Figuring out the degree of semantic similarity between two sentences is important for many practical applications of natural language processing. The goal is to determine the similarity of sentences on a scale from "sentences are unrelated" to "sentences are equivalent". In this thesis we examined application of di erent neural network architectures to solve this problem. We proposed models based on Recurrent neural networks, which convert text sequence to constant sized vector. We followed up with suitable representation of unknown words. Our experiments showed that simple architectures achieved better results on the used dataset. We see a future extension of this thesis by using bigger training dataset. 1

In this thesis, we develop a physical understanding of the effects of viscosity and geometry on the dynamics of interfacial flows in drops and bubbles. We first consider the coalescence of pairs of conical water droplets surrounded by air. Droplet pairs can form cones under the influence of an electric field and have been observed to coalesce or recoil depending on the angle of this cone. With high resolution numerical simulations we show the coalescence and non-coalescence of these drop pairs is negligibly affected by the electric field and can be understood through a purely hydrodynamic process. The coalescence and recoil dynamics are shown to be self similar, demonstrating that for these conical droplet pairs viscosity has a negligible effect on the observed behavior. We generalize this result to the coalescence and recoil of droplets with different cone angles, and focus on droplets coalescing with a liquid bath and flat substrate. From the simulations of these droplets with different cone angles, an equivalent angle is found that describes the coalescence and recoil behavior for all water cones of any cone angle. While viscosity is found to negligibly affect the coalescence of conical water drops, it plays a key role in regulating the coalescence process of bursting gas bubbles. When these gas bubbles burst, a narrow liquid jet is formed that can break up into tiny liquid jet drops. Through consideration of the effects of viscosity, we show that these jet drops can be over an order of magnitude smaller than previously thought. Here, viscosity plays a key role in balancing surface tension and inertial forces and determining the size of the jet drops. Finally, we investigate the drainage of surfactant free, ultra-viscous bubbles where surface tension serves only to set the initial shape of the bubble. We use interferometry to find the thickness profiles of draining bubble films up to the point the of rupture. A theoretical film drainage model considering the balance of viscous and gravitational stresses is developed and numerically computed. The numerical results are found to be consistent with the experimentally obtained thickness profiles. In this work we provide insight into the role of viscosity in the outlined interfacial flows. The results of this thesis will advance the understanding of drop production in clouds, the marine climate, and the degassing of glass melts.

The expression ''Hire for attitude, train for skills'' is used as a motive to create a matching program where companies and job seekers' soft qualities are measured and compared against each other. Are there better or worse methods for this purpose, and how do they compare with each other? By associating soft qualities with companies and job seekers, it is possible to generate a value for how well they match. Therefore, data has been collected on several companies and job seekers. Their associated qualities are then translated into numerical vectors that can be used for matching purposes, where vectors closer together are more equal than vectors with greater distances. When it comes to analyzing and comparing the qualities, several methods have been used and compared with a subsequent discussion about their suitability. One consequence of the lack of a proper standard for presenting the qualities of companies and job seekers is that the data is messy and varied. An expected conclusion from the result is that the most flexible method is the one that generates the most accurate results.

Classification is a general framework that can be applied tovarious tasks such as object recognition, prediction, diagnosis or learning. There exist at least two different approaches for classification, namely rule-based and similarity-based classification. The two approaches have different strengths and weaknesses. The former requires a domain theory in order to make inferences from the test instance to its class. The latter does not have this requirement and approximates the class of a test instance via its similarity to a set of known instances.In this thesis the above two approaches are integrated in the realm of Case-Based Reasoning (CBR). CBR treats new cases according to their similarity to stored cases. Similarity is calculated by a similarity measure, which is the crucial factor for classification accuracy. In this work, rule-based domain knowledge is incorporated into the similarity measures of CBR in order to increase classification accuracy. Several novel integration methods are introduced, implemented and evaluated. Since knowledge about real world domains is typically imperfect, the approach does not assume that the domain theories are accurate or complete. Rather, a systematic analysis of different knowledge types is presented that shows the effect of imperfect knowledge on classification accuracy. The analysis is conducted partly empirically in artificial and in real world domains, and partly formally.

This study investigated the influence of chronological age and stimuli on the explanations of "same" and "different" by young children. Seventy children, between by young children. the ages of three-years, six-months and nine-years, six-months, selected on the basis of chronological age, normal speech and language development, and normal verbal maturity, were involved as subjects. A test consisting of a series of verbal tasks was administered to each child. The experimenter recorded and later analyzed and classified all responses for each child, following specific guidelines for judging appropriateness of response and assigning each appropriate response to one of three classifications.

