With the emergence of the Internet, many innovative and interesting applications from different countries have been stimulated and e-commerce is also getting more and more pervasive. Under this scenario, tremendous amount of information expressed in different languages are exchanged and shared by not only organizations but also individuals in the modern global environment. A large proportion of information is typically formatted and available as textual documents and managed by using categories. Consequently, the development of a practical and effective technique to deal with the problem of cross-lingual category integration (CLCI) becomes a very essential and important issue. Several category integration techniques have been proposed, but all of them deal with category integration involving only monolingual documents. In response, in this study, we combine the existing cross-lingual text categorization techniques with an existing monolingual category integration technique (specifically, Enhanced Naive Bayes) and proposed a CLCI solution to address cross-lingual category integration. Our empirical evaluation results show that our proposed CLCI technique demonstrates its feasibility and superior effectiveness.

In this study, a methodology for separation of chiral molecules, by using enhanced ultrafiltration system was developed. Benzoin was the model chiral molecule studied. In the scope of developing this methodology, some parameters were investigated in the preliminary ultrafiltration experiments in order to set the operation conditions for enhanced ultrafiltration experiments. Due to the slight solubility of benzoin in pure water, 15% (v/v) Polyethylene glycol (PEG 400) and 30 % (v/v) Dimethyl sulfoxide (DMSO) were selected as cosolvents. Because of the high retention capacity of RC-10000 Da membranes for benzoin, a membrane saturation strategy was developed. In polymer enhanced ultrafiltration (PEUF) experiments bovine serum albumin (BSA) was used as ligand. Effects of ligand concentration and pH on total benzoin retention and on enantiomeric excess (ee %) were investigated. Benzoin concentration was almost kept constant at ~10 ppm and ~50 ppm for 15% (v/v) PEG 400 and 30 % (v/v) DMSO cosolvents, respectively. It was observed that the increase either in pH or in BSA concentration yielded an increase in total benzoin retention. In 15% (v/v) PEG 400-water, with BSA concentration of 10000 ppm, at pH 10, total benzoin retention reached to 48.7%. For this cosolvent, at different pH values and at different BSA concentrations, all ee % values were about or less than 10%. When 50000 ppm BSA was dissolved in 30 % (v/v) DMSO-water, total benzoin retention increased to 41.3% at pH 10 and ee % reached 16.7 % at pH 11. In enzyme enhanced ultrafiltration (EEUF) experiments, specific to benzoin, apo form of Benzaldehyde Lyase (BAL, E.C. 4.1.2.38) was used as ligand. These experiments were performed with constant ~ 10 ppm benzoin concentration in only 15% (v/v) PEG 400 &ndash / water solvent. Effect of BAL concentration on total benzoin retention and ee% was investigated. It was found that / for all the studied BAL concentrations in the range of 650- 1936 ppm total benzoin retention and ee % were kept almost constant at ~75% and ~60%, respectively.

