Structured Approaches for Exploring Interpersonal Relationships in Natural Language Text

Human relationships have long been studied by scientists from domains like sociology, psychology, literature, etc. for understanding people's desires, goals, actions and expected behaviors. In this dissertation we study inter-personal relationships as expressed in natural language text. Modeling inter-personal relationships from text finds application in general natural language understanding, as well as real-world domains such as social networks, discussion forums, intelligent virtual agents, etc. We propose that the study of relationships should incorporate not only linguistic cues in text, but also the contexts in which these cues appear. Our investigations, backed by empirical evaluation, support this thesis, and demonstrate that the task benefits from using structured models that incorporate both types of information. We present such structured models to address the task of modeling the nature of relationships between any two given characters from a narrative. To begin with, we assume that relationships are of two types: cooperative and non-cooperative. We first describe an approach to jointly infer relationships between all characters in the narrative, and demonstrate how the task of characterizing the relationship between two characters can benefit from including information about their relationships with other characters in the narrative. We next formulate the relationship-modeling problem as a sequence prediction task to acknowledge the evolving nature of human relationships, and demonstrate the need to model the history of a relationship in predicting its evolution. Thereafter, we present a data-driven method to automatically discover various types of relationships such as familial, romantic, hostile, etc. Like before, we address the task of modeling evolving relationships but don't restrict ourselves to two types of relationships. We also demonstrate the need to incorporate not only local historical but also global context while solving this problem. Lastly, we demonstrate a practical application of modeling inter-personal relationships in the domain of online educational discussion forums. Such forums offer opportunities for its users to interact and form deeper relationships. With this view, we address the task of identifying initiation of such deeper relationships between a student and the instructor. Specifically, we analyze contents of the forums to automatically suggest threads to the instructors that require their intervention. By highlighting scenarios that need direct instructor-student interactions, we alleviate the need for the instructor to manually peruse all threads of the forum and also assist students who have limited avenues for communicating with instructors. We do this by incorporating the discourse structure of the thread through latent variables that abstractly represent contents of individual posts and model the flow of information in the thread. Such latent structured models that incorporate the linguistic cues without losing their context can be helpful in other related natural language understanding tasks as well. We demonstrate this by using the model for a very different task: identifying if a stated desire has been fulfilled by the end of a story.

Recent educational changes at higher educational level in Pakistan: English language teachers' perceptions and practices

Higher Education Institutes (HEIs) of any country could be a source of providing professionals to the country in many fields. By doing so, HEIs could play a pivotal role in the economic growth of the country. In Pakistan, it seems that, in the wake of this realization, steps have been taken to reform Higher Education. Drawing on the Triple I model of educational change covering Initiation, Implementation and Institutionalization (Fullan, 2007) this study focuses on the planning and implementation of reforms in the Education system of Pakistan at higher education level that have been introduced by the Higher Education Commission (HEC) since its inception in 2002. Kennedy’s model of hierarchical subsystems affecting innovation and Chin and Benne’s (1985) description of strategies for implementing change also provided guidelines for analyzing the changes in education in the country to highlight the role that the authorities expect the language teacher to play in the process of implementing these changes. A qualitative method is followed in this study to gather data from English language teachers at three universities of the Khyber Pakhtunkhwa province of Pakistan. A questionnaire was developed to look into the perceptions of English language teachers regarding the impact of these reforms. This was followed up by interviews. Responses from 28 teachers were received through questionnaire out of which 9 teachers were interviewed for detailed analysis of their perceptions. Thematic Content analysis was used to analyze and interpret the data. Some of the most significant changes that the respondents reported knowledge of included the introduction of Semester System, extending the Bachelors degree to four years from two years, promotion of research culture, and increased teachers’ autonomy in classroom practices. Implications of these reforms for English teachers’ professional development were also explored. The data indicate that the teachers generally have a positive attitude towards the changes. However, the data also show concerns that teachers have about the practical effectiveness of these changes in improving English language teaching and learning in Pakistani Universities. Some of the areas of concern are worries regarding resources, the assessment system, the number of qualified teachers, and instability in the educational policy. They are concerned about the training facilities and quality of the professional training available to them. Moreover, they report that training opportunities for their professional development are not available to all the teachers equally. Despite the HEC claims of providing regular training opportunities, the majority of the teachers did not receive any formal training in the last three years, while some teachers were able to access these opportunities multiple times. Through the recent reforms HEC has empowered the teachers in conducting the learning/teacher processes but this extra power has reduced their accountability and they can exercise these powers without any check on them. This empowerment is limited to the classroom and there appears to be no or minimal involvement in decision making at the top level of policy making. Such lack of involvement in the policy decisions seems to be generating a lack of sense of ownership among the teachers (Fullan 2003a:6). Although Quality Enhancement Cells have been developed in the universities to assure the desired quality of education, they might need a more active role to contribute in achieving the level of enhancement in education expected from them. Based on the perceptions of the respondents of this study and the review of the relevant literature, it is argued that it is unlikely for the reforms to be institutionalized if teachers are not given the right kind of awareness at the initiation stage and are not prepared at the implementation stage to cope with the challenge of a complex process. The teachers participating in this study, in general, have positive and enthusiastic attitudes towards most of the changes, in spite of some reservations. It could also be interesting to see if the power centers of the Pakistani Higher Education appreciate this enthusiasm and channel it for a strong Higher Education system in the country.

