Learning user and product distributed representations using a sequence model for sentiment analysis

In product reviews, it is observed that the distribution of polarity ratings over reviews written by different users or evaluated based on different products are often skewed in the real world. As such, incorporating user and product information would be helpful for the task of sentiment classification of reviews. However, existing approaches ignored the temporal nature of reviews posted by the same user or evaluated on the same product. We argue that the temporal relations of reviews might be potentially useful for learning user and product embedding and thus propose employing a sequence model to embed these temporal relations into user and product representations so as to improve the performance of document-level sentiment analysis. Specifically, we first learn a distributed representation of each review by a one-dimensional convolutional neural network. Then, taking these representations as pretrained vectors, we use a recurrent neural network with gated recurrent units to learn distributed representations of users and products. Finally, we feed the user, product and review representations into a machine learning classifier for sentiment classification. Our approach has been evaluated on three large-scale review datasets from the IMDB and Yelp. Experimental results show that: (1) sequence modeling for the purposes of distributed user and product representation learning can improve the performance of document-level sentiment classification; (2) the proposed approach achieves state-of-The-Art results on these benchmark datasets.

Multipurpose portable system for environmental, industrial, and medical applications

This dissertation is about the research carried on developing an MPS (Multipurpose Portable System) which consists of an instrument and many accessories. The instrument is portable, hand-held, and rechargeable battery operated, and it measures temperature, absorbance, and concentration of samples by using optical principles. The system also performs auxiliary functions like incubation and mixing. This system can be used in environmental, industrial, and medical applications. ^ Research emphasis is on system modularity, easy configuration, accuracy of measurements, power management schemes, reliability, low cost, computer interface, and networking. The instrument can send the data to a computer for data analysis and presentation, or to a printer. ^ This dissertation includes the presentation of a full working system. This involved integration of hardware and firmware for the micro-controller in assembly language, software in C and other application modules. ^ The instrument contains the Optics, Transimpedance Amplifiers, Voltage-to-Frequency Converters, LCD display, Lamp Driver, Battery Charger, Battery Manager, Timer, Interface Port, and Micro-controller. ^ The accessories are a Printer, Data Acquisition Adapter (to transfer the measurements to a computer via the Printer Port and expand the Analog/Digital conversion capability), Car Plug Adapter, and AC Transformer. This system has been fully evaluated for fault tolerance and the schemes will also be presented. ^

What Google Maps can do for Biomedical Data Dissemination: Examples and a Design Study

Background: Biologists often need to assess whether unfamiliar datasets warrant the time investment required for more detailed exploration. Basing such assessments on brief descriptions provided by data publishers is unwieldy for large datasets that contain insights dependent on specific scientific questions. Alternatively, using complex software systems for a preliminary analysis may be deemed as too time consuming in itself, especially for unfamiliar data types and formats. This may lead to wasted analysis time and discarding of potentially useful data. Results: We present an exploration of design opportunities that the Google Maps interface offers to biomedical data visualization. In particular, we focus on synergies between visualization techniques and Google Maps that facilitate the development of biological visualizations which have both low-overhead and sufficient expressivity to support the exploration of data at multiple scales. The methods we explore rely on displaying pre-rendered visualizations of biological data in browsers, with sparse yet powerful interactions, by using the Google Maps API. We structure our discussion around five visualizations: a gene co-regulation visualization, a heatmap viewer, a genome browser, a protein interaction network, and a planar visualization of white matter in the brain. Feedback from collaborative work with domain experts suggests that our Google Maps visualizations offer multiple, scale-dependent perspectives and can be particularly helpful for unfamiliar datasets due to their accessibility. We also find that users, particularly those less experienced with computer use, are attracted by the familiarity of the Google Maps API. Our five implementations introduce design elements that can benefit visualization developers. Conclusions: We describe a low-overhead approach that lets biologists access readily analyzed views of unfamiliar scientific datasets. We rely on pre-computed visualizations prepared by data experts, accompanied by sparse and intuitive interactions, and distributed via the familiar Google Maps framework. Our contributions are an evaluation demonstrating the validity and opportunities of this approach, a set of design guidelines benefiting those wanting to create such visualizations, and five concrete example visualizations.

