M.S. Theses

Permanent URI for this collection

https://digitalarchive.library.bogazici.edu.tr/handle/123456789/11645

Browse

Now showing 1 - 20 of 505

Implementation of the analysis part of a pascal translator
(Thesis (M.S.) - Bogazici University. Institute for Graduate Studies in Science and Engineering, 193., 1983.) Ataç, H. Cem.; Tosun, Oğuz.
This thesis describes the implementation of the analysis part of a single pass translator for the programming language STANDARD PASCAL. The translation process includes syntactic analysis and semantic analysis at the declaration level, but excludes code generation. The details of the implementation are discribed, with a short descripticin of alternative approaches in each section. The source and test runs of the program are given in the appendixes A and B respectively.
An interactive information retrieval system: |SORGU
(Thesis (M.S.) - Bogazici University. Institute for Graduate Studies in Science and Engineering, 1984., 1984.) Kural, Namık.; Balman, Tunç.
There are several interactive information retrieval software packages that maintain easy access to data base items and structures. It is relatively easy for persons famil iar with computer to use these packages. But to persons who are unfamiliar, especially the managers, using these packages is not easy. The syntax of these packages is extraordinarily different for these persons and usually do not wish to spend time learning this syntax. SORGU is a turkish based interactive information retrieval software package. The queries are formed using menus. It does not have any syntax convention. Anybody who has knowledge about the fundamental concepts of data base management systems can easily use SORGU. It is a completely menu driven inquiry system that provides additional facil ities to change the item 'naming conventions, supports multiple enquiries and user defined report structures. Because of the ease of use training time is minimized.
A multi-tasking executive
(Thesis (M.S.) - Bogazici University. Institute for Graduate Studies in Science and Engineering, 1984., 1984.) Yılmazer, Sedat.; Balman, Tunç.
This thesis describes the design and implementation of a real-time multi-tasking priority-driven executive, FOX. The name stands for "a East ObJect oriented executive. FOX permits the micro-computer user to enjoy the multi-tasking capabilities of the high performance machines like minicomputers. FOX tries to unify the concept of multi-tasking and team study of projects. FOX uses the concept of exchange for inter task communication and synchronization. For interrupts, FOX uses special exchanges called flags. There is no limit on the number of tasks and number of exchanges that can be created under FOX. There are 16 flags managed by FOX. Eigth of these flags are assigned to eight external interrupt sources via an interrupt priority controller. The remaining eight can be used by the programmers for fast and easy event processing. Each task running under FOX has an eight bit priority level, which is used by the task dispacher. All task exchange and interrupt binding is done dynamicaly by FOX to ease the team study of projects of large code size.
A BCPL translator using recursive descent technique
(Thesis (M.S.)- Bogazici University. Institute for Graduate Studies in Science and Engineering, 1984., 1984.) Kadifeli, Vasıl.; Balman, Tunç.
This thesis describes the implementation of a BCPL translator on a CDCCYBER 170/815 computer system using Recursive Descent Technique. The translator has been written in Pascal. The output of the translator is CDC COMPASS assembler instructions. A minimal run-time library of COMPASS routines have been prepared to provide for the interface of the BCPL language to the operating system.
Deadlock detection problem in computing systems :|a simulation approach using a priority based deadlock detection algorithm
(Thesis (M.S.) - Bogazici University. Institute for Graduate Studies in Science and Engineering, 1989., 1989.) Akgün, Sema F.; Tosun, Oğuz.
In this thesis, the deadlock detection problem in computing systems is examined in detail. Deadlock models and some published algorithms on deadlock detection are discussed. A modified priority based algorithm is introduced and some more modifications are offered to make the algorithm correct and more efficient. The final version of the algorithm is simulated for a single-site system. To show the effects of these modifications. the simulation results obtained with modifications are compared with the results obtained without them. It is observed that after the modifications. the system performed better. For further simulation studies. a distributed system model is offered.
Generating causal relations from mathematical models
(Thesis (M.S.)- Bogazici University. Institute for Graduate Studies in Science and Engineering, 1989., 1989.) Güngör, Tunga.; Kuru, Selahattin,
System analysis is an important topic for many scientific disciplines. By a system, we mean an orderly, interconnected arrangement of parts describing a phenomenon. The formal representation of systems is done through mathematical modeling. Thus the analysis of a system is performed by analysing the mathematical model of the system. The mathematical model of a system contains implicit information about the system it describes - how the system behaves under various conditions, what are the relationships between the variables, how the cause-effect sequence of the variables be arranged, etc. Currently, the exposition of this information in order to understand the system truly is usually performed by humans. This thesis is a step towards the automation of this process and introduces for this purpose some techniques. These techniques are based on the use of some well-known mathematical tools: partial derivatives and total differentials of the closed form functions defining a system. Partial derivatives and total differentials are analysed to make the causal relations implied by the model explicit. The mathematical models addressed by our work are restricted to models of the form of algebraic equations.
