe, Mn and Zn or by UV-VIS spectrophotometry to detect P [2-3]. In this work the elemental composition of two cultivars (Longal and Judia) of Castanea sativa Miller were studied by means of PIXE (Proton Induced X-ray Emission), using a broad proton beam at CTN-IST, Lisbon. Element identification (from Si to Pb) and quantification (following the method described in [4]) was done. Preliminary results show different composition according with the cultivar, not only in the major and minor elements (as it is described in the bibliography) but also in the trace elements, which includes the presence of Pb and other heavy elements. These preliminary results are part of a wider project that aims to evaluate chestnut fruits decontamination treatment with high energy electron beam and X-rays technology.

KNOWING THE ENEMY: NAZI FOREIGN INTELLIGENCE IN WAR, HOLOCAUST, AND POSTWAR

“Knowing the Enemy: Nazi Foreign Intelligence in War, Holocaust and Postwar,” reveals the importance of ideologically-driven foreign intelligence reporting in the wartime radicalization of the Nazi dictatorship, and the continued prominence of Nazi discourses in postwar reports from German intelligence officers working with the U.S. Army and West German Federal Intelligence Service after 1945. For this project, I conducted extensive archival research in Germany and the United States, particularly in overlooked and files pertaining to the wartime activities of the Reichssicherheitshauptamt, Abwehr, Fremde Heere Ost, Auswärtiges Amt, and German General Staff, and the recently declassified intelligence files pertaining to the postwar activities of the Gehlen Organization, Bundesnachrichtendienst, and Foreign Military Studies Program. Applying the technique of close textual analysis to the underutilized intelligence reports themselves, I discovered that wartime German intelligence officials in military, civil service, and Party institutions all lent the appearance of professional objectivity to the racist and conspiratorial foreign policy beliefs held in the highest echelons of the Nazi dictatorship. The German foreign intelligence services’ often erroneous reporting on Great Britain, the Soviet Union, the United States, and international Jewry simultaneously figured in the radicalization of the regime’s military and anti-Jewish policies and served to confirm the ideological preconceptions of Hitler and his most loyal followers. After 1945, many of these same figures found employment with the Cold War West, using their “expertise” in Soviet affairs to advise the West German Government, U.S. Military, and CIA on Russian military and political matters. I chart considerable continuities in personnel and ideas from the wartime intelligence organizations into postwar West German and American intelligence institutions, as later reporting on the Soviet Union continued to reproduce the flawed wartime tropes of innate Russian military and racial inferiority.

X-ray fluorescence spectroscopic determination of heavy metals and trace elements in aerial parts of Origanum sipyleum L from Turkey

Purpose: To determine the heavy metal and trace element composition of the powdered aerial parts of Origanum sipyleum L. and its water extract. Methods: The heavy metal and trace elements content of the powdered plant material and 2 % aqueous extract were evaluated by x-ray fluorescence spectroscopy with silicon drift detector SDD at a resolution of 145 eV and 10,000 pulses. The process conditions were 0.1 g sample weight, process time of 300 s at a voltage of 25 kV and 50 kV, and at a current of 0.5 and 1.0 mA under helium atmosphere. Results: The major elements, K, Ca and Na, known as macronutrients, constituted 11990, 10490 and 970 ppm of the powdered drug and 8910, 2991 and 810 ppm of the water extract, respectively. Among other constituents, arsenic, lead and uranium levels were < 1, 2.1 and < 3 ppm, respectively, in the powdered material while in the aqueous extract, the levels were < 1, < 2 and 200 ppm, respectively. Conclusion: O. sipyleum is a potential source of macro- and micronutrients from which useful food additives and health supplements can be derived.

The Integration of Foreign Students in the Portuguese School of the 1st Cycle of Basic Education: the Case of a School Grouping in the Municipality of Viseu

The present study aims to understand whether the foreign students who have different nationalities but the Portuguese are integrated into the school of the 1st Cycle of Basic Education. With this purpose, a descriptive and phenomenological research was conducted, making use of documental analysis, as well as semi-structured interviews and sociometric tests. These two data collecting tools were applied to students attending from the 1st to the 4th school years, in three 1st Cycle of Basic Education schools, within a school grouping in Viseu. The data obtained through the interviews allow us to conclude that foreign students, in general, feel integrated both in the school and in the class they belong to. However, the analysis of the results of the sociometric tests reveals other data, allowing us to conclude that one of the students is neither integrated in the school, nor in the class he is part of.

