1,608
Archived Theses
0
DOIs Assigned
0%
DOI Rate
Discipline
Analyzing source codes and detecting similarities
Plagiarism in academic institutions is often expressed as copying someone else's work (i.e., another students or from sources such as books). By reason of the fact that innovations and developments are occurred in technology of late years, massive increase of software applications is monitored. Concordantly, plagiarism issue becomes more significant day by day. Plagiarism of programming source codes is an undesirable situation in the many fields of software development world. Especially in educational field, it is obviously realized that plagiarism in programming courses increases consistently. The aim of this study is attempting to answer questions such as "which codes are similar?", "what similarity ratios are?" in order to prevent plagiarism among collage students who attend programming courses. There are many methods and tools are available to find similarities between program codes. Generally traditional methods are preferable while detecting similarities among source codes. One of these traditional methods is finding metrics in software documents. However, different approaches are seen in recent years while solving plagiarisim problems. N-gram method that belongs Natural Language Process (NLP) can be given as example of different approaches. While developing the proposed methodology, metric extraction method (fingerprint system), N-gram algortihm and Vector Space Model (VSM) were considered. Information Retrieval (IR) System and Cosine Normalization (CN) methods were utilized to calculate similarity ratios. Experimental study was performed on datasets of two different kinds. First type was created by collecting assignments of students who attend programming courses in Software Engineering Department of Celal Bayar University. Second type is yielded by changing source code examples in different forms. The results obtained provide convincing evidence that the study is fit the purpose.The experimental results about proposed methods give success when compared with the previous methods.
An integrated measurement and evaluation method for software developers' performance
Up to now, several criterions have been determined in order to evaluate software developers' performance: Productivity, Engagement, Attention to Quality, Code Base Knowledge and Management, Adherence to Coding Guidelines and Techniques, Learning and Skills, Personal Responsibility and etc. However, there isn't any universally accepted methodology to measure and evaluate software developers' performance. There are three main reasons of this situation: Firstly, each part of software creation is unique. There is no compelling reason to assemble two times the same parts of software as it might be duplicated by copying it. This makes it truly difficult to make a formal and thorough correlation between two parts of software. Secondly, the current technology is something that changes at a truly fast pace. So, each time a methodology in respect to a certain wave of technology is dependable enough, it is for the most part as of recently old. Thirdly, there is a gigantic zone for innovativeness in discovering the diverse answers for a unique issue. About the thesis subject, the background has been prepared before the study of thesis by observing and analyzing the researches which have been done before and the case studies which have been published before. With this background study, the common criteria set about the measurement and evaluation of software developers' performance have been presented. Some information has been got from some software developers and managers by doing survey technic on internet so as to evaluate the use of the common criteria in real work life and identify criteria which are used in real work life but not seen in researches before. In the light of the survey results, a measurement and evaluation criteria set about the software developers' performance have been created.
A new query expansion approach using co-occurrence based semantic relationship of terms
In this thesis, we propose a new semantic-based query expansion approach, which uses co-occurrence information of words to acquire the semantics information of the terms. To do this, we first constructed word-context matrix using large English text corpus. Then we applied Singular Value Decomposition on the matrix to reduce the dimensionality. We validated the accuracy of semantic relationship on TOEFL synonym test. Our query expansion approach uses this semantic relationship to improve the existing query expansion methods available in the literature, namely Bose-Einstein and Kullback-Leibler. We evaluated our approach on Milliyet collection, which is a Turkish IR test bed containing more than 400K documents and 72 queries. The experimentation shows that our approach clearly improves the all exsiting QE methods in terms of major IR performance measures such as MAP, r-precision and recall.