Scheduling problem in distributed hard real-time computer systems :|a simulation approach to dynamic task scheduling using focused addressing and bidding
(Thesis (M.S.) - Bogazici University. Institute for Graduate Studies in Science and Engineering, 1990., 1990.) Tüzün, Z. Dilek (Duman).; Tosun, Oğuz.
The unique feature, the time constraint, of hard real-time systems makes them different from the traditional computer systems because in such systems the required tasks must be executed not only functionally correctly but in a timely manner. In this thesis, the scheduling problem of hard real-time tasks in distributed systems is examined in detail. Previous work on the algorithms proposed for scheduling in hard real-time systems is reviewed. A heuristic algorithm which considers not only CPU scheduling but also general resource requirements of tasks is chosen to be evaluated. A set of heuristics that can be used by this algorithm is studied through a sequence of simulations. The heuristic function which is observed to perform the best is incorporated in the distributed scheduling algorithm. In this algorithm the determination of a good destination node for a locally nonguaranteed task, is based on a technique that combines bidding and focused addressing algorithms. Simulation studies are conducted in order to evaluate the performance of the algorithm in a wide range of application enviroil.ments. The performance of the algorithm' is also compared to that of three other distributed scheduling algorithms. It is observed that though this algorithm is sensitive to the characteristics of the environments, it performs well in a wide range of environments, compared with the other algorithms.
An intelligent database interface for Turkish
(Thesis (M.S.)- Bogazici University. Institute for Graduate Studies in Sciences and Engineering, 1991., 1991.) Darcan, Osman Nuri.; Kuru, Selahattin,
In this thesis, a portable natural language interface system for communicating with databases in Turkish is developed. The system does a two step transformation from a Turkish query in user's view to an intermediate meaning repl'esentation language D&Q and finally to a target database language SQL. It is composed of domain independent runtime modules for different processing stages, namely language processing, internal query generation and translation to SQL. Modules refer to the knowledge base in which diverse knowledge about the domain and the database are maintained. Two additional modules, to wita spelling corrector and a history keeper are incorporated in the system. A syntactic parser is used in analyzing queries. for the syntactic parser. a formalization of a subset of Turkish grammar based on the simple principle of general categorization incorporated with the notion of modifications between words is proposed and a grammar that consists of a collection of rervrite rllles for the formal representation of sentences is discussed. A decision tree which works with suffix strip off approach is used for the morphological parsing of n.ouns. Parse trees produced for different types of sentences using the formalized grammar are given. Regarding to the meaning representation, an "intelllgent" meaning representation generator which has a rule based reasoning capability is designed. The interpretations of some modification relations are discussed in details. Finally the interpretation of a full sentence is shown. The system is tested on an imaginary stude-course-instructor database, all examples refers to this database. Possible extensions for both the parser and the meaning representation generator are also proposed at the end of the thesis.
Using Pade approximation in system modelling and simulation with state-space representation
(Thesis (M.S.) - Bogazici University. Institute for Graduate Studies in Science and Engineering, 1992., 1992.) Ceylan, Ülkü (Kolağasıoğlu).; Akın, H. Levent.
Today, in a variety of application the statistical characteristics of a system response is important in order for analysis and model the systems. In this study, we mainly made an investigation to the system analyzing and modelling methods. Especially, we considered Autoregressive Moving Average (ARMA) and Pade' approximation methods to find the modelled system transfer function coefficients. Theie are several algorithms to calculate these coefficients. In our study we used Modified yule Walker Algorithm (MYWE) and AKAIKE algorithms for ARMA and a new Pade' algorithm developed by Biyiksiz for Pade' approximation. When these three methods were simulated, it was seen that Pade' is mainly less sensitive to the coefficient quantization error and arithmetic round-off error accumulation introduced by finite word length. On the other hand it is not a good approximation for higher orders.It was seen that if the lower orders were used, Pade' approximation gave really good results compared to the MYWE and AKAIKE. But these ARMA models also are not guaranteed to give stable solutions for higher orders. In some cases for higher or lower order ARMA models produced good results especially for higher orders. But these orders should be choosen with one of the methodologies described for model order selection. An extension of research was done to the state-space error sensitivity. When the mentioned errors were investigated for different representation types of the state-space approach, it was shown that Pade' algorithm was less sensitive to such errors especially for some of the representation types.