The increased level of greenhouse gases due to human activity is the main factor for climate change. CO2 is the main constitute among these gases. Subsurface storage of CO2 in geological systems such as coal reservoirs is considered as one of the promising perspectives. Coal can be safely and effectively utilized to both store CO2 and recover CH4. By injecting CO2 into the coal beds, methane is released with CO2 adsorption in the coal matrix and this process is known as enhanced coal bed methane recovery (ECBM). Zonguldak Coal Basin is one of the Turkey&amp / #8217 / s important coal resources. Since the coal seams in Bartin-Amasra field are found relatively deeper parts of the basin comparing to other places, this basin was not studied detailed enough yet. Bartin-Amasra basin was found convenient for enhanced coalbed methane recovery. The lithologic information taken from the Turkish Hard Coal Enterprise (TTK) was examined and the depths of the coal seams and the locations of the wells were visualized to perform a reliable correlation between seams existed in the area. According to the correlations, 63 continuous coal layers were found. A statistical reserve estimation of each coal layer for methane was made by using Monte Carlo simulation method. Uncertainty is an important parameter in risk analysis, for this reason the results were determined at probabilities of P10, P50 and P90. Enhanced coalbed methane recovery was simulated with CMG-GEM module using Coal Layer #26 which has more initial gas in place. The effects of adsorption, cleat spacing, compressibility, density, permeability, permeability anisotropy, porosity and water saturation parameters were examined in enhanced coalbed methane recovery by the simulation runs. The initial methane in place found in all these coal layers both in free and adsorbed states were estimated using probabilistic calculations resulted in possible reserve (P10) of 72.97 billion scf, probable reserve (P50) of 47.74 billion scf and proven reserves (P90) of 30.46 billion scf. Since the Amasra coal reservoir is not saturated with water, almost 10% of the total gas in place was found to be in the cleats as free gas. Coal layer #26 has an area of 4099 acres, average thickness of 6.23 ft and depth of 545 m (Karadon formation). P50 reserve estimation was 6.47 billion scf in matrix and 0.645 billion scf in fracture. Although the decrease in cleat porosity was less when shrinkage and swelling effects included, the decrease in cleat permeability as a function of porosity diminished the methane production. Cumulative methane production was enhanced with the injection of carbon dioxide (ECBM) approximately 23% than that of CBM recovery. Although closing the wells to production because of CO2 breakthrough had a negative effect on methane production initially, there was no difference between ultimate methane productions whether the wells remained open or closed, but more carbon dioxide was sequestered when the production ceased at the wells. Injected carbon dioxide amount of 5192 tonnes/year in base case was only capable to sequester only 0.3% of the yearly carbon dioxide emission of Zonguldak &Ccedil / atalagzi Power Plant nearby. Considering the gas in place capacity of the coal layer #26 as 15% of the resource area-A, it can be said that the project aiming ECBM recovery rather than carbon dioxide sequestration would be successful. In spite of water saturated coal reservoirs where the water production is required initially, it can be possible to start immediately the injection of CO2 with methane production for a dry coal reservoir. Cleat permeability being one of the most crucial parameter in the coal reservoir affected the rate of methane production. The more free gas was found in higher porosity cleat systems. Although the cumulative methane production was increased when the cleat porosity rose, methane recovery percentages were remained almost constant. The lower the cleat spacing the higher the rate of transfer between fracture and matrix was observed. The rate of gas desorption from the coal matrix and subsequent diffusion to both butt and face cleats was higher than the rate of flow in the face cleats, then production was flow-limited, pressure-driven and was defined by Darcy&amp / #8217 / s Law. The cumulative CH4 production was higher when the coal was denser. The change in coal compressibility affected slightly the cleat porosity and therefore the cleat permeability due to the change in reservoir pressure. Langmuir volume is defined as maximum adsorption capacity. Kozlu formation (deeper than Karadon formation) having lower Langmuir volume resulted in higher ultimate recovery because of lower Langmuir pressure than that of Karadon formation. In base case (Karadon formation), although the higher Langmuir volume was used, less methane production was observed. Permeability anisotropy generated the CO2-CH4 front in elliptic shape.

Nanomaterials are widely investigated by researches and because of their unique properties they have been utilized in many different devices. Nanowires are one of these materials which show deviated mechanical, chemical, physical and optical, properties from their bulk counterparts. These deviations in properties of the nanowires are based on both their high surface to volume ratio and quantum confinement effect. Lately optical properties of nanowires have received great attention as they also exhibit good light sensitivity. Germanium is a semiconductor, which has been used widely as an active material in infrared light detectors. Due to excellent light detection of germanium its nanostructures have also been widely studied in optoelectronic devices. Germanium nanowires have been used in many devices such as field effect transistors, diodes, field emitters and photodetectors. Synthesis of high quality and high aspect ratio germanium nanowires could make important contributions to these devices. There are several synthesis methods for germanium nanowires. These are electrochemical etching, solvothermal, supercritical v fluidic, laser ablation, chemical vapor deposition and vapor transport methods. Among these methods, high quality, single crystalline, defect free germanium nanowires using accessible solid powder precursors could be synthesized with vapor transport method. In the first part of this thesis, germanium nanowire growth with vapor transport method is investigated. One of the most advantageous features of this method is using solid powder precursors instead of toxic gases. Until now, three different kinds of solid germanium precursors have been reported in vapor transport method, all of them are investigated and the resulting nanowires are compared in this thesis. Vapor transport method enables high control over the morphology of the nanowires. The most important parameters which affect the morphology of the nanowires are temperature, pressure and precursor type. Therefore, a detailed parametric study is provided based on these parameters and their effect on the final diameter of the nanowires is determined. The as &ndash / synthesized nanowires contain a very thick oxide layer on their surface. Therefore, oxide removal with acid etching is also investigated in this thesis. In the second part of this thesis, utilization of the germanium nanowire networks in fully transparent, flexible and network enhanced photodetectors is investigated. In order to obtain a germanium nanowire network, the as-synthesized nanowires are transferred from growth substrate to the device substrate by sonication and vacuum filtration. Silver nanowires and single walled carbon nanotubes are used as fully transparent electrodes. Both rigid and flexible photodetectors are fabricated and their current-voltage characteristics and photoresponse behaviors with different germanium nanowire densities are determined.