Simulation and comparison of a new network protocol
The active queue management (AQM) algorithms in use today are found to be vulnerable to a type of denial of service attack called low-rate denial of service (LDoS) attack which degrades TCP connection throughput. Robust Random Early Detection (RRED), an AQ M algorithm, is designed to filter out these attacking packets before they are fed to the RED. Performance of this algorithm is evaluated and compared with other active queue management algorithm during the presence of LDoS attack. However, the analysis only focused on average throughput of TCP during the attack period. In this work, we adopted the detection and filtering part from RRED and integrated it to our Orange AQM and expanded the analysis to instantaneous and average TCP throughput; used sensitivity, specificity and accuracy measures to determine the performance of the filter. In all the cases, the result of the analysis shows that Robust Orange algorithm performs better than RRED in terms of TCP throughput; Robust Orange is more sensitive in filtering out attack packets than RRED and Robust Orange is better in receiving normal user's packet than RRED. The accuracy measure of Robust Orange is better than that of RRED.
Spatial-temporal agricultural information system to optimize pesticide usage by creating a buffer-zone
The estimation of agriculture yields is a challenging and suitable task for every farmer. In the world economy, agriculture sector plays a big role. In most countries worldwide, their populations most likely above seventy percent practice and depend on agriculture produces for feed their lives and earning incomes. Today, farmers do not only produce yields but also generate huge amounts of agricultural data. This data can be collected, stored and analyzed for useful information. In an attempt to increase agriculture produce, farmers tend to apply excessive pesticides. This has led to spray drift problem whenever liquid sprays are applied and hence causing harm to adjacent crops and other non-target areas. A lot of research has been carried done to mitigate the problem of pesticide spray drift by erosion. However, research has been carried out on pesticide drift by agents such as wind, boom height, driving speed but not to the extreme. Therefore, our research focuses on creating a buffer zone as measure to counter rampant spray drift during pesticide application. In this research, a GIS model was used to capture spray area (farm) points. The spray drift distance that was obtained using the mentioned agents above, was appended to these geographic points of the farm to create a buffer zone area. This thus can be essential for optimizing pesticide usage and better management.
Detecting - embedding data in image files
Steganography is a kind of information hiding technique that hide a data (text, sound, image?) in an appropriate carrier for example an image or an audio file. The carrier can then be sent to a receiver without anyone else knowing that it contains a hidden message. The objective is to hide the existence of the message and make it diffucult to read with some cryptographic methods. This study gives a new approach to BMP Steganography algorithm which is based on Least Significant Bit (LSB) insertion technique. Due to security option requirements some extra steps are added to LSB technique. This study accepts BMP images as a cover file and any message can be embedded into the cover file without significant changesperceived by a naked eye. Both encoding and decoding can be done using the visual interface of the tool.
Developing a field force communication system based on a third party application market study
The aim of this study is to develop a Field Force communication system which is based on a market study and its results for Actemium.All requirements are gathered with requirement analysis and used in the market research to find out all alternative solutions. Those solutions are analyzed and compared with each other to choose the best suited solution. Finally, a Field Force system is designed based on selected solution.Requirement analysis is made by organizing meetings and interviews to specify the scope and needs of the organization. To finalize requirement analysis, all requirements, including business requirements, are evaluated with the management.In-house development is chosen as best solution. Actemium has sufficient knowledge and experience to develop software solutions just like Metrack solution. Metrack is a base solution that is customized for customer requirements. The same development method is used for Field Force. Base solution is designed and it will be customized for customer requirements. Microsoft .NET Framework and its components are used in the solution such as Windows Communication Foundation (WCF), CF.NET. SQL Server 2008 and SQL Server Compact 3.5 are used as data sources in the solution.
Machine learning models for autoverification of medical laboratory test results
Understanding the job of biochemistry specialists while accepting and checking blood test results during in a day where they spend a lot of time, we decided to develop a learning system that can help them. In our experiment we used machine learning models, LibSVM and ANN for classifying the data sets. Because our experiment was started with Artificial Neural Networks (ANN) datasets were obtained from the prior experiment and that approach was tested in a view of Support Vector Machines. We used Replace Missing Values filter for cleaning up the data instances of null values, and correlation of attributes was done with Correlation Attribute Evaluator of WEKA software.