Determination of rare earth elements in geological samples using the agilent SVDV ICP-OES.

2016

Towards Unstructured Knowledge Integration in Natural Language Processing

In the last decades, Artificial Intelligence has witnessed multiple breakthroughs in deep learning. In particular, purely data-driven approaches have opened to a wide variety of successful applications due to the large availability of data. Nonetheless, the integration of prior knowledge is still required to compensate for specific issues like lack of generalization from limited data, fairness, robustness, and biases. In this thesis, we analyze the methodology of integrating knowledge into deep learning models in the field of Natural Language Processing (NLP). We start by remarking on the importance of knowledge integration. We highlight the possible shortcomings of these approaches and investigate the implications of integrating unstructured textual knowledge. We introduce Unstructured Knowledge Integration (UKI) as the process of integrating unstructured knowledge into machine learning models. We discuss UKI in the field of NLP, where knowledge is represented in a natural language format. We identify UKI as a complex process comprised of multiple sub-processes, different knowledge types, and knowledge integration properties to guarantee. We remark on the challenges of integrating unstructured textual knowledge and bridge connections with well-known research areas in NLP. We provide a unified vision of structured knowledge extraction (KE) and UKI by identifying KE as a sub-process of UKI. We investigate some challenging scenarios where structured knowledge is not a feasible prior assumption and formulate each task from the point of view of UKI. We adopt simple yet effective neural architectures and discuss the challenges of such an approach. Finally, we identify KE as a form of symbolic representation. From this perspective, we remark on the need of defining sophisticated UKI processes to verify the validity of knowledge integration. To this end, we foresee frameworks capable of combining symbolic and sub-symbolic representations for learning as a solution.

Textos científicos de autoria de graduandos em química: análise dos professores

This study aimed to analyze the aspects of scientific language considered by chemistry professors in the evaluation of scientific texts authored by undergraduate chemistry students. Interviews were conducted with four professors who had evaluated reports and scientific articles written by students enrolled in undergraduate chemistry courses. The professors highlighted aspects pertaining to the structure and general organization of scientific texts typically adopted by the scientific community, and mentioned the importance of certain rhetoric elements in scientific language. This study can be an impetus for further investigation into the importance of language in chemistry education.

La protection du consommateur dans le commerce international passé par Internet: une analyse comparée des systèmes juridiques européen, français, canadien et québécois

Avec le développement exponentiel de l'Internet et son corollaire l'expansion du commerce en ligne, le sort de la protection du cyberconsommateur devient un sujet préoccupant en ce 21ième siècle. En effet, dans ce monde virtuel où l'on emploie des méthodes et technologies nouvelles et plus encore des clauses abusives dans les contrats unilatéraux, s'installe indubitablement un sentiment de méfiance entre le cyberconsommateur et le cybercommerçant. Pour rétablir cette confiance et favoriser le commerce par Internet, des lois nationales, internationales et des normes communautaires ont été adoptées aux fins de l'encadrement rigoureux du processus contractuel. Toutefois, en raison de la présence fréquente d'éléments d'extranéité dans les contrats de consommation en ligne, la question fondamentale qui vient tout naturellement à l'esprit de tous ceux qui entreprennent des études aujourd'hui, en la matière, est celle de savoir si les règles classiques de droit international privé sont dépassées par le développement trop rapide de ce type de commerce ou si au contraire elles y sont adaptées. On pourrait en outre se demander si l'encadrement juridique offert au cyberconsommateur est à même de lui procurer le même niveau de protection dont il bénéficie dans le commerce traditionnel. La présente étude tente d'apporter certains éléments de réponse en analysant dans un premier temps, le droit substantiel interne de protection du consommateur dans les systèmes juridiques européen, français, canadien et québécois en vu de scruter des zones de conflits susceptibles d'exister dans le cycle de vie de ce contrat. Dans la seconde partie, elle démontre que les méthodes classiques de résolution des conflits de juridiction et des conflits de lois en droit international privé, bien que nécessitant des adaptations, sont effectivement applicables au contexte de l'internet et ce, dans l'objectif privilégié de la protection du cyberconsommateur. Le bilan de l'analyse et de l'appréciation des critères de ces règles de conflits nous conduiront à l'examen des nouvelles mesures qui s'imposent.