"Enhanced index fund" is an investment strategy, combining active and passive management elements, for index tracking and return enhancing through disciplined market timing, stock selection and leverage activities. Though enhanced index funds have been well developed globally, there is only one enhanced index fund in Taiwan - "Polaris/P-Shares Taiwan Dividend+ ETF". Taiwan's stock market falls between weak form and semi-strong form efficiency. With the growth of Taiwan's mutual fund industry size, the enhanced index funds have very good chance to become the main investment instruments of institutional investors, index investors, and pensions. This study attempts to build enhanced index funds, then analyzes the performance and checks the feasibility of launching such products in Taiwan. In this study, we select "TSEC Taiwan 50 index (TW50)" as the benchmark index. The innovative "Multi-section Adjustment Model" divides the original weights of constituent stocks into two sections. Each section is adjusted through parameters. The "multi-factor model section" is responsible for the delivery of enhanced return, while the "cash dividend yield section" is used to provide excess cash dividend yield. The investment target is set for less than 1.5 percent tracking error, at least 1 percent tracking difference, and higher cash dividend yield than the benchmark. Building methodology can be divided into "fixed parameter model" and "floating parameters model" according to its update frequency. Empirical studies show that: (1) The enhanced index fund built from the "fixed parameter model" not only exhibits risk slightly lower than the benchmark, but also enjoys higher return. (2) In the short-term, the performance of the enhanced index fund built from "floating parameters model" is difficult to predict; in the long-term, however, the risk is lower and the return is higher than TW50. The cumulative return from the "fixed parameter model" is higher than the "floating parameters model" by about 2 percent. (3) The effectiveness of the parameters used to control the optimal weight distribution is decreasing over time, so it is necessary to update parameters regularly. (4) Raising "enhancement multiplier" will cause higher tracking error, but also bring higher tracking difference. This result proves that "multi-factor model section" works nicely and has its contribution. (5) As the "section allotment" and/or "fixed rate" getting lower, there will be more and more weights distributed to the cash dividend yield level, resulting in higher cash dividend yield. It means the "cash dividend yield section" has its merit as well. (6) Regular parameter updates to the "floating parameters model" helps to reduce the tracking error and, at the same time, maintain positive tracking difference. Considering the perpetual life of real world funds, "floating parameters model" should be a better building methodology. "Multi-Section Adjustment Model" has following advantages: (1) Its concept is intuitive and easy to use. (2) Sections can be customized based on investment objectives. (3) It is easy to analyze the impacts and trade-off among the parameters.

The main objective of this study was to evaluate the feasibility of using nanoscale zero-valent iron (nZVI) as the source of hydrogen to enhance in situ anaerobic biodegradation of trichloroethylene (TCE). In the first part of this study, microcosms were constructed to evaluate the effects of different controlling factors [e.g., different redox conditions (aerobic and anaerobic conditions), different microorganisms (in situ microorganisms, activated sludge, and anaerobic sludge), and different sources of substrates and electron donors (phenol, cane molasses, hydrogen, and nZVI)] on TCE biodegradation. In the second part of this study, batch experiments were conducted to evaluate the feasibility of hydrogen production by nZVI and bimetallic particles. Results from the microcosm study indicate that in-situ microorganisms were capable of degrading TCE under aerobic and anaerobic conditions. Results also show that TCE removal was more effective by activated sludge and anaerobic sludge. Aerobic biodegradation of TCE was enhanced by the addition of phenol and cane molasses. Under anaerobic conditions, TCE removal could be improved when cane molasses and hydrogen were supplied. In addition, anaerobic TCE degradation was more effective with the presence of hydrogen. Results of microcosms conducted with the addition of nZVI reveal that TCE was degraded completely in both live and autoclaved microcosms. This indicates that chemical reductive dechlorination seemed to dominate the removal of TCE in microcosms. Therefore, further studies with higher TCE concentrations or lower nZVI doses need to be conducted to determine the effects of the produced hydrogen on TCE biodegradation. Results from the hydrogen production experiments indicate that efficiency of hydrogen production by nZVI ranged from 30% to 76%. Higher dose of nZVI addition resulted in higher amount of hydrogen production. The total amounts of hydrogen production were correlated with the doses of nZVI. In addition, rates and efficiency of hydrogen production by bimetallic particles were better than those of nZVI. Results of the batch experiments reveal that nZVI and bimetallic particles had good efficiency on hydrogen production. This indicates that nZVI and bimetallic particles have high potential to be used as hydrogen producers. In this study, a simple system consisted of only water and nZVI or bimetallic particles was applied to produce hydrogen. Although TCE in microcosms with nZVI addition was totally consumed by nZVI, results of microcosms with hydrogen addition demonstrated that hydrogen was able to improve the efficiency of anaerobic TCE biodegradation. Thus, it may be feasible to use nZVI as the source of hydrogen to enhance in situ anaerobic biodegradation of TCE. The advantages of using nZVI as the source of hydrogen include: (1) rapid removal of significant contaminant concentrations in the early stage of nZVI injection; (2) creation of a more reducing environment; (3) safer than liquid hydrogen, which is stored in steel containers; and (4) direct hydrogen supply without transfer of biological mechanisms compared to commercial hydrogen release compounds and other organic substrates. Results of this study suggest that biological reductive dechlorination of TCE can be enhanced if proper doses of nZVI are supplied in situ. Knowledge and comprehension obtained in this study will be helpful in designing an enhanced in situ anaerobic bioremediation system for a TCE-contaminated site.