Machine translation from Turkish to other Turkic languages and an imlementation for the Azeri languages
(Thesis (M.S.)- Bogazici University. Institute for Graduate Studies in Science and Engineering, 1993., 1993.) Hamzaoğlu, İlker.; Kuru, Selahattin,
Machine translation is the application of computers to the translation of texts from one natural language into another. There are three different approaches proposed for machine translation; direct translation, transfer-based translation and interlingua-based translation. Since none of these approaches are suitable for the problem of machine translation from Turkish to other Turkic languages, we propose a lexicon-based approach. As the sentence syntax is similar for Turkish and the Azeri language, chosen as a representative of Turkic languages, we do not employ any syntactic analysis. Morphological and semantic analyses are carried out for the translation. Translation can not be viewed as a word for word translation even though the source and the target languages have a similar syntactic structure. This is due to the existence of ambiguous words with multiple meanings. The subject of the ambiguity in translation from Turkish to Azeri is explained and possible ways to resolve the ambiguities are put forward. A translator from ITurkish to Azeri is implemented using the proposed approach. The results obtained from the translator show that the proposed approach is feasible for machine translation from Turkish to the Azeri language.
Performance evaluation of a frame relay based IP router to interconnect ethernet lans
(Thesis (M.S.) - Bogazici University. Institute for Graduate Studies in Science and Engineering, 1995., 1995.) Küçükateş, Roy.; Ersoy, Cem.
Many organizations have a substantial number of computers in operation. All these computers were isolated from each other. Whereas, the jobs done on these machines were strongly correlated. In order to share the computing power and special hardware as the printing machines or extensive disk storage these computers have been interconnected with Local Area Networks (LAN). For geographically distant parts of organizations and companies, the local area networks, due to electrical limitations, are impossible to use. As a solution to this problem, interconnection of LANs over a Wide Area Network (WAN) is proposed and implemented. In this thesis, a gateway is proposed as an interconnection machine. The main communication protocol of nodes in LANs is taken as Internet Protocol (IP). LANs are consisting of well known implementations of IEEE 802.3 standards: Ethernet networks. As a WAN protocol Frame Relay (FR) is proposed. FR is chosen because of its speed compared to traditional communication methods. FR is not the fastest way for interconnection but an economical and sufficient method for the problem.In this thesis, an IP router is proposed to interconnect LANs over FR service. The design is simulated and evaluated on a powerful computer network analysis tool, OPNET, implemented by using PCs and also the system's performance is measured. This thesis covers Ethernet, FR, Internet Protocol, Routers, Ethernet to FR routing problems, computer simulation results and the implemented system's performance measurement results.
Earthquake prediction using neural networks
(Thesis (M.S.) - Bogazici University. Institute for Graduate Studies in Science and Engineering, 1995., 1995.) Dimililer, Nazife.; Alpaydın, Ethem.; Ertüzün, Ayşın.
Earthquake Prediction is a mainly unsolved problem. A large number of different approaches have been tried and only a small number of attempts were fruitful. A few of these are explained briefly in this thesis. One of the most succesful earthquake prediction sytems in use today is the Canada-Nevada, CN, algorithm. It is discussed and contrasted to the neural networks implemented in the project. For this project the earthquake prediction problem is treated as a time senes prediction problem and neural networks that have been used for ordinary time senes prediction with some success have been applied to the problem. The data used was treated as a two dimensional time series with two variables; the magnitude of the present earthquake, and the time elapsed since the previous earthquake. The neural network architectures implemented were the" multilayer perceptron network with sigmoidal activation function, NADINE, and a mult,ilayer network with chaotic activation function. The results were not succesful because of the complex nature of input data and the earthquake generation process.
A virtual path routing algorithm for ATM networks based on the equivalent bandwidth concept
(Thesis (M.S.) - Bogazici University. Institute for Graduate Studies in Science and Engineering, 1998., 1998.) Bür, Kaan.; Ersoy, Cem.
