34254 research outputs found
Sort by
Key Silicon Photonics Component Development for High-Speed Optical Interconnect Applications
No full text\ue3\ue3This thesis aims to develop key silicon photonics devices for high-speed optical interconnects applications. The devices developed include optical mode (de)multiplexer, vertical-junction silicon ring modulator, and broadband silicon ring modulator.
\ue3\ue3Through mode division multiplexing (MDM), two orthogonal light of TE0 and TE1 modes with the same wavelength are allowed to propagate in a multimode optical waveguide simultaneously, thus doubles the data transmission capacity. MDM is another approach to increase data throughput besides wavelength division multiplexing. We experimentally demonstrate a TE0/TE1 mode (de)multiplexer to achieve low-loss (0.8509 dB for TE0 and 1.0431 dB for TE1) and low-channel crosstalk (-22 dB for TE0 and -24 dB for TE1) mode conversion over 130 nm wavelength range (1480~1610 nm). We further confirmed its high fabrication tolerance by characterizing the same devices from 8 intra-wafer dies and 4 wafer-to-wafer dies. All results revealed almost identical device performance.
\ue3\ue3We employed and optimized the profile of vertical PN junction which was applied to the micro-ring modulator (MRM). The benefit of vertical PN junction design over traditional lateral PN counterpart is to enable efficient phase modulation efficiency. As a result, the optimized MRM revealed 9.735 dB extinction ratio at 20 Gbps NRZ signal (1V Vpp). We also characterized the MRM device to extract the functionality of metal heater and PN phase shifter in a MRM. High incident laser intensity would lead to two-photon absorption and asymmetric spectral response of a MRM. When operating the MRM at 40 Gbps NRZ we observed an extinction ratio of 4.9870 dB.
\ue3\ue3We implemented a broadband directional coupler design in a MRM to achieve broadband operation with identical quality factor. The broadband spectral response is maintained under -50 to + 10 nm linewidth variation, but its optical coupling coefficient is sensitive to the linewidth variation. This device could also serve as a process monitoring device
A study on the operation efficiency of container Terminal by introducing the function of identified individual performance
No full textContainer terminals, as hubs between oceans and land, are pivotal for container transport. Other than a vast hinterland, sufficient heavy machineries are required for smooth operations at terminals. Thus, whether operators are skilled affects the flow of work. This study focuses on whether an individual performance efficiency identification system at container terminals brings positive effects to the overall work and the operators. We collected three years of work data through the system and took into account personnel backgrounds. We then conducted secondary data research so as to analyze overall work and individual work respectively. Results show that, with the installation of the individual performance efficiency identification system, work hours reduce significantly, and personnel performance efficiency and performance appraisal results improve significantly
Applying Topic Model on Patent Data As Innovative Indicators for Financial Forecasting
No full textWith the rapid development of technology and the Internet, companies pay more attention to their technology and innovation capabilities. the patent can not only improve the company's business reputation but can even increase the company's revenue. However, the patent application process is very complicated and rigorous, so it usually takes 3-5 years to obtain a patent. The complexity of the composition of patent documents also increases the complexity of text analysis. To assist companies in making accurate technology investment decisions, this thesis uses indicators extracted from different aspects of patent information and financial ratios to study the impact of patent information on company performance from different perspectives and uses machine learning methods to predict company performance. And an innovation indicator is proposed to conduct an in-depth discussion on the impact of the life cycle of each patent topic on company performance. After experiments, the patent topic indicator has a certain impact on company performance, and the patent topic in the growth stage is one of the important indicators
Design of Neural Network Processing Unit Based on Modified Posit Format and Block Floating Point Processing Technique
No full textHow to effectively encode the data of a neural network model will not only affect the amount of data transmission required during the process of the model, but also the design of the acceleration circuit architecture used to perform the model operation. In recent years, some low-bit floating-point encoding formats have been proposed, among which posit number is a new encoding format that has attracted much attention, and also been applied to the expression of neural network data. This thesis first proposes a modified posit number coding format for neural network data, because the data range of general network data is quite limited such that the field used to represent the exponent part can be reallocated for the mantissa. The experimental results show that the proposed encoding method for 8 bits can improve the inference accuracy of several major neural network models by about 3% compared with the original posit number representation. The proposed encoding method can also help simplify the implementation of posit number