Toxicity assessment of industrial- and sunscreen-derived ZnO nanoparticles

Lo scopo della presente tesi è l’analisi della tossicità di nanoparticelle di ossido di zinco (nano-ZnO) verso gli organismi acquatici. In particolare, il presente studio valuta per la prima volta l'inibizione della crescita della diatomea Thalassiosira pseudonana indotta sia da nanoparticelle di dervazione industriale, che da nanoparticelle auto-estratte in laboratorio da un filtro solare. Gli esperimenti, condotti presso il Laboratorio di Ingegneria dell'Università di Miami, hanno mostrato che la tossicità indotta dalle nanoparticelle di ossido di zinco è influenzata dal tipo di nanoparticelle, nonché dalla loro concentrazione nella soluzione acquosa e dal tempo di esposizione. In particolare le nanoparticelle di derivazione industriale, più piccole rispetto alle nanoparticelle estratte dal filtro solare, hanno indotto un’inibizione della crescita superiore, specialmente a concentrazioni inferiori. Questo andamento suggerisce che ad alte concentrazioni la tossicità di nano-ZnO potrebbe essere influenzata dall’aggregazione di nanoparticelle (indipendentemente dalle dimensioni di partenza delle nanoparticelle), mentre a concentrazioni inferiori la tossicità potrebbe essere influenzata dalle dimensioni di partenza delle nanoparticelle, così come dal tipo di nanoparticelle e dal tempo di esposizione.

Protein and Drug Design Algorithms Using Improved Biophysical Modeling

This thesis focuses on the development of algorithms that will allow protein design calculations to incorporate more realistic modeling assumptions. Protein design algorithms search large sequence spaces for protein sequences that are biologically and medically useful. Better modeling could improve the chance of success in designs and expand the range of problems to which these algorithms are applied. I have developed algorithms to improve modeling of backbone flexibility (DEEPer) and of more extensive continuous flexibility in general (EPIC and LUTE). I’ve also developed algorithms to perform multistate designs, which account for effects like specificity, with provable guarantees of accuracy (COMETS), and to accommodate a wider range of energy functions in design (EPIC and LUTE).

Physical Insights, Steady Aerodynamic Effects, and a Design Tool for Low-Pressure Turbine Flutter

The successful, efficient, and safe turbine design requires a thorough understanding of the underlying physical phenomena. This research investigates the physical understanding and parameters highly correlated to flutter, an aeroelastic instability prevalent among low pressure turbine (LPT) blades in both aircraft engines and power turbines. The modern way of determining whether a certain cascade of LPT blades is susceptible to flutter is through time-expensive computational fluid dynamics (CFD) codes. These codes converge to solution satisfying the Eulerian conservation equations subject to the boundary conditions of a nodal domain consisting fluid and solid wall particles. Most detailed CFD codes are accompanied by cryptic turbulence models, meticulous grid constructions, and elegant boundary condition enforcements all with one goal in mind: determine the sign (and therefore stability) of the aerodynamic damping. The main question being asked by the aeroelastician, ``is it positive or negative?'' This type of thought-process eventually gives rise to a black-box effect, leaving physical understanding behind. Therefore, the first part of this research aims to understand and reveal the physics behind LPT flutter in addition to several related topics including acoustic resonance effects. A percentage of this initial numerical investigation is completed using an influence coefficient approach to study the variation the work-per-cycle contributions of neighboring cascade blades to a reference airfoil. The second part of this research introduces new discoveries regarding the relationship between steady aerodynamic loading and negative aerodynamic damping. Using validated CFD codes as computational wind tunnels, a multitude of low-pressure turbine flutter parameters, such as reduced frequency, mode shape, and interblade phase angle, will be scrutinized across various airfoil geometries and steady operating conditions to reach new design guidelines regarding the influence of steady aerodynamic loading and LPT flutter. Many pressing topics influencing LPT flutter including shocks, their nonlinearity, and three-dimensionality are also addressed along the way. The work is concluded by introducing a useful preliminary design tool that can estimate within seconds the entire aerodynamic damping versus nodal diameter curve for a given three-dimensional cascade.

