An ontology mapping neural network (OMNN) is proposed in order to learn and infer correspondences among ontologies. It extends the Identical Elements Neural Network (IENN)s ability to represent and map complex relationships. The learning dynamics of simultaneous (interlaced) training of similar tasks interact at the shared connections of the networks. The output of one network in response to a stimulus to another network can be interpreted as an analogical mapping. In a similar fashion, the networks can be explicitly trained to map specific items in one domain to specific items in another domain. Representation layer helps the network learn relationship mapping with direct training method. The OMNN approach is tested on family tree test cases. Node mapping, relationship mapping, unequal structure mapping, and scalability test are performed. Results show that OMNN is able to learn and infer correspondences in tree-like structures. Furthermore, OMNN is applied to several OAEI benchmark test cases to test its performance on ontology mapping. Results show that OMNN approach is competitive to the top performing systems that participated in OAEI 2009.

This thesis explores four research areas that are examined using DNS traffic analysis. The tools used for this analysis are presented first. The four topics examined are domain mapping, response time of anti-phishing block lists to find the phishing sites, automated identification of malicious fast-flux hosting domains, and identification of distributed denial of service attacks. The first three approaches yielded successful results, and the fourth yields primarily negative lessons for using DNS traffic analysis in such a scenario. Much of the analysis concerns the anti-phishing response time, which has yielded tentative results. It is found that there is significant overlap between the automatically identified fast-flux sites and those sites on the block list. It appears that domains were being put onto the list approximately 11 hours after becoming active, in the median case, which is very nearly the median lifetime of a phishing site. More recently collected data indicates that this result is extremely difficult to verify. While further work is necessary to verify these claims, the initial indication is that finding and listing phishing sites is the bottleneck in propagating data to protect consumers from malicious phishing sites.

This research explores the role of affect in the domain of human wayfinding by asking if increased motivation will alter the performance across various routes of increasing complexity. Participants were asked to perform certain navigation tasks within an indoor Virtual Reality (VR) environment under either motivated and not-motivated instructions. After being taught to navigate along simple and complex routes, participants were tested on both the previously learned routes and new routes that could be implicitly derived from the prior spatial knowledge. Finally, participants were tested on their ability to follow schematized instructions to explore familiar and unfamiliar areas in the VR environment. Performance of the various spatial tasks across the motivated and control groups indicated that motivation improved performance in all but the most complex conditions. Results of the empirical study were used to create a theoretical model that accounts for the influence of affect on the access of route knowledge. Results of the research suggest the importance of including past knowledge and affect of the traveler as components of future wayfinding systems.

Driven by the explosive growth of information available online, the World-Wide-Web is currently witnessing a trend towards personalized information access. As part of this trend, numerous personalized news services are emerging. The goal of this project is to develop a prototype algorithm for using bookmarks to develop a personal profile. Ultimately, we imagine this might be used to construct a personalized RSS reader for reading news online. A reader returns a large number of news stories. To increase user satisfaction it is useful to rank them to bring the most interesting to the fore. This ranking is done by implementing a personalized profile. One way to create such a profile might be to extract it from users bookmarks. In this paper, we describe a process for learning user interest from bookmarks and present an evaluation of its effectiveness. The goal is to utilize a user profile based on bookmarks to personalize results by filtering and re-ranking the entries returned from a set of user defined feeds.

Bayesian networks (BNs) have proven to be a modeling framework capable of capturing uncertain knowledge and have been applied successfully in many domains for over 25 years. The strength of Bayesian networks lies in the graceful combination of probability theory and a graphical structure representing probabilistic dependencies among domain variables in a compact manner that is intuitive for humans. One major challenge related to building practical BN models is specification of conditional probability distributions. The number of probability distributions in a conditional probability table for a given variable is exponential in its number of parent nodes, so that defining them becomes problematic or even impossible from a practical standpoint. The objective of this dissertation is to develop a better understanding of models for compact representations of local probability distributions. The hypothesis is that such models should allow for building larger models more efficiently and lead to a wider range of BN applications.

As the Internet moves towards a three-layer architecture consisting of overlay networks on top of the IP network layer on top of WDM-based physical networks, incorporating the interaction between and among network layers is crucial for efficient and effective implementation of survivability. <br/><br/> This dissertation has four major foci as follows: First, a first-of-its-kind analysis of the impact of overlay network dependency on the lower layer network unveils that backhaul, a link loop that occurs at any two or more lower layers below the layer where traffic is present, could happen. This prompts our proposal of a crosslayer survivable mapping to highlight such challenges and to offer survivability in an efficient backhaul-free way. The results demonstrate that the impact of layer dependency is more severe than initially anticipated making it clear that independent single layer network design is inadequate to assure service guarantees and efficient capacity allocation. Second, a forbidden link matrix is proposed masking part of the network for use in situations where some physical links are reserved exclusively for a designated service, mainly for the context of providing multiple levels of differentiation on the network use and service guarantee. The masking effect is evaluated on metrics using practical approaches in a sample real-world network, showing that both efficiency and practicality can be achieved. Third, matrix-based optimization problem formulations of several crosslayer survivable mappings are presented; examples on the link availability mapping are particularly illustrated. Fourth, survivability strategies for two-layer backbone networks where traffic originates at each layer are investigated. Optimization-based formulations of performing recovery mechanisms at each layer for both layers of traffic are also presented. Numerical results indicate that, in such a wavelength-based optical network, implementing survivability of all traffic at the bottom layer can be a viable solution with significant advantages. <br/><br/> This dissertation concludes by identifying a roadmap of potential future work for crosslayer survivability in layered network settings.