processing units. The area of decoding circuit of the exponent values can be reduced by about 30%. In addition, this thesis also applies the proposed modified floating point format to an existing neural network accelerator design. In order to reduce the area of the overall accelerator circuit, this thesis adopts the block floating-point technique. However, the overall accuracy may be degraded a lot. Therefore, this thesis further extends the original block floating-point method. When each layer of block data has been computed and stored back to the external memory, they will be scaled by shifting the same amount such that the average length of mantissa field can be maximized. The scale factor will be recorded in the internal memory in order to align the feature data when processing the next layer. Experiment results show that the modified block floating-point method can increase the accuracy of inference by about 2%. The proposed block floating-point method has been applied to a network accelerator which contains 576 multipliers, and the implementation results show that the area can be reduced by 45% compared with the direct implementation based on posit number processing units
The Power of Market Intermediaries: The Influence of Awards on Participants and Category Evolution in the U.S. Book Market
No full textOur conception of categories structures every part of our lives, and among markets that are generally categorized by categories, most studies focus on the product level, with less discussion of the category level. However, focusing on the product level, or on a single set of market participants and products, limits our understanding of the evolution of entire categories, and makes it difficult to understand how categories change between emergence and demise. This paper constructs a panel data set from 2007 to 2021 comprising Goodreads' book and review data, combining three representative and well-known book awards, and adding Amazon's bestseller list data to measure whether the awards' conferring of category status affects producers' and audiences' choices of categories by fixed-effects Poisson model, as well as the impact on the levels within categories, to find out the process of category evolution. By examining producers' and audiences' reactions to the conferring of category status, it is hoped to fill in the gaps in the concept of category growth. The data results show that awards do influence producers' and audiences' choices of product categories, and that category growth comes from the wider use of labels by producers and audiences as categories obtain conferred status
Electronic Word-of-Mouth and Japanese Reviews in the Hotel Industry of Taiwan
No full textOnline consumer ratings and textual comments are playing an increasingly important role in guiding potential consumers to make purchasing decisions. However, it is puzzling to find that the star ratings and review texts are inconsistent in many cases\ue2such as a review full of negative words, but with a full star rating. This research is aimed to examine whether ratings or reviews by Japanese travelers have an impact on the actual number of visitors in the hotel industry of Taiwan, and whether there is any inconsistency between ratings and sentiment of reviews. This paper studies the reviews left by Japanese on TripAdvisor from 2012 to 2019, combined with the actual number of Japanese guests in each hotel, in order to identify the important factors that have influenced Japanese tourists\ue2 choice of hotels in Taiwan over the past eight years. These findings can help to encourage the return of Japanese visitors after the pandemic, and accelerate the Taiwanese tourism industry\ue2s recovery.
In this study, the fixed-effects model is used to analyze the long-term effects of the eWOM variables on the number of Japanese visitors, which is the analysis of value-to-value correlations. In order to examine the correlation between the change of the eWOM variables and the change of the number of Japanese visitors, the first-difference estimator is extended so as to use the design of panel data. The results show that ratings and sentiment of reviews are inconsistent in Japanese reviews. Under the long-term observation of the fixed-effects estimator, the review volume and sentiment positively affect the number of Japanese tourists. However, under the short-term change of the first-difference estimator, an increase in the review volume in the previous month leads to a decrease in the number of Japanese visitors in the next month
Exploring the Interaction between Local Government and Social Innovation Organization from the Perspective of Actor Network Theory -A Case Study of Taoyuan City
No full textWith the impact of globalization and rapid economic development, various public issues have become diverse and complex, and the stakeholders involved are also very diverse. How to use "social enterprise" or "social innovation" to solve social problems has also received more attention. Governments of various countries also continue to focus on the development of various legislations, regulations and the formulation of administrative guidelines to help create a good environment for the development of the social innovation ecosystem. To understand the network interaction between local government, social innovation organizations and other multiple actors, this study takes Taoyuan City as the research object to gain an in-depth understanding of the interaction process and dilemma between local government and different actors under the policy background.