The coexistence of services with different quality of service (QoS) requirements in today's networks makes the efficient use of resources a major issue. It is desirable to improve network efficiency by adaptively assigning resources to services that have different bandwidth demands. Implementing Broadband Integrated Services Digital Networks (B-ISDN) therefore requires a network control scheme that can absorb unexpected traffic fluctuations. ATM technology provides this flexibility by virtualizing network resources through the use of the VP concept. The traffic demand of new services in a B-ISDN environment may be highly bursty and difficult to predict. The implementation of the equivalent bandwidth concept provides an efficient method to estimate capacity requirements. In this study, a method for designing a VP-based ATM network is proposed. The developed heuristic algorithm uses the equivalent bandwidth concept to compute the capacity requirements of the connection requests. This way, the desired QoS defined by the cell loss probability is guaranteed. The algorithm applies VP routing and separation techniques to minimize the maximum link utilization under processing delay constraints. The quality of the solutions achieved by the heuristic design algorithm is compared to several competitors under varying network topologies and traffic conditions. The observations on the algorithm performance show that the developed method is able to facilitate an efficient use of network resources through the introduction of VPs.
A simulation model for bus based shared memory multiprocessor systems
(Thesis (M.S.) - Bogazici University. Institute for Graduate Studies in Science and Engineering, 1998., 1998.) Bekmezci, İlker.; Tosun, Oğuz.
Simulation modelling is one of the most common methods to predict the performance of multiprocessor computer systems. In this study, a new simulation model is developed and a simulator based on this model is implemented to predict the performance of the bus based shared memory multiprocessor systems. The main inputs of the simulator are representation of the architecture and workload parameters. There are several proposed algorithms to process these inputs, including path finding, mailbox location, and memory units clustering algorithms. A number of protocols are also devised and implemented in the simulator. The verification and validation studies are very important for a simulator. The verification study is realised by the help of several analytical models, which are available in the literature. The validation is realised by the performance results of the TOMP prototype. In the last part of the thesis, several sample runs are provided to analyse and compare certain architectures under various workload conditions.
Multivariate decision trees for machine learning
(Thesis (M.S.)- Bogazici University. Institute for Graduate Studies in Science and Engineering, 2000., 2000.) Yıldız, Olcay Taner.; Alpaydın, Ethem.
In this thesis, we detail and compare univariate, linear decision tree methods using a set of simulations on twenty standard data sets. For univariate decision tree methods, we have used the ID3 algorithm and for multivariate decision tree methods, we have used the CARD algorithm. For linear and nonlinear methods, we have used neural networks at each decision node. We also propose to use the LDA algorithm in constructing linear multivariate trees. Univariate decision trees at each decision node consider the value of only one feature leading to axis-aligned splits. In a linear multivariate decision tree, each decision node divides the input space into two with an arbitrary hyperplane leading to oblique splits. In a nonlinear one, a multilayer perceptron at each node divides the input space arbitrarily, at the expense of increased complexity. We propose hybrid trees where the decision node may be linear or nonlinear depending on the outcome of a statistical test on accuracy. We also propose to use linear discriminant anlysis at each decision node. Our results indicate that if the data set is small and has few classes, then a univariate technique does not overfit and can be sufficient and the univariate ID3 has better performance than multivariate linear methods. ID3 learns fast, learns simple and interpretable rules. If the variables are highly correlated, then the univariate method is not sufficient and we may resort to multivariate methods. We have shown that ID_LDA has better performance than CART in terms of accuracy, node size and very significantly in learning time. It has also smaller learning time than ID-LP and the same accuracy. ID-LDA generates smaller trees than ID3 and CART. This shows that to generate a linear multivariate tree, using ID-LDA is preferable over CART, and may be preferable over ID-LP if learning time is critical.
Identifying peptide motifs using genetic algorithms
(Thesis (M.S.) - Bogazici University. Institute for Graduate Studies in Science and Engineering, 2000., 2000.) Tanrıseven, Deniz.; Ersoy, Cem.; Sezerman, Uğur.
Finding the ligand motifs binding to the receptor molecules is crucial in vaccine and drug design, especially for the MHC-peptide problem. In this work, for determining the peptide motifs binding to specific MHC molecules, we have used regression analysis. In order to find the the optimum regression line, genetic algorithm (GA) techniques are used because in traditional regression analysis methods, you may not be able to reach the optimum solution. The optimum regression line generated by the GA also determines the factors on the MHC molecules that makes the peptide bind to these MHC molecules. The efficiency of the GA is tested by doing several tests on its different parameters, and the optimum set of parameters are determined for this problem. Results have shown that we are able to predict second position of a peptide motif with 95 per cent exact match or 100 per cent close match within one standard deviation of the predicted equation. We have divided last position's data into two parts in order to explain it with two regression lines. Predictions for the last position of the peptide motif with the first regression line resulted in 80 per cent exact match. Second regression line resulted in 75 per cent exact match.