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.

Clinical education reflective ecological model for health science majors

The clinical education is an integral part of the Health Science majors’ curriculum programs of the University of Aveiro’s School of Health (i.e., Nursing, Physical Therapy, Radiology, Radiotherapy and Speech-Language Pathology) and aims to develop clinical competences in order to generate excellent health care professionals. The organization was based on the Ecological Model of Clinical-Reflective Training, which was characterized by inter-institutional interaction and student’s reflection on actions on a professional setting. This study encompassed two moments of clinical internships in the Nursing, Physical Therapy, Radiology and Radiotherapy majors. The Clinical Internship I provided the 123 students with a global view of the health care professional activities. The Clinical Internship II, with 119 students, developed competences of each health professional. Questionnaires with categorical scales from 1 to 5 evaluated the organization and efficiency of the two internships. The results revealed averages over 3 in all items. In conclusion, the Ecological Model of Clinical-Reflective Training was well accepted by students and clinical supervisors. Applications in the health care area were demonstrated.

Towards Distributed Model Transformations with LinTra

Performance and scalability of model transformations are becoming prominent topics in Model-Driven Engineering. In previous works we introduced LinTra, a platform for executing model transformations in parallel. LinTra is based on the Linda model of a coordination language and is intended to be used as a middleware where high-level model transformation languages are compiled. In this paper we present the initial results of our analyses on the scalability of out-place model-to-model transformation executions in LinTra when the models and the processing elements are distributed over a set of machines.

EFFECTIVE FUNCTION CHOICE IN THE R SCRIPTING LANGUAGE

This project examines the current available work on the explicit and implicit parallelization of the R scripting language and reports on experimental findings for the development of a model for predicting effective points for automatic parallelization to be performed, based upon input data sizes and function complexity. After finding or creating a series of custom benchmarks, an interval based on data size and time complexity where replacement becomes a viable option was found; specifically between O(N) and O(N3) exclusive. As data size increases, the benefits of parallel processing become more apparent and a point is reached where those benefits outweigh the costs in memory transfer time. Based on our observations, this point can be predicted with a fair amount of accuracy using regression on a sample of approximately ten data sizes spread evenly between a system determined minimum and maximum size.

