111
Thesis Title: Improved SPI-Calculus for Reasoning on Cryptographic Protocols
Author: Saleh Mansour Bani Hani, Supervisor: Dr. Hasan Al-Refai and Dr. Mourad Maouche, Year: 2012
Faculty: Information Technology, Department: Computer Science

Abstract: Most of cryptographic protocols are subjects to very subtle attacks. Therefore, many researchers have developed tools to model and analyze protocols to guarantee their security properties. Among these developed tools, the SPI calculus has proved to be a powerful formal method has useful for analyzing and reasoning on cryptographic protocols. However, such research work assumed that the environment will be restricted to some predefined rules and assumptions without solving the case of interacting with real systems that matched with such restrictions. In this thesis, we introduced an improved version of SPI calculus called the PH-SPI calculus. Such calculus provides the ability for evaluating each action in the running processes and making suitable decision to be more suitable with open environment. Therefore, we have enhanced the evaluation function to make it capable for making suitable decisions to handle the open and changeability in environment behaviors as well as evaluating and validating every action in protocol processes for proving the main security properties as authentication and confidentiality. In the PH-SPI calculus, the message is structured to be the same as in real protocols that have a tuple of messages. Also, it includes all operators needed for such protocols such as timestamp, hash function, digital signature and asymmetric key cryptosystem. Hence, PH-SPI calculus ready is to be used for real protocols such as e-commerce protocols.

Keywords: Cryptographic Protocol, Cryptographic Protocol Analysis, SPI Calculus, Evaluation Function, Testing Equivalence

112
Thesis Title: An Agent Modeling Formalism Enhancing AUML Class Diagram
Author: Mahmoud Adnan Ibrahim Sawalha, Supervisor: Prof. Said Ghoul, Year: 2008
Faculty: Information Technology, Department: Computer Science

Abstract: Multi-Agent Systems (MAS) has been used successfully for years with different purposes. It is used in systems that using some kind of intelligence and automation. Nowadays, there are a lot of modeling languages used to model MAS. One of the well-known MAS modeling languages is Agent Unified Modeling Language (AUML). AUML is an agent modeling language based on Unified Modeling Language (UML 2.0), it enhances some of UML diagrams and it doesn't use or enhance the remain of UML diagrams. Even, AUML is the closest agent modeling language to UML; it still has some serious weaknesses that have not been solved yet while dealing with agents. This study enhanced the agent class diagram in the agent modeling language AUML and presents a new agent class diagram that solves some of the weaknesses of AUML by using strengthens of some other agent modeling languages.

Keywords: Multi-Agent Systems (MAS), Agent Unified Modeling Language (AUML)

113
Thesis Title: Character recognition system for artistic Arabic style
Author: Abed Alssalam Aqel Khaleel Abu Odeh, Supervisor: Dr. Moayad A. Fadhil, Year: 2008
Faculty: Information Technology, Department: Computer Science

Abstract: Despite the tremendous achievement which has been made on the reading mechanism in the field of desktop publishing, the door remains open to the great efforts should be done in this field for the Arabic language contrary to the languages based on Latin and Chinese characters. The characteristics of the Arabic text, such as the complexity of characters and overlapping, and some characters do not accept the contact only one side in addition to the varied forms of characters in accordance with its position in the middle, beginning or the end of the word, all these factors increase the specificity and uniqueness of the research in this area. The recognition of Arabic characters pass through several steps; the preprocessing, segmentation, feature extraction, recognition and post processing. However, the Arab character recognition depends mainly on the segmentation, because the segmentation has many problems resulting from the characteristics of Arabic characters such as the complexity of characters and overlap, a number of researches have emerged dealing with this problem in the printed and handwritten text, and reached a series of solutions and proposals and recommendations to solve the problem of segmentation.

Keywords: Arabic Language, Complexity, Overlapping, Preprocessing, Segmentation, Feature Extraction, Recognition and Post Processing

114
Thesis Title: Enhancement of weight calculation in ranking of internet search engines
Author: Hisham Khaleel Hamed Abu Jalban, Supervisor: Dr. Moayad A. Fadhil, Year: 2008
Faculty: Information Technology, Department: Computer Science

Abstract: This thesis proposes new factors for indexing documents; enhance the weight calculations for the ranking algorithm and combining the advanced link-based ranker with the term-based ranker, in order to enhance the ranking score for documents, which led to improve the recall and precision of the search engine and for making more relevant documents appear at the beginning of the results list. The proposed system which is made especially to test the new enhancements on indexer and ranker has been tested and evaluated, with various queries, and used to compare the advanced link-based page ranker with the original link-based ranker and term-based ranker. The relevance of the returned documents (recall, precision) has been considerably improved in comparison between the advanced link-based ranker with original link-based ranker and term-based ranker by more than 13% on queries’ results tested on more than 3000 Web pages.

Keywords: Indexing Documents, Ranking Algorithm, Link-Based

115
Thesis Title: Prediction of the Dead Sea Water Level Using Neural Networks
Author: "Mohammed Khaled" Yousef Shambour, Supervisor: Dr. Rashid Al-Zubaidy, Year: 2008
Faculty: Information Technology, Department: Computer Science

Abstract: The Dead Sea (DS) basin plays a major role for regional economic development (industry, tourism and agriculture) in Jordan. Different studies stated that the water level of the DS is dropping an average of 3 feet per year. Accordingly there is a need to provide accurate and reliable estimates for the water level to help the researchers and geologists of the DS to make different kind of studies giving results, so they can understand the state of the DS and its behavior and stop the dropping of the DS water level. Neural Networks (NN) are computational models with the capacity to learn, to generalize, or to organize data based on parallel processing. Among all kinds of networks, the most widely used are Backpropagation (BP), Levenberg-Marquardt (L-M), and Generalized Regression Neural Networks (GRNN) that are capable of representing non-linear functional mappings between inputs and outputs. Different NN based DS water level prediction models are built and compared to determine the most effective neural networks work in prediction. It is known that DS water level depends on many factors such as Air temperature, Salinity, Humidity and other environmental information. Our NN models capture different subsets of those effects; reflect them within our models to identify the most effective set, which has significant impact on the water level of DS. Finally, we can say that the proposed GRNN model provides best significant performance results comparing with other NN models using Mean Square Error (MSE).

Keywords: Dead Sea, Neural Networks, Backpropagation, Levenberg-Marquardt, GRNN, MSE