This study adopts Actor\ue2network theory as the core framework to understand the interaction process of actors with different characteristics in the social innovation ecosystem through the translation stage, and to analyze the impact of resources provided by local governments on social innovation organizations.
In order to understand the interaction process, secondary data analysis and semi-structured in-depth interviews are used to explore the effective resources of social innovation and entrepreneurship and the context of actors' involvement in the social innovation ecosystem, clarify the interactive relationship of multiple actors, and maximize the use of resources , and also for other local governments to refer to this interactive model to create a local social innovation ecosystem and create a more diffuse social impact.
The study found that the effective resources of social innovation and entrepreneurship including funds, competitions, counseling and matchmaking, space applications and intangible resources. The social innovation ecosystem in Taoyuan City is mainly composed of the Department of Youth Affairs, Taoyuan, which achieve complementary roles and has a good connection model. And the suggestions for the government to promote the social innovation ecosystem are Import diverse and corresponding resources according to the needs of different stages, plan effective resources and check the alignment of the goals of both parties, let the multiple actors join in and creating a friendly social innovation environment
Growth and characterization of GaN/InGaN/InAlN single quantum well on GaN template by Molecular Beam Epitaxy
No full textIn this research, we used plasma assisted molecular beam epitaxy system to grow GaN/InGaN/InAlN single quantum well on GaN template, which is on Patterned sapphire substrate.
It is difficult to grow high indium content and high quality InGaN quantum well and the electron-hole pairs are not easy to recombine in light emitting diodes device because of the different lattice constants and binding energies between InN and GaN. In order to solve this problem, we can use InAlN as the barrier of InGaN quantum well which has larger band off set to ensure the quantum confinement and also the matched lattice constant to reduce the defects in the blue, green and red InGaN/InAlN quantum wells.
In this thesis, we demonstrated the growth of blue light emitted InGaN/InAlN quantum well and improved its optical property. The experiments were divided to three series which are growth temperature dependence of InAlN barrier, indium flux dependence of InAlN barrier and different after growth cooling rate dependent. The samples were analyzed by SEM, PL, XRD and XPS. In series I and II, even though the growth temperature was increased and indium flux was decreased, the excess of iudium droplets were still found on the surface after growth which damaged the crystal quality. To prevent indium droplets diffused out, the cooling rate of the temperature after growth was increased in series III. By increasing the cooling rate, the indium droplets were eliminated and optical property of InGaN quantum well has been improved when InAlN barrier was grown at 630 oC with the cooling rate 20 oC per minute. The emission peak of InGaN/InAlN quantum well has been found at 2.8eV by PL spectrum while the indium content of InAlN is verified to be 25.9% in the XPS spectrum. It is verified we has grown the lattice matched InAlN barrier for the blue light emitted InGaN quantum well in our study
A study of big five personality traits and professional functions on customer satisfaction: The mediation effect of job performance
No full textIn recent years, due to the impact of the economic downturn and the test of the global outbreak of the new crown epidemic, the unemployment rate has increased, and many industries have been affected. Only the market demand for human services has not decreased significantly, but has increased, especially for home care and baby care. Babysitting and other related categories, this study focuses on confinement people who go to the house to study and discuss. Take the YesMa company in the market as an example. From 2021 to 2022, it will receive an average of more than 1,300 cases at home and abroad each year. The postpartum confinement mode of consumers is based on the common confinement center in the market and the delivery of cooked confinement meals. Changing to confinement at home and hiring professional confinement staff to serve at home, the scale of the home care market has gradually expanded; also because the Internet has become the most popular in recent years, it is a consideration of whether to choose to live in a "confinement center" after childbirth It needs to bear relatively high expenses, so the value measurement between "economic ability considerations" and "necessity of living in a confinement center" often causes