Funtional Near Infrared Spectroscopy Study of Language, Joint Attention and Motor Skills

Near infrared spectroscopy (NIRS) is an emerging non-invasive optical neuro imaging technique that monitors the hemodynamic response to brain activation with ms-scale temporal resolution and sub-cm spatial resolution. The overall goal of my dissertation was to develop and apply NIRS towards investigation of neurological response to language, joint attention and planning and execution of motor skills in healthy adults. Language studies were performed to investigate the hemodynamic response, synchrony and dominance feature of the frontal and fronto-temporal cortex of healthy adults in response to language reception and expression. The mathematical model developed based on granger causality explicated the directional flow of information during the processing of language stimuli by the fronto-temporal cortex. Joint attention and planning/ execution of motor skill studies were performed to investigate the hemodynamic response, synchrony and dominance feature of the frontal cortex of healthy adults and in children (5-8 years old) with autism (for joint attention studies) and individuals with cerebral palsy (for planning/execution of motor skills studies). The joint attention studies on healthy adults showed differences in activation as well as intensity and phase dependent connectivity in the frontal cortex during joint attention in comparison to rest. The joint attention studies on typically developing children showed differences in frontal cortical activation in comparison to that in children with autism. The planning and execution of motor skills studies on healthy adults and individuals with cerebral palsy (CP) showed difference in the frontal cortical dominance, that is, bilateral and ipsilateral dominance, respectively. The planning and execution of motor skills studies also demonstrated the plastic and learning behavior of brain wherein correlation was found between the relative change in total hemoglobin in the frontal cortex and the kinematics of the activity performed by the participants. Thus, during my dissertation the NIRS neuroimaging technique was successfully implemented to investigate the neurological response of language, joint attention and planning and execution of motor skills in healthy adults as well as preliminarily on children with autism and individuals with cerebral palsy. These NIRS studies have long-term potential for the design of early stage interventions in children with autism and customized rehabilitation in individuals with cerebral palsy.

Intersubjectivity and sentiment:from language to knowledge

Intersubjectivity is an important concept in psychology and sociology. It refers to sharing conceptualizations through social interactions in a community and using such shared conceptualization as a resource to interpret things that happen in everyday life. In this work, we make use of intersubjectivity as the basis to model shared stance and subjectivity for sentiment analysis. We construct an intersubjectivity network which links review writers, terms they used, as well as the polarities of the terms. Based on this network model, we propose a method to learn writer embeddings which are subsequently incorporated into a convolutional neural network for sentiment analysis. Evaluations on the IMDB, Yelp 2013 and Yelp 2014 datasets show that the proposed approach has achieved the state-of-the-art performance.

Integration of a graph-based model indexer in commercial modelling tools

Softeam has over 20 years of experience providing UML-based modelling solutions, such as its Modelio modelling tool, and its Constellation enterprise model management and collaboration environment. Due to the increasing number and size of the models used by Softeam’s clients, Softeam joined the MONDO FP7 EU research project, which worked on solutions for these scalability challenges and produced the Hawk model indexer among other results. This paper presents the technical details and several case studies on the integration of Hawk into Softeam’s toolset. The first case study measured the performance of Hawk’s Modelio support using varying amounts of memory for the Neo4j backend. In another case study, Hawk was integrated into Constellation to provide scalable global querying of model repositories. Finally, the combination of Hawk and the Epsilon Generation Language was compared against Modelio for document generation: for the largest model, Hawk was two orders of magnitude faster.

Digital literacy matters. Increasing workforce productivity through blended English language programmes

Current workplace demands newer forms of literacies that go beyond the ability to decode print. These involve not only competence to operate digital tools, but also the ability to create, represent, and share meaning in different modes and formats; ability to interact, collaborate and communicate effectively using digital tools, and engage critically with technology for developing one’s knowledge, skills, and full participation in civic, economic, and personal matters. This essay examines the application of the ecology of resources (EoR) model for delivering language learning outcomes (in this case, English) through blended classroom environments that use contextually available resources. The author proposes the implementation of the EoR model in blended learning environments to create authentic and sustainable learning environments for skilling courses. Applying the EoR model to Indian skilling instruction contexts, the article discusses how English language and technology literacy can be delivered using contextually available resources through a blended classroom environment. This would facilitate not only acquisition of language and digital literacy outcomes, but also consequent content literacy gain to a certain extent. This would ensure satisfactory achievement of not only communication/language literacy and technological literacy, but also active social participation, lifelong learning, and learner autonomy.