The new trend on the Web has totally changed todays information access environment. The traditional information overload problem has evolved into the qualitative level beyond the quantitative growth. The mode of producing and consuming information is changing and we need a new paradigm for accessing information. Personalized search is one of the most promising answers to this problem. However, it still follows the old interaction model and representation method of classic information retrieval approaches. This limitation can harm the potential of personalized search, with which users are intended to interact with the system, learn and investigate the problem, and collaborate with the system to reach the final goal. This dissertation proposes to incorporate interactive visualization into personalized search in order to overcome the limitation. By combining the personalized search and the interac- tive visualization, we expect our approach will be able to help users to better explore the information space and locate relevant information more efficiently. We extended a well-known visualization framework called VIBE (Visual Information Browsing Environment) and implemented Adaptive VIBE, so that it can fit into the per- sonalized searching environment. We tested the effectiveness of this adaptive visualization method and investigated its strengths and weaknesses by conducting a full-scale user study. We also tried to enrich the user models with named-entities considering the possibility that the traditional keyword-based user models could harm the effectiveness of the system in the context of interactive information retrieval. The results of the user study showed that the Adaptive VIBE could improve the precision of the personalized search system and could help the users to find out more diverse set of information. The named-entity based user model integrated into Adaptive VIBE showed improvements of precision of user annotations while maintaining the level of diverse discovery of information.

Wireless ad hoc and sensor networks have received attention from research communities over the last several years. The ability to operate without a fixed infrastructure is suitable for a wide range of applications which in many cases require protection from security attacks. One of the first steps to provide security is to distribute cryptographic keys among nodes for bootstrapping security. The unique characteristics of ad hoc networks create a challenge in distributing keys among limited resource devices. In this dissertation we study the impact on secure connectivity achieved through key pre-distribution, of jamming attacks which form one of the easiest but efficient means for disruption of network connectivity. In response to jamming, networks can undertake different coping strategies (e.g., using power adaptation, spatial retreats, and directional antennas). Such coping techniques have impact in terms of the changing the initial secure connectivity created by secure links through key predistribution. The objective is to explore how whether predistribution techniques are robust enough for ad hoc/sensor networks that employ various techniques to cope with jamming attacks by taking into account challenges that arise with key predistribution when strategies for coping with jamming attacks are employed. In the first part of this dissertation we propose a hybrid key predistribution scheme that supports ad hoc/sensor networks that use mobility to cope with jamming attacks. In the presence of jamming attacks, this hybrid scheme provides high key connectivity while reducing the number of isolated nodes (after coping with jamming using spatial retreats). The hybrid scheme is a combination of random key predistribution and deployment-based key predistribution schemes that have complementary useful features for secure connectivity. In the second part we study performance of these key predistribution schemes under other jamming coping techniques namely power adaptation and directional antennas. We show that the combination of the hybrid key predistribution and coping techniques can help networks in maintaining secure connectivity even under jamming attacks.

Adaptive hypermedia is one of the most popular approaches of personalized information access. When the field started to emerge, the expectation was that soon nearly all published hypermedia content could be adapted to the needs, preferences, and abilities of its users. However, after a decade and a half, the gap between the amount of total hypermedia content available and the amount of content available in a personalized way is still quite large. In this work we are proposing a novel way of speeding the development of new adaptive hypermedia systems. The gist of the approach is to extract the adaptation functionality out of the adaptive hypermedia system, encapsulate it into a standalone system, and offer adaptation as a service to the client applications. Such a standalone adaptation provider reduces the development of adaptation functionality to configuration and compliance and as a result creates new adaptive systems faster and helps serve larger user populations with adaptively accessible content. To empirically prove the viability of our approach, we developed PERSEUS server of adaptation functionalities. First, we confirmed that the conceptual design of PERSEUS supports realization of a several of the widely used adaptive hypermedia techniques. Second, to demonstrate that the extracted adaptation does not create a significant computational bottleneck, we conducted a series of performance tests. The results show that PERSEUS is capable of providing a basis for implementing computationally challenging adaptation procedures and compares well with alternative, not-encapsulated adaptation solutions. As a result, even on modest hardware, large user populations can be served content adapted by PERSEUS.

Wireless Sensor Networks (WSNs) have unique characteristics which constrain them; including small energy stores, limited computation, and short range communication capability. Most traditional security algorithms use cryptographic primitives such as Public-key cryptography and are not optimized for energy usage. Employing these algorithms for the security of WSNs is often not practical. At the same time, the need for security in WSNs is unavoidable. Applications such as military, medical care, structural monitoring, and surveillance systems require information security in the network. As current security mechanisms for WSNs are not sufficient, development of new security schemes for WSNs is necessary. New security schemes may be able to take advantage of the unique properties of WSNs, such as the large numbers of nodes typical in these networks to mitigate the need for cryptographic algorithms and key distribution and management. However, taking advantage of these properties must be done in an energy efficient manner. The research examines how the redundancy in WSNs can provide some security elements. The research shows how multiple random delivery paths (MRDPs) can provide data integrity for WSNs. Second, the research employs multiple sinks to increase the total number of duplicate packets received by sinks, allowing sink voting to mitigate the packet discard rate issue of a WSN with a single sink. Third, the research examines the effectiveness of using multiple random paths in maintaining data confidentiality in WSNs. Last, the research examines the use of a rate limit to cope with packet flooding attacks in WSNs.