An extended semantic analyzer for ADA'95
(Thesis (M.S.) - Bogazici University. Institue for Graduate Studies in Science and Engineering, 2001., 2001.) Karahan, Mehmet Özgür.; Özturan, Can.
An extended semantic rule is a rule which we expect a program in a particular language to obey in addition to the standard semantic rules enforced by the compiler for the language. Such extended semantic rules may be necessary to ensure that software has certain software quality attributes. In this work, an Extended Semantic Rule Set (ESRS) that contains certain semantic restrictions for Ada'95 programs is defined and the effectiveness of that rule set to increase software quality attributes of Ada codes is examined. The work done in this thesis also describes the design and implementation of a software tool, a SemantiC Analyzer (SCA) that checks the compliance of a given Ada code to the Semantic Rule Set. The effectiveness of the Semantic Analyzer is discussed on sample input and its output is analyzed for a large set of previously compiled Ada code.
Market basket analysis for data mining
(Thesis (M.S.) - Bogazici University. Institute for Graduate Studies in Science and Engineering, 2001., 2001.) Ulaş, Mehmet Aydın.; Alpaydın, Ethem.
Most of the established companies have accumulated masses of data from their customers for decades. With the e-commerce applications growing rapidly, the companies will have a significant amout of data in months not in years. Data Mining, also known as Knowledge Dicovery in Databases (KDD), is to find trends, patterns, correlations, anomalies in these databases which can help us to make accurate future decisions. Mining Association Rules is one of the main application areas of Data Mining. Given a set of customer transactions on items, the aim is to find correlations between to sales of items. We consider Association Mining in large database of customer transactions. We give an overview of the problem and explain aooroaches that have been used to attack this problem. We then give the description of the Apriori Algorithm and show results that are taken from Gima Türk A.Ş. a large Turkish supermarket chain. We also use two statistical methods: Principal Component Analysis and k-means to detect correlations between sets of items.
Incremental neural network construction algorithms for training multilayer perceptrons
(Thesis (M.S.) - Bogazici University. Institute for Graduate Studies in Science and Engineering, 2002., 2002.) Aran, Oya.; Alpaydın, Ethem.
The problem of determining the architecture of a multilayer perceptron together with the disadvantages of the standard backpropagation algorithm, directed the research towards algorithms that determine not only the weights but also the structure of the net~vork necessary for learning the data. In this work we propose two algorithms: the Constructive Algorithm using Statistical Tests (CAST), and Constructive Algorithm with Multiple Operators using Statistical Tests (MOST). The first one constructs a single hidden layer network by adding hidden nodes one by one. The algorithm checks the difference between the errors of the current and candidate networks and decides whether to select the candidate network or not by using a statistical test for comparing the accuracies of the two networks. The networks that are constructed by MOST can have more than one hidden layer. The algorithm uses node removal, addition and layer addition and determines the number of nodes in layers by heuristics. To our krowledge, MOST is the only algorithm that constructs a multilayer perceptron with multiple hidden layers with multiple units per layer. The results of the algorithms are promising and near optimal.
Web service orchestration standards
(Thesis (M.S.)-Bogazici University. Institute for Graduate Studies in Science and Engineering, 2003., 2003.) Patlak, Çiğdem.; Bingöl, Haluk.
Web services are a standards-based, loosely coupled new distributed computing architecture designed and specified to foster cross-platform program-to-program communications. Currently the Web services landscape is in an evolving state with core specifications almost mature, whereas more specific standards for complex business requirements have not been finalized. The principal goal underlying this thesis is to highlight the potential of Web service solutions as an enabling distributed computing technology by studying the role, standards and deficiencies of Web service technologies. The thesis study includes a state-of-the-art study on Web services technology and its radical changes on application innovation and development. Due to competing vendor specifications in the field of "Web Service Orchestration", a major part of the study is devoted to taking different viewpoints on existing business process languages such as BPEL4WS, WSCI and BPML so as to produce an extensive comparative study. As a general scheme for Web service adoption does not exist, the work is concluded with a financial Web service use case to investigate the feasibility of the Web services paradigm for application development in the financial services industry.

Browse

Browsing M.S. Theses by Issue Date

Results Per Page

Sort Options