Rough sets and neural networks both offer good theoretical background for data processing and analysis. However, both of them suffer from few issues. This thesis will investigate methods by which these two concepts are merged, and few such solutions will be implemented and compared with conventional algorithm to study the benefits of this approach.

Brazing and soldering, as advanced manufacturing processes, are of significant importance to industrial applications. It is widely accepted that joining by brazing or soldering is possible if a liquid metal wets the solids to be joined. Wetting, hence spreading and capillary action of liquid metal (often called filler) is of significant importance. Good wetting is required to distribute liquid metal over/between the substrate materials for a successful bonding. Topographically altered surfaces have been used to exploit novel wetting phenomena and associated capillary actions, such as imbibitions (a penetration of a liquid front over/through a rough, patterned surface). Modification of surface roughness may be considered as a venue to tune and control the spreading behavior of the liquids. Modeling of spreading of liquids on rough surface, in particular liquid metals is to a large extent unexplored and constitutes a cutting edge research topic. In this dissertation the imbibitions of liquid metal has been considered as pertained to the metal bonding processes involving brazing and soldering fillers. First, a detailed review of fundamentals and the recent progress in studies of non-reactive and reactive wetting/capillary phenomena has been provided. An imbibition phenomenon has been experimentally achieved for organic liquids and molten metals during spreading over topographically modified intermetallic surfaces. It is demonstrated that the kinetics of such an imbibition over rough surfaces follows the Washburn-type law during the main spreading stage. The Washburn-type theoretical modeling framework has been established for both isotropic and anisotropic non-reactive imbibition of liquid systems over rough surfaces. The rough surface domain is considered as a porous-like medium and the associated surface topographical features have been characterized either theoretically or experimentally through corresponding permeability, porosity and tortuosity. Phenomenological records and empirical data have been utilized to verify the constructed model. The agreement between predictions and empirical evidence appears to be good. Moreover, a reactive wetting in a high temperature brazing process has been studied for both polished and rough surfaces. A linear relation between the propagating triple line and the time has been established, with spreading dominated by a strong chemical reaction.

Under certain loading conditions, surfaces topography coupled with materials degree of brittleness can significantly compromise the mechanical performance of structures. The foregoing remains valid even if roughness is intentionally introduced for engineering reasons. In either case, stress can concentrate. The case of the stress concentration in surfaces having randomly distributed pits is a problem that, although being very practical, yet it remains unsolved. The complexity of a random configuration renders difficult the problem of analytically finding relationships between surface parameters and markers indicative of mechanical failure. Another difficulty is the reproducibility of replicates of specimens possessing random rough surfaces, for destructive testing followed by statistical analysis. An experimental technique to produce highly controlled replicates of random rough surfaces (including modeling of degradation growth) was developed. This method was used to experimentally and statistically study the effects on fracture of early randomly degraded surfaces of poly methyl methacrylate (PMMA) versus topographical parameters. Growth of degradation was assumed to go from an engineering surface to one whose heights are normally distributed. (Early stage of degradation is meant to be that level of roughness which is in the neighborhood of the critical flaw size for a given material). Among other findings, it was found that neither stress nor strain alone can be used to predict fracture at this early stage of degradation. However, fracture location was found to be strongly correlated to the ratio of the root-mean square roughness (RMS) to auto correlation length (ACL), above some RMS threshold. This correlation decreases as the material becomes less brittle (i.e., decrease of Young’s modulus or increase of percent of elongation). Simultaneously, a boundary value problem involving traction-free random rough surfaces was solved using a perturbation method, assuming elastic and isotropic conditions. For small RMS/ACL ratio, the solution for the RMS stress concentration factor, kt was found to be: kt = 1 + 2*SQRT(2)*(RMS/ACL), which agrees very well with the experimental work. Finally, a generalization of stress concentration factor formulas for several geometrical configurations and loading conditions into the Modified Inglis Formula was proposed. Finite element analysis was carried out and comparison was made with both experimental and analytical results. Applications of these results are broad. In surface engineering, for example, our analytical solution can be coupled with Fick’s Law to find critical conditions under which a film could become unstable to random roughness. Additionally, in design and maintenance of surfaces in service, it can be used to preliminarily assess how stress concentrates in surfaces where well defined notches cannot be used as an approximation.