Indiana University-Purdue University Indianapolis (IUPUI) / How is charitable giving influenced by other donors’ charitable giving? Do people give more in the presence of other donors who are similar to themselves? Most research suggests that individuals are positively influenced by others who are similar across a variety of behaviors. In the charitable giving contexts, people are more likely to donate (or donate more) to the same cause if others who are similar donate. Yet, prior research has paid little attention to potential non-linear effects of similarity on charitable giving. Is there a certain amount of similarity that is too much? My dissertation investigates this research question through two different methodological approaches, a systematic literature review and an experimental study. The findings suggest the curvilinear effects of similarity on charitable giving (i.e. self-other oversimilarity hypothesis); that is, individuals are more likely to donate (and donate more) in the presence of other generous donors who are moderately similar to themselves. Yet, individuals are less likely to donate (and donate) less in the presence of other generous donors who are in high similarity to themselves. In other words, too much similarity between donors may actually backfire in charitable giving contexts when others give generously. This dissertation consists of a brief overview of similarity (Chapter 1), a systematic literature review (Chapter 2), an experimental study (Chapter 3) and a research proposal (Chapter 4). Chapter 1 in this dissertation identifies the importance of similarity in social relationships. Chapter 2 investigates the effects of similarity on charitable giving and identifies the literature gap. Chapter 3 attempts to fill the gap via developing and testing self-other oversimilarity hypothesis. It further offers practical implications for nonprofit fundraising practices on how to apply similarity between donors to motivate more funding. In order to provide additional empirical evidence that may contribute to theory and practice, and to address certain limitations of the current experimental study, Chapter 4 proposes a new research project to further test self-other oversimilarity hypothesis in the presence of a stingy donor.

No description available.

The researcher of this quasi-experimental 2 x 2 factorial design study investigated the impact of an academic integrity module and Turnitin on undergraduate student similarity index scores. Similarity index scores were used to measure suggested plagiarism rates of student papers. The purposive sample was made up of 96 undergraduate education students enrolled in four sections of the same course in a Southeastern university. One main factor was submitting assignments through Turnitin, with two levels: yes or no. The second factor was completing the academic integrity module, with two levels: yes or no. The four intact groups were randomly assigned to the main factors. Although the findings from this study indicated that neither main factor nor their interaction were statistically significant at the .05 alpha level, the mean similarity index score of participants who submitted their paper through Turnitin was lower than the mean similarity index score of participants who did not submit their papers through Turnitin. Similarly, the mean similarity index score of participants who completed the academic integrity module was lower than the mean similarity index score of participants who did not complete the academic integrity module. Although not at the statistically significant level, the plagiarism rate as measured by the mean similarity index score of the group of participants who completed the academic integrity module and submitted papers through Turnitin was the lowest followed by the group of participants that completed the academic integrity module but did not submit papers through Turnitin, then the group of participants that did not complete the academic integrity module and did not submit papers through Turnitin. The group of participants who did not complete the academic integrity module but submitted papers through Turnitin had the highest mean similar index score suggesting a higher plagiarism rate as compared to the other groups. Finally, the researcher acknowledged the limitations of this study and made recommendations for future consideration. Other variables such as gender, demographics, major, and credit hours completed could be investigated. Replication of the study is recommended. Also, a qualitative approach could enhance future studies.

For decades, the United States has been known as the nation of immigrants due to the increasing number of immigrant students in the public school system. Although the population of immigrant students steadily increases annually, American society still pressures immigrants into acculturation to fulfill the United States ideals of academic achievement despite the United States claim of multiculturalism (Malcolm & Lowery, 2011). This research focuses on 1st - and 2nd generation immigrant students’ strife of acceptance in U.S. culture, while sill preserving their own native culture, and the influence it has on academic achievement. The researcher interviewed eight (8) adult participants who are either 1st - or 2nd generation immigrant college students. This qualitative case study research aims to determine if forced acculturation or assimilation using stereotypes and perceptions of similarity effects how immigrant students develop their cultural identity, and the influence it has on academic achievement. Four major themes emerged from the participants’ responses: parental approval, peer pressure, environmental influence, and feelings about their ethnic group. Basic findings supported that immigrant students’ cultural identity is threatened by stereotypes and perceptions of similarity.

Hybrid rocket propulsion technology shows promise for the next generation of sounding rockets and small launch vehicles. This paper seeks to provide details on the process of developing hybrid propulsion systems to the academic and amateur rocket communities to assist in future research and development. Scaling hybrid rocket motors for use in sounding rockets has been a challenge due to the inadequacies in traditional boundary layer analysis. Similarity scaling is an amendment to traditional boundary layer analysis which is helpful in removing some of the past scaling challenges. Maintaining geometric similarity, oxidizer and fuel similarity and mass flow rate to port diameter similarity are the most important scaling parameters. Advances in composite technologies have also increased the performance through weight reduction of sounding rockets through and launch vehicles. Technologies such as Composite Overwrapped Pressure Vessels (COPV) for use as fuel and oxidizer tanks on rockets promise great advantages in flight performance and manufacturing cost. A small scale COPV, carbon fiber ablative nozzle and a N class hybrid rocket motor were developed, manufactured and tested to support the use of these techniques in future sounding rocket development. The COPV exhibited failure within 5% of the predicted pressure and the scale motor testing was useful in identifying a number of improvements needed for future scaling work. The author learned that small scale testing is an essential step in the process of developing hybrid propulsion systems and that ablative nozzle manufacturing techniques are difficult to develop. This project has primarily provided a framework for others to build upon in the quest for a method to easily develop hybrid propulsion systems sounding rockets and launch vehicles.