Discerning Cultural Depictions and Mistakes in Kafe Piter, a Russian Language Textbook for Finnish Students

This thesis presents the results of an analysis of the content in the series of Russian textbooks Kafe Piter, which is widely used in Finnish educational institutions for adult learners at the time that the research is conducted. The purpose of this study is to determine and describe how a textbook may purvey an image of a foreign country (in this case, Russia). Mixed-methods research with a focus on the qualitative content analysis of Kafe Piter is performed. The guidelines for textbook evaluation of cultural content proposed by Byram (1993) are used in this study as the basis for creating a qualitative analysis checklist, which is adopted according to the needs of the current research. The selection of the categories in the checklist is based on major themes where direct statements about Russia, Russian people and culture appear in the textbook. The cultural content and the way in which it is presented in Kafe Piter are also compared to the intercultural competence objectives of the Common European Framework of Reference for Languages. Because the textbook was not written by a native Russian speaker, it was also important to investigate the types of mistakes found in the books. A simple quantitative analysis in the form of descriptive statistics was done, which consisted of counting the mistakes and inaccuracies in Kafe Piter. The mistakes were categorized into several different groups: factual or cultural, lexicosemantic, grammatical, spelling and punctuation mistakes. Based on the results, the cultural content of Kafe Piter provides a rich variety of cultural information that allows for a good understanding of the Russian language and Russian culture. A sufficient number of cross-cultural elements also appear in the textbook, including cultural images and information describing and comparing Russian and Finnish ways of life. Based on the cultural topics covered in Kafe Piter, we conclude that the textbook is in line with the intercultural competence objectives set out in the Common European Framework of Reference for Languages. The results of the study also make it clear that a thorough proofreading of Kafe Piter is needed in order to correct mistakes - more than 130 cultural and linguistic mistakes and inaccuracies appear in the textbook.

Machine Translation and the Disruption of Foreign Language Learning Activities

This study examines the question of how language teachers in a highly technologyfriendly university environment view machine translation and the implications that this has for the personal learning environments of students. It brings an activity-theory perspective to the question, examining the ways that the introduction of new tools can disrupt the relationship between different elements in an activity system. This perspective opens up for an investigation of the ways that new tools have the potential to fundamentally alter traditional learning activities. In questionnaires and group discussions, respondents showed general agreement that although use of machine translation by students could be considered cheating, students are bound to use it anyway, and suggested that teachers focus on the kinds of skills students would need when using machine translation and design assignments and exams to practice and assess these skills. The results of the empirical study are used to reflect upon questions of what the roles of teachers and students are in a context where many of the skills that a person needs to be able to interact in a foreign language increasingly can be outsourced to laptops and smartphones.

Machine translation and the disruption of foreign language learning activities

This study examines the question of how language teachers in a highly technology-friendly university environment view machine translation and the implications that this has for the personal learning environments of students. It brings an activity-theory perspective to the question, examining the ways that the introduction of new tools can disrupt the relationship between different elements in an activity system. This perspective opens up for an investigation of the ways that new tools have the potential to fundamentally alter traditional learning activities. In questionnaires and group discussions, respondents showed general agreement that although use of machine translation by students could be considered cheating, students are bound to use it anyway, and suggested that teachers focus on the kinds of skills students would need when using machine translation and design assignments and exams to practice and assess these skills. The results of the empirical study are used to reflect upon questions of what the roles of teachers and students are in a context where many of the skills that a person needs to be able to interact in a foreign language increasingly can be outsourced to laptops and smartphones.