Master of Science / School of Family Studies and Human Services / Deborah Norris / Risk and challenge in children’s play have steadily declined over the last 30 years due to adult fears about injuries and litigation, among other factors. This societal trend is important to remedy because not only do children miss out on the numerous crucial benefits in every domain that play, and specifically risk and challenge in play, provides, but research suggests it also can lead to a host of other problems like childhood obesity, more injuries as children create their own risk and challenge in inappropriate ways, and childhood psychopathology. Data on children in care demonstrate a large number of children enrolled in pre-kindergarten programs today, therefore it is important to understand young children’s risky play in the education context and the role that early childhood practitioners play in either supporting or hindering that play. The present study used an original survey derived from the literature to examine early childhood practitioners’ beliefs and perceptions about preschool children’s risky play, practitioner’s risky play practices, and the factors that influence those beliefs and practices. The results showed that practitioners generally had more positive than negative beliefs about risky play, but only rarely or occasionally allowed risky play to occur in their classrooms or centers. A variety of both global and situational factors influenced practitioners’ decisions to allow risky play or not. Participants’ beliefs and practices were positively correlated, and beliefs and practices were both negatively correlated with influences. Numbers of years of experience in the field and education level were not found to be significant predictors of participants’ risky play beliefs and practices. These results have implications for professional development trainings as well as teacher education programs.

We investigate the use of the concept of indiscernibilityand vagueness in spatial information systems. To representindiscernibility and vagueness we use rough sets, respectivelyfuzzy sets. We introduce a theoretical model to supportapproximate queries in information systems and we show howthose queries can be used to perform uncertain classi.cations.We also explore how to assess quality of uncertainclassi.cations and ways to compare those classi.cations to eachother in order to assess accuracies. We implement the querylanguage in an SQL relational language to demonstrate thefeasibility of approximate queries and we perform an experimenton real data using uncertain classi.cations.

Knowledge discovery is an important process in data analysis, data mining and machine learning. Typically knowledge is presented in the form of rules. However, knowledge discovery systems often generate a huge amount of rules. One of the challenges we face is how to automatically discover interesting and meaningful knowledge from such discovered rules. It is infeasible for human beings to select important and interesting rules manually. How to provide a measure to evaluate the qualities of rules in order to facilitate the understanding of data mining results becomes our focus. In this thesis, we present a series of rule evaluation techniques for the purpose of facilitating the knowledge understanding process. These evaluation techniques help not only to reduce the number of rules, but also to extract higher quality rules. Empirical studies on both artificial data sets and real world data sets demonstrate how such techniques can contribute to practical systems such as ones for medical diagnosis and web personalization. In the first part of this thesis, we discuss several rule evaluation techniques that are proposed towards rule postprocessing. We show how properly defined rule templates can be used as a rule evaluation approach. We propose two rough set based measures, a Rule Importance Measure, and a Rules-As-Attributes Measure, %a measure of considering rules as attributes, to rank the important and interesting rules. In the second part of this thesis, we show how data preprocessing can help with rule evaluation. Because well preprocessed data is essential for important rule generation, we propose a new approach for processing missing attribute values for enhancing the generated rules. In the third part of this thesis, a rough set based rule evaluation system is demonstrated to show the effectiveness of the measures proposed in this thesis. Furthermore, a new user-centric web personalization system is used as a case study to demonstrate how the proposed evaluation measures can be used in an actual application.