Characterisation of atmospheric single particles in industrial and regional background environments using aerosol time of flight mass spectrometry

The composition of atmospheric particles is an important factor in determining their impact on climate and health. In this study, an aerosol time-of-flight mass spectrometer (ATOFMS) was used to measure the chemical composition of ambient single particles at two contrasting locations – an industrial site in Dunkirk, France and a regional background site in Corsica. The ATOFMS data were combined with meteorological information and other particle measurements to determine the various sources of the particles observed at the sites. The particle classes detected in Dunkirk included carbonaceous species from fossil fuel combustion and biomass burning, metal-containing types from local industries and seasalt. Highest particle number concentrations and mass concentrations of PM2.5, black carbon, organics, nitrate, ammonium and several metallic species (Fe, Mn, Pb, Zn) were found during periods heavily influenced by local industry. Particles from a ferromanganese alloy manufacturing facility were identified by comparing ambient ATOFMS data with single particle mass spectra from industrial chimney filters and ores. Particles from a steelworks were identified based on comparison of the ambient data with previous studies. Based on these comparisons, the steelworks was identified as the dominant emitter of Fe-rich particles, while the ferromanganese alloy facility emitted Mn-rich particles. In Corsica, regional transport of carbonaceous particles from biomass burning and fossil fuel combustion was identified as the major source of particles in the Mediterranean background aerosol. Throughout the campaign the site was influenced by air masses altering the composition of particles detected. During North Atlantic air masses the site was heavily influenced by fresh sea salt. Regional stagnation was the most common type of air mass regime throughout the campaign and resulted in the accumulation of carbonaceous particles during certain periods. Mass concentrations were estimated for ATOFMS particle classes, and good agreement was found between the major carbonaceous classes and other quantitative measurements. Overall the results of this work serve to highlight the excellent ability of the ATOFMS technique in providing source-specific composition and mixing state information on atmospheric particles at high time resolution.

Metropolitan Governance and Institutional Design: Transportation in the Jakarta Metropolitan Region

In the Jakarta Metropolitan Region (JMR), the lack of co-ordination and appropriate governance has resulted in paralyzing traffic jams at the metropolitan scale that cannot be resolved by a single government entity. The issue of metropolitan governance is especially crucial here as the JMR lacks an established and formally pre-designed system of governance (e.g., in a constitution or other legal regulations). Instead, it relies on the interaction, coordination and cooperation of a multitude of different stakeholders, ranging from local and regional authorities to private entities and citizens. This chapter offers a discussion on the various governance approaches relating to an appropriate institutional design required for transportation issues at the metropolitan scale. The case used is a regional Bus Rapid Transit (BRT) system as an extension to the metropolitan transport system. Institutional design analysis is applied to the case and three possible improvements - i) a ‘Megapolitan’ concept, ii) a regional spatial plan and iii) inter-local government cooperation; were identified that correspond to current debates on metropolitan governance approaches of regionalism, localism and new regionalism. The findings, which are relevant to similar metropolitan regions, suggest that i) improvements at the meso-level of institutional design are more readily accepted and effective than improvements at the macro-level and ii) that the appropriate institutional design for governing metropolitan transportation in the JMR requires enhanced coordination and cooperation amongst four important actors - local governments, the regional agency, the central government, and private companies.