Many enhanced index funds are based on a quantitative model to control active risk and to acquire active return. In this thesis we first construct a multiple-factor model (MFM) and then use statistical methods to evaluate the significance and stability of factor explanatory power. Significant and stable factors are utilized to fine tune weights of T50 index fund portfolio by an intuitive weight allocation model to achieve the effect of return enhancement. Empirical studies show that the multiple-factor model can explain the excess stock return effectively; the average R-Square of multiple-factor model reaches 49%. After analyzing the sensitivity of parameter of enhanced index weight allocation, the study finds that the original weight retention rate has linear relationship with active return and active risk of the T50 index fund. Adjusting the retention rate allows us to control the active return and active risk of T50 index fund. Furthermore, adjusting the original weight retention rate according to the Adj-R2 of multiple-risk factor model can effectively improve the stability of active return. The study finds also that the expected rates of return which are calculated by multiple-risk factor model could not differentiate among future performance of the first your guarantee portfolios. Thus, the study adjusts the range of weight allocation to T50 constituent stocks with higher and lower expected return rates. The result shows that this adjustment increased the IR of the enhanced index funds.

The work presented in this dissertation suggests novel design of chemical plasmonic sensors which have been developed based on Localized Surface Plasmon Resonance (LSPR), and Surface-enhanced Raman scattering (SERS) phenomena. The goal of the study is to understand the SERS phenomena for 3D hybrid (organic/inorganic) templates and to design of the templates for trace-level detection of selected chemical analytes relevant to liquid explosives and hazardous chemicals. The key design criteria for the development of the SERS templates are utilizing selective polymeric nanocoatings within cylindrical nanopores for promoting selective adsorption of chemical analyte molecules, maximizing specific surface area, and optimizing concentration of hot spots with efficient light interaction inside nanochannels. The organic/inorganic hybrid templates are optimized through a comprehensive understanding of the LSPR properties of the gold nanoparticles, gold nanorods, interaction of light with highly porous alumina template, and the choice of physical and chemical attributes of the selective coating. Furthermore, novel method to assemble silver nanoparticles in 3D as the active SERS-active substrate has been demonstrated by uniform, in situ growth of silver nanoparticles from electroless deposited silver seeds excluding any adhesive polymer layer on template. This approach can be the optimal for SERS sensing applications because it is not necessary to separate the Raman bands of the polyelectrolyte binding layer from those of the desired analyte. The fabrication method is an efficient, simple and fast way to assemble nanoparticles into 3D nanostructures. Addressable Raman markers from silver nanowire crossbars with silver nanoparticles are also introduced and studied. Assembly of silver nanowire crossbar structure is achieved by simple, double-step capillary transfer lithography. The on/off SERS properties can be observed on silver nanowire crossbars with silver nanoparticles depending on the exact location and orientation of decorated silver nanoparticles nearby silver nanowire crossbars. As an alternative approach for the template-assisted nanostructure design, porous alumina membrane (PAM) can be utilized as a sacrificial template for the fabrication of the nanotube structure. The study seeks to investigate the design aspects of polymeric/inorganic hybrid nanotube structures with plasmonic properties, which can be dynamically tuned by external stimuli such as pH. This research suggests several different organic/inorganic nanostructure assemblies by various template-assisted techniques. The polymeric/inorganic hybrid nanostructures including SERS property, pH responsive characteristics, and large surface area will enable us to understand and design the novel chemical plasmonic sensors.