Knowledge discovery is an important process in data analysis, data mining and machine learning. Typically knowledge is presented in the form of rules. However, knowledge discovery systems often generate a huge amount of rules. One of the challenges we face is how to automatically discover interesting and meaningful knowledge from such discovered rules. It is infeasible for human beings to select important and interesting rules manually. How to provide a measure to evaluate the qualities of rules in order to facilitate the understanding of data mining results becomes our focus. In this thesis, we present a series of rule evaluation techniques for the purpose of facilitating the knowledge understanding process. These evaluation techniques help not only to reduce the number of rules, but also to extract higher quality rules. Empirical studies on both artificial data sets and real world data sets demonstrate how such techniques can contribute to practical systems such as ones for medical diagnosis and web personalization. In the first part of this thesis, we discuss several rule evaluation techniques that are proposed towards rule postprocessing. We show how properly defined rule templates can be used as a rule evaluation approach. We propose two rough set based measures, a Rule Importance Measure, and a Rules-As-Attributes Measure, %a measure of considering rules as attributes, to rank the important and interesting rules. In the second part of this thesis, we show how data preprocessing can help with rule evaluation. Because well preprocessed data is essential for important rule generation, we propose a new approach for processing missing attribute values for enhancing the generated rules. In the third part of this thesis, a rough set based rule evaluation system is demonstrated to show the effectiveness of the measures proposed in this thesis. Furthermore, a new user-centric web personalization system is used as a case study to demonstrate how the proposed evaluation measures can be used in an actual application.

With the popularization of Internet and the wide use of electronic mails, the number of spam mails grows continuously. The matter has made e-mail users feel inconvenient. If e-mail servers can be integrated with data mining and artificial intelligence techniques and learn spam rules and filter out spam mails automatically, they will help every person who is bothered by spam mails to enjoy a clear e-mail environment. In this research, we propose an architecture called union defense to oppose against the spread of spam mails. Under the architecture, we need a rule-based data mining and artificial intelligence algorithm. Rough sets theory will be a good choice. Rough sets theory was proposed by Palwak, a logician living in Poland. It is a rule-based data mining and artificial intelligence algorithm and suitable to find the potential knowledge of inexact and incomplete data out. This research developed a spam filter based on rough sets theory. It can search for the characteristic rules of spam mails and can use these rules to filter out spam mails. This system set up by this research can be appended to most of existing e-mail servers. Besides, the system support Chinese, Japanese and Korean character sets and overcome the problem that most spam filters only can deal with English mails. We can develop a rule exchange approach between e-mail servers in the future works to realize union defense.

The observed multiscale phenomenon of rough surfaces, i.e. the smaller mountains mount on the bigger ones successively, renders the hierarchical structures which are described by the fractal geometry. In this situation, when two rough surfaces are loaded together with a higher load, the smaller asperities will undergo plastic flow and immerge into the bigger asperities below them. In other words, the higher load needs to be supported by the bigger asperities. However, when the GW model was proposed in 1966, its analytical method considered that the length-scale of asperities is fixed, which is independent of load (or surface separation). In such condition, the analytical results for a specific asperity length-scale can only suit the situation of a certain narrow range of load. In this research, a new model, called the multiscale GW model, has been developed, which takes into account the relationship between the load and the asperity length-scale. At first, based on the Nayak¡¦s model the multiscale asperity properties with different surface parameters have been derived, and based on the material yielding theory a criterion for determining the optimal asperity length-scale, which functions as supporting the load, is developed. Then both of the above are integrated into the GW model to build the multiscale GW model. The new model is compared with traditional one qualitatively and quantitatively and show their essential differences. The effects of surface parameters and material parameters are discussed in this model. Finally a comparison with the experiment is made, and reveal the good coincidence.

Growing volume of spam mails have not only decreased the productivity of people but also become a security threat on the Internet. Mail servers should have abilities to filter out spam mails which change time by time precisely and manage increasing spam rules which generated by mail servers automatically and effectively. Most paper only focused on single aspect (especially for spam rule generation) to prevent spam mail. However, in real word, spam prevention is not just applying data mining algorithm for rule generation. To filter out spam mails correctly in a real world, there are still many issues should be considered in addition to spam rule generation. In this paper, we integrate three modules to form a complete anti-spam system, they are spam rule generation module, spam rule reinforcement module and spam rule exchange module. In this paper, rule-based data mining approach is used to generate exchangeable spam rules. The feedback of user¡¦s returns is reinforced spam rule. The distributing spam rules are exchanged through machine-readable XML format. The results of experiment draw the following conclusion: (1) The spam filter can filter out the Chinese mails by analyzing the header characteristics. (2) Rules exchanged among mail improve the spam recall and accuracy of mail servers. (3) Rules reinforced improve the effectiveness of spam rule.