Effect of web holes on web crippling strength of cold-formed steel channel sections under end-one-flange loading condition - Part II: Parametric study and proposed design equations

A parametric study of cold-formed steel sections with web openings subjected to web crippling under end-one-flange (EOF) loading condition is undertaken, using finite element analysis, to investigate the effects of web holes and cross-section sizes. The holes are located either centred above the bearing plates or with a horizontal clear distance to the near edge of the bearing plates. It was demonstrated that the main factors influencing the web crippling strength are the ratio of the hole depth to the depth of the web, the ratio of the length of bearing plates to the flat depth of the web and the location of the holes as defined by the distance of the hole from the edge of the bearing plate divided by the flat depth of web. In this study, design recommendations in the form of web crippling strength reduction factor equations are proposed, which are conservative when compared with the experimental and finite element results.

Improving photofermentative hydrogen production through metabolic engineering and DOE (Design of Experiments)

A l’heure actuelle, les biocarburants renouvelables et qui ne nuit pas à l'environnement sont à l'étude intensive en raison de l'augmentation des problèmes de santé et de la diminution des combustibles fossiles. H2 est l'un des candidats les plus prometteurs en raison de ses caractéristiques uniques, telles que la densité d'énergie élevée et la génération faible ou inexistante de polluants. Une façon attrayante pour produire la H2 est par les bactéries photosynthétiques qui peuvent capter l'énergie lumineuse pour actionner la production H2 avec leur système de nitrogénase. L'objectif principal de cette étude était d'améliorer le rendement de H2 des bactéries photosynthétiques pourpres non sulfureuses utilisant une combinaison de génie métabolique et le plan des expériences. Une hypothèse est que le rendement en H2 pourrait être améliorée par la redirection de flux de cycle du Calvin-Benson-Bassham envers du système de nitrogénase qui catalyse la réduction des protons en H2. Ainsi, un PRK, phosphoribulose kinase, mutant « knock-out » de Rhodobacter capsulatus JP91 a été créé. L’analyse de la croissance sur des différentes sources de carbone a montré que ce mutant ne peut croître qu’avec l’acétate, sans toutefois produire d' H2. Un mutant spontané, YL1, a été récupéré qui a retenu l'cbbP (codant pour PRK) mutation d'origine, mais qui avait acquis la capacité de se développer sur le glucose et produire H2. Une étude de la production H2 sous différents niveaux d'éclairage a montré que le rendement d’YL1 était de 20-40% supérieure à la souche type sauvage JP91. Cependant, il n'y avait pas d'amélioration notable du taux de production de H2. Une étude cinétique a montré que la croissance et la production d'hydrogène sont fortement liées avec des électrons à partir du glucose principalement dirigés vers la production de H2 et la formation de la biomasse. Sous des intensités lumineuses faibles à intermédiaires, la production d'acides organiques est importante, ce qui suggère une nouvelle amélioration additionnel du rendement H2 pourrait être possible grâce à l'optimisation des processus. Dans une série d'expériences associées, un autre mutant spontané, YL2, qui a un phénotype similaire à YL1, a été testé pour la croissance dans un milieu contenant de l'ammonium. Les résultats ont montré que YL2 ne peut croître que avec de l'acétate comme source de carbone, encore une fois, sans produire de H2. Une incubation prolongée dans les milieux qui ne supportent pas la croissance de YL2 a permis l'isolement de deux mutants spontanés secondaires intéressants, YL3 et YL4. L'analyse par empreint du pied Western a montré que les deux souches ont, dans une gamme de concentrations d'ammonium, l'expression constitutive de la nitrogénase. Les génomes d’YL2, YL3 et YL4 ont été séquencés afin de trouver les mutations responsables de ce phénomène. Fait intéressant, les mutations de nifA1 et nifA2 ont été trouvés dans les deux YL3 et YL4. Il est probable qu'un changement conformationnel de NifA modifie l'interaction protéine-protéine entre NifA et PII protéines (telles que GlnB ou GlnK), lui permettant d'échapper à la régulation par l'ammonium, et donc d'être capable d'activer la transcription de la nitrogénase en présence d'ammonium. On ignore comment le nitrogénase synthétisé est capable de maintenir son activité parce qu’en théorie, il devrait également être soumis à une régulation post-traductionnelle par ammonium. Une autre preuve pourrait être obtenue par l'étude du transcriptome d’YL3 et YL4. Une première étude sur la production d’ H2 par YL3 et YL4 ont montré qu'ils sont capables d’une beaucoup plus grande production d'hydrogène que JP91 en milieu d'ammonium, qui ouvre la porte pour les études futures avec ces souches en utilisant des déchets contenant de l'ammonium en tant que substrats. Enfin, le reformage biologique de l'éthanol à H2 avec la bactérie photosynthétique, Rhodopseudomonas palustris CGA009 a été examiné. La production d'éthanol avec fermentation utilisant des ressources renouvelables microbiennes a été traitée comme une technique mature. Cependant, la plupart des études du reformage de l'éthanol à H2 se sont concentrés sur le reformage chimique à la vapeur, ce qui nécessite généralement une haute charge énergetique et résultats dans les émissions de gaz toxiques. Ainsi le reformage biologique de l'éthanol à H2 avec des bactéries photosynthétiques, qui peuvent capturer la lumière pour répondre aux besoins énergétiques de cette réaction, semble d’être plus prometteuse. Une étude précédente a démontré la production d'hydrogène à partir d'éthanol, toutefois, le rendement ou la durée de cette réaction n'a pas été examiné. Une analyse RSM (méthode de surface de réponse) a été réalisée dans laquelle les concentrations de trois facteurs principaux, l'intensité lumineuse, de l'éthanol et du glutamate ont été variés. Nos résultats ont montré que près de 2 moles de H2 peuvent être obtenus à partir d'une mole d'éthanol, 33% de ce qui est théoriquement possible.