Numeric model to estimate pesticide deposition and losses from aerial spraying.

Pesticides applications have been described by many researches as a very inefficient process. In some cases, there are reports that only 0.02% of the applied products are used for the effective control of the problem. The main factor that influences pesticides applications is the droplet size formed on spraying nozzles. Many parameters affects the dynamic of the droplets, like wind, temperature, relative humidity, and others. Small droplets are biologically more active, but they are affected by evaporation and drift. On the other hand, the great droplets do not promote a good distribution of the product on the target. In this sense, associated with the risk of non target areas contamination and with the high costs involved in applications, the knowledge of the droplet size is of fundamental importance in the application technology. When sophisticated technology for droplets analysis is unavailable, is common the use of artificial targets like water-sensitive paper to sample droplets. On field sampling, water-sensitive papers are placed on the trials where product will be applied. When droplets impinging on it, the yellow surface of this paper will be stained dark blue, making easy their recognition. Collected droplets on this papers have different kinds of sizes. In this sense, the determination of the droplet size distribution gives a mass distribution of the material and so, the efficience of the application of the product. The stains produced by droplets shows a spread factor proportional to their respectives initial sizes. One of methodologies to analyse the droplets is a counting and measure of the droplets made in microscope. The Porton N-G12 graticule, that shows equaly spaces class intervals on geometric progression of square 2, are coulpled to the lens of the microscope. The droplet size parameters frequently used are the Volumetric Median Diameter (VMD) and the Numeric Median Diameter. On VMD value, a representative droplets sample is divided in two equal parts of volume, in such away one part contains droplets of sizes smaller than VMD and the other part contains droplets of sizes greater that VMD. The same process is done to obtaining the NMD, which divide the sample in two equal parts in relation to the droplets size. The ratio between VMD and NMD allows the droplets uniformity evaluation. After that, the graphics of accumulated probability of the volume and size droplets are plotted on log scale paper (accumulated probability versus median diameter of each size class). The graphics provides the NMD on the x-axes point corresponding to the value of 50% founded on the y-axes. All this process is very slow and subjected to operator error. So, in order to decrease the difficulty envolved with droplets measuring it was developed a numeric model, implemented on easy and accessfull computational language, which allows approximate VMD and NMD values, with good precision. The inputs to this model are the frequences of the droplets sizes colected on the water-sensitive paper, observed on the Porton N-G12 graticule fitted on microscope. With these data, the accumulated distribution of the droplet medium volumes and sizes are evaluated. The graphics obtained by plotting this distributions allow to obtain the VMD and NMD using linear interpolation, seen that on the middle of the distributions the shape of the curves are linear. These values are essential to evaluate the uniformity of droplets and to estimate the volume deposited on the observed paper by the density (droplets/cm2). This methodology to estimate the droplets volume was developed by 11.0.94.224 Project of the CNPMA/EMBRAPA. Observed data of herbicides aerial spraying samples, realized by Project on Pelotas/RS county, were used to compare values obtained manual graphic method and with those obtained by model has shown, with great precision, the values of VMD and NMD on each sampled collector, allowing to estimate a quantities of deposited product and, by consequence, the quantities losses by drifty. The graphics of variability of VMD and NMD showed that the quantity of droplets that reachs the collectors had a short dispersion, while the deposited volume shows a great interval of variation, probably because the strong action of air turbulence on the droplets distribution, enfasizing the necessity of a deeper study to verify this influences on drift.

Development of the Simulation Model for the CoSES Laboratory Test Microgrid in Modelica

Evolution of the traditional consumer in a power system to a prosumer has posed many problems in the traditional uni-directional grid. This evolution in the grid model has made it important to study the behaviour of microgrids. This thesis deals with the laboratory microgrid setup at the Munich School of Engineering, built to assist researchers in studying microgrids. The model is built in Dymola which is a tool for the OpenModelica language. Models for the different components were derived, suiting the purpose of this study. The equivalent parameters were derived from data sheets and other simulation programs such as PSCAD. The parameters were entered into the model grid and tested at steady state, firstly. This yielded satisfactory results that were similar to the reference results from MATPOWER power flow. Furthermore, fault conditions at several buses were simulated to observe the behaviour of the grid under these conditions. Recommendations for further developing this model to include more detailed models for components, such as power electronic converters, were made at the end of the thesis.