disagreements between couples, and the proportion of online voices ranks second " mother-baby sitter " has been affected by social news events over the years, it has caused negative comments from the public. Most people think that the quality of mother-baby sitter varies greatly from person to person. With transparent information, the sense of uncertainty and insecurity is extremely high. After all, a stranger is invited to come to the house. Knowledge sharing\ue2 and \ue2customized confinement meals\ue2 have their advantages in terms of flexibility, and still attract many pregnant women, especially more than half of the non-first-born families will consider issues such as the care, pick-up and placement of children at home. Therefore, we can confirm that what the public cares most about is the level of professional functions of mother-baby sitter, the quality of work performance, personality traits, and whether there are objective measurement and evaluation data for reference. Therefore, this study distributed questionnaires to hundreds of mother-baby sitter and customers of YeaMa Company. After recycling and selection, 207 valid questionnaires were obtained. After statistical analysis, the results are as follows:
1. Conscientious personality traits of confinement mother-baby sitter are positively correlated with work performance
2. The personality traits of confinement mother-baby sitter are open and positively correlated with work performance
3. The impact of professional functions is positively related to job performance
4. The impact of professional functions is positively correlated with customer satisfaction
5. Job performance will positively affect customer satisfaction
6. The personality traits of confinement mother-baby sitters who are conscientious will positively affect customer satisfaction through work performance
7. The personality traits of confinement mother-baby sitter who are open will positively affect customer satisfaction through work performance
8. Professional Functions Positively Affect Customer Satisfaction Through Job Performance
Based on the results of the above research, it is found that it is consistent with our initial hypothesis, and because of this research, we will provide it for future research and reference in the practical industry
Impacts of educational programs to reduce blood culture contamination-- systematic review and meta-analysis
No full textBackground: Administration of appropriate antimicrobial therapy is one of the key factors for improving the survival rate and prognosis among bloodstream infection patients. Blood culture and antimicrobial testing (AST) are the gold standard method to identify bloodstream pathogens and appropriate antibiotics. Previous studies showed
that sample collection is the major contamination source, which mainly caused by not comply with the blood collection protocol. However, whether training program or reeducation could improve the recognition of correct blood collection for culture and change the practice behavior remain unclear, this study will evaluate the effectiveness of training programs for phlebotomists by meta-analysis. Objectives: The purpose of this study is to explore the impact of blood sampling training program for medical staff on the reduction of clinical contamination rate of blood culture. Materials and methods:This study is set in a medical environment, and analyzes the contamination rate of blood culture specimens of personnel before/after education and training. The study conducted a literature search in three databases. After the literature search was completed, duplicate literature was deleted and included in the research design of clinical executive education training as quasi-experimental studies. We assessed the quality of included studies using ROBINS-I. This meta-analysis integrated Odds Ratio (OR) in a random effect model to describe the difference in pollution risk, and the results were presented in a forest plot. Results: According to the keywords, 23 articles were finally selected after screening and reviewing. In the acute and severe intensive care unit without education and training, the contamination rate of blood samples before education and training was 5% (summarized percentage [95%CI] = 0.03 [0.04~0.06]). The risk of blood sample culture contamination was significantly reduced by 57%(summarized OR [95%CI] = 0.43 [0.37~0.50]) after education and training; and the contamination rate
with increased feedback significantly reduced by 59% (p-value<0.00001). Conclusion: The results of this study indicate that in the acute and severe medical environment, education and training can effectively reduce the contamination rate of blood culture
collection by medical staff in a high-pressure working environment. The feedback mechanism after education and training can also reduce the pollution rate, but its
effectiveness needs to be further explored