Dynamic Simulation Methods for Evaluating Motor Vehicle and Roadway Design and Resolving Policy Issues, 1990

This study had three objectives: (1) to develop a comprehensive truck simulation that executes rapidly, has a modular program construction to allow variation of vehicle characteristics, and is able to realistically predict vehicle motion and the tire-road surface interaction forces; (2) to develop a model of doweled portland cement concrete pavement that can be used to determine slab deflection and stress at predetermined nodes, and that allows for the variation of traditional thickness design factors; and (3) to implement these two models on a work station with suitable menu driven modules so that both existing and proposed pavements can be evaluated with respect to design life, given specific characteristics of the heavy vehicles that will be using the facility. This report summarizes the work that has been performed during the first year of the study. Briefly, the following has been accomplished: A two dimensional model of a typical 3-S2 tractor-trailer combination was created. A finite element structural analysis program, ANSYS, was used to model the pavement. Computer runs have been performed varying the parameters defining both vehicle and road elements. The resulting time specific displacements for each node are plotted, and the displacement basin is generated for defined vehicles. Relative damage to the pavement can then be estimated. A damage function resulting from load replications must be assumed that will be reflected by further pavement deterioration. Comparison with actual damage on Interstate 80 will eventually allow verification of these procedures.

Review of Geometric Design Criteria and Lighting Design Criteria, 1968

The investigations for this report were initiated in October, 1967, to perform the following: l. Review the current Iowa State Highway Commission roadway geometric design standards and criteria for conformance with national policies and recent research findings with special attention to high way safety. 2. Review the current Iowa State Highway Commission roadway lighting design standards and criteria for conformance with national policies and recent research findings with special attention to high way safety

Categories, Allegories, and Circuit Design

Languages based upon binary relations offer an appealing setting for constructing programs from specifications. For example, working with relations rather than functions allows specifications to be more abstract (for example, many programs have a natural specification using the converse operator on relations), and affords a natural treatment of non-determinism in specifications. In this paper we present a novel pictorial interpretation of relational terms as simple pictures of circuits, and a soundness/completeness result that allows relational equations to be proved by pictorial reasoning.