<p>Learning materials have multiple forms, such as books, overhead slides, computer files, blackboard notes by teachers, narration to the notes, video/audio tapes etc. Since the forms are highly inhomogeneous, it becomes difficult to collect and practically use them by a particular learner for individual study at home. Such multiple media are also expensive in management, since human resources are needed to keep the material repositories in order. One solution of the problem lies in centralized active digital repositories. Such repositories aim to simplify the learner’s work and boost learning efficiency. With introduction of interactivity and live communication tools such repositories become learning platforms exceeding the functionality of “passive” digital libraries. Such learning platforms could be easily used both for on-campus and distance education.</p><p>This dissertation presents an evaluation of a digital repository of interactive multimedia content in the field of Heat and Power Technology: Computerized Educational Platform (CompEdu HPT). The platform evaluation consisted of integration of the tool into the university curriculum and then collection of feedback from students and teachers. The evaluation concerned usefulness of the platform for learning, aspects of instruction improvement, collecting observations about how the platform is used by students, as well as their opinions about the IT application direction chosen. The methods included: online feedback forms, questionnaires, interviews, discussions and observations.</p><p>The evaluation demonstrated that the main strength of the platform is the integration of learning materials in one portable package. The students appreciated structured and logically arranged information that was available for easy access. Coverage of a broad area of knowledge related to heat and power technology was also pointed out as an advantage with reflection on the very low price of acquisition of the materials. The most popular elements of the content in use included: simulations, lecture notes, the print function, the glossary, and calculation exercises. A major part of the students declared the high value of CompEdu in facilitating home study. Nevertheless, not all the students had a positive impression: around one-fifth of them did not find the platform useful and expressed preference for more traditional learning media. The majority of the negative opinions concerned content quality, which directly related to weaknesses of the content production and review process.</p><p>The evaluation emphasized the importance of material quality and amount as the key issue for a good learning platform with relatively smaller importance of presentation forms. The evaluation also considered aspects of functionality from the user point of view. Differentiation between popularity of simulations showed that simulations used by teachers during lectures have higher educational value than those for individual use only. The popularity of the printing option indicated a need for adaptation of digital materials for paper publishing. The general conclusion for practical use of multimedia tools in education was that high usability and simplicity of information access should be the focus point of any chosen approach in the direction.</p><p>The CompEdu evaluation suggested that after thorough content review and addition of an efficient search mechanism the platform can successfully deliver rich learning content. The platform gave an extensive real-case illustration of how multimedia can be used in educational practice. Due to the evaluation, the CompEdu e-learning group has collected rich experience and know-how in the field of active knowledge repositories. The experience will be used for development of a more sophisticated learning platform working in the global Internet environment with major focus on information accessibility by easy search.</p>

<p>The existence of trihalomethanes (THMs) in a drinking water plant of Nicaragua has been investigated in order to see whether the concentration exceeded the maximum contaminant level recommended by the environmental protection agency of the United States (USEPA) and the Nicaragua guidelines. The influence of pH, temperature, chlorine dose and contact time on the formation of THMs were studied. The contents of organic matter measured by surrogate parameters such as total organic carbon, dissolved organic carbon, ultraviolet absorbance and specific ultraviolet absorbance were also determined in order to show which type of organic matter is most reactive with chlorine to form THMs. Models developed by other researchers to predict the formation of trihalomethanes were tested to see whether they can be used to estimate the trihalomethane concentration. In addition, empirical models were development to predict the THM concentration of the drinking water plant analysed. The raw water was treated by conventional and enhanced coagulation and these processes were compared with regard to the removal of natural organic matter (NOM). The significance of the results was assessed using statistic procedures.</p><p>The average concentration of THMs found at the facility is below the USEPA and Nicaragua guideline values. Nevertheless the maximum contaminant level set by USEPA is sometimes exceeded in the rainy season when the raw water is rich in humic substances. Comparison between the water treated by conventional and enhanced coagulation shows that enhanced coagulation considerably diminished the trihalomethane formation and the value after enhanced coagulation never exceeded the guidelines. This is because enhanced coagulation considerably decreases the organic matter due to the high coagulant dose applied. The study of the trihalomethane formation when varying pH, time, temperature and chlorine dose using water treated by conventional and enhanced coagulation showed that higher doses of chlorine, higher pH, higher temperature and a longer time increases the formation of THMs. However, combinations of two and three factors are the opposite. The predicted THM formation equations cannot be used for the water at this facility, since the results shown that the measured THM differs significantly from the THM concentration predicted. Two empirical models were developed from the data for enhanced coagulation, using linear and non-linear regression. These models were tested using the database obtained with conventional coagulation. The non-linear model was shown to be able to predict the formation of THMs in the Boaco drinking water plant.</p>