Bio-QA-GNN: Reasoning with Language Models and Knowledge Graphs for Interpretable Biomedical Question Answering

Nowadays the idea of injecting world or domain-specific structured knowledge into pre-trained language models (PLMs) is becoming an increasingly popular approach for solving problems such as biases, hallucinations, huge architectural sizes, and explainability lack—critical for real-world natural language processing applications in sensitive fields like bioinformatics. One recent work that has garnered much attention in Neuro-symbolic AI is QA-GNN, an end-to-end model for multiple-choice open-domain question answering (MCOQA) tasks via interpretable text-graph reasoning. Unlike previous publications, QA-GNN mutually informs PLMs and graph neural networks (GNNs) on top of relevant facts retrieved from knowledge graphs (KGs). However, taking a more holistic view, existing PLM+KG contributions mainly consider commonsense benchmarks and ignore or shallowly analyze performances on biomedical datasets. This thesis start from a propose of a deep investigation of QA-GNN for biomedicine, comparing existing or brand-new PLMs, KGs, edge-aware GNNs, preprocessing techniques, and initialization strategies. By combining the insights emerged in DISI's research, we introduce Bio-QA-GNN that include a KG. Working with this part has led to an improvement in state-of-the-art of MCOQA model on biomedical/clinical text, largely outperforming the original one (+3.63\% accuracy on MedQA). Our findings also contribute to a better understanding of the explanation degree allowed by joint text-graph reasoning architectures and their effectiveness on different medical subjects and reasoning types. Codes, models, datasets, and demos to reproduce the results are freely available at: \url{https://github.com/disi-unibo-nlp/bio-qagnn}.

Integrated Proteomics Identified Up-regulated Focal Adhesion-mediated Proteins In Human Squamous Cell Carcinoma In An Orthotopic Murine Model.

Understanding the molecular mechanisms of oral carcinogenesis will yield important advances in diagnostics, prognostics, effective treatment, and outcome of oral cancer. Hence, in this study we have investigated the proteomic and peptidomic profiles by combining an orthotopic murine model of oral squamous cell carcinoma (OSCC), mass spectrometry-based proteomics and biological network analysis. Our results indicated the up-regulation of proteins involved in actin cytoskeleton organization and cell-cell junction assembly events and their expression was validated in human OSCC tissues. In addition, the functional relevance of talin-1 in OSCC adhesion, migration and invasion was demonstrated. Taken together, this study identified specific processes deregulated in oral cancer and provided novel refined OSCC-targeting molecules.

Electronic And Structural Study Of Pt-modified Au Vicinal Surfaces: A Model System For Pt-au Catalysts.

Two single crystalline surfaces of Au vicinal to the (111) plane were modified with Pt and studied using scanning tunneling microscopy (STM) and X-ray photoemission spectroscopy (XPS) in ultra-high vacuum environment. The vicinal surfaces studied are Au(332) and Au(887) and different Pt coverage (θPt) were deposited on each surface. From STM images we determine that Pt deposits on both surfaces as nanoislands with heights ranging from 1 ML to 3 ML depending on θPt. On both surfaces the early growth of Pt ad-islands occurs at the lower part of the step edge, with Pt ad-atoms being incorporated into the steps in some cases. XPS results indicate that partial alloying of Pt occurs at the interface at room temperature and at all coverage, as suggested by the negative chemical shift of Pt 4f core line, indicating an upward shift of the d-band center of the alloyed Pt. Also, the existence of a segregated Pt phase especially at higher coverage is detected by XPS. Sample annealing indicates that the temperature rise promotes a further incorporation of Pt atoms into the Au substrate as supported by STM and XPS results. Additionally, the catalytic activity of different PtAu systems reported in the literature for some electrochemical reactions is discussed considering our findings.