Wireless multicast streaming
This thesis covers performance analysis of Multimedia Broadcast Multicast System (MBMS) streaming delivery method on emulated UMTS environment considering Reed Solomon Forward Error Correction (FEC) algorithm, tune-in delay, rebuffering effect, MPEG-7, Electronic Service Guide (ESG).This thesis introduces MPEG-7 based ESG for mobile TV that provides a multimedia query for MBMS services and sessions, retrieves a tree view of available services and a categorized view according to the genre grouping criteria. The prototype covers OMA BCAST ESG fragments and extends content fragment of ESG by MPEG-7. The proposed ESG prototype has been developed using Visual Studio .Net 2005 Smartphone Emulator.The thesis also covers research on YouTube codecs, interfaces and whether YouTube and MBMS would work together. Another thesis research covers comparison and testing of Xenon Streamer and Darwin Streaming Server (DSS).Keywords: ESG, FEC, MBMS, MPEG-7, mobile TV, multimedia query, performance analysis, rebuffering effect, tune-in delay, UMTS emulation, wireless streaming.
Root-suffix seperation of Turkish words
Defining structure of a language and getting inferences depending on this structure supply obtaining information about that language. The applications in real world prepare an infrastructure, in which new decisions can be developed, by using the obtained information as new automatic methods in a large sample space involving a huge word source. This infrastructure should be based on a trustworthy skeleton and an integration of rules which depends on this skeleton?s accuracy.In this study, it is aimed to get the most absolute structure as possible as it can be, to prepare this structure as a module that works in high performance by not making concessions in its correctness, to make this module work with the projects implemented before and to be the base to the projects which will be done in future.The work called ?DEVELOPMENT OF A METHOD TO DETERMINE ROOT AND SUFFIXES FOR TURKISH WORDS TO GENERATE LARGE SCALE TURKISH CORPUS?, which was developed by Özlem AKTAŞ, is accepted as the base in this work. Some structures on that project, whose ratio of the correctness is approximately %99, were standardized by making some revisions. Standardization works were done by using the XML structures which were used to send information between the modules, and also used as the result of the program that was run for any morphological work and the source of the rules to separate the suffixes correctly. The suffixes are allowed to be tagged as readable and efficient for performance by being abbreviated and also named according to some predefined rules systematically.The results have formed more explicit step by step. The duration of process will be decreased by developing new modules in the future.
Messaging systems / designing and implementing messaging architectures for software applications
In software applications, as an architecture multi-tier (N-tier) architecturesgenerally are used. Multi-tier architectures isolates presentation, business logic anddata operations from each other by dividing them generally into at least three layers.Software framework is an application that is used to develop complicatedapplications and it is designed with reusable manner. It can be used for allapplications that are in the same domain. While developing a software application,developers write less code by using software frameworks so requirement analysisphase of software development phases can be much longer than implementationphase.The main goal of this thesis is researching software frameworks, applicationarchitectures ?especially N-tier application architectures?, enterprise integration patternsand implementing enterprise integration patterns to produce a re-usable softwareframework that can used in wide range application domain. While producing a softwareframework, different enterprise integration patterns are researched and used; Messagetypes, message routers, message dispatchers, Pipes and Filters are main patterns that areused to design and implement the software framework.Rota Framework is a software framework that can be used to develop a softwareapplication. Rota Framework uses specialized components and messaging patterns tocommunicate between layers of applications. It can be used with wide variety ofplatforms: Web applications, desktop application, web services and etc?
Suffix tree indexing for music information retrieval
This thesis intended for fast and reliable data retrieval from music databases. It introduces new data reduction and indexing approaches for both polyphonic and monophonic music sequences.The study contributes to the literature from three aspects. These are data reduction, suffix tree indexing and tree alignment on external memory. In terms of data reduction, we present a new melody extraction approach for polyphonic music sequences. The new melody extraction approach considers the pitch histogram, and entropy of music sequences. Consequently, accompany channels of the MIDI music sequences are determined for data reduction. In terms of indexing, we present a new suffix tree construction approach for streaming music sequences. Current suffix tree construction algorithms have leaks about indexing music sequences. Hence, we adapted the physical structure of suffix trees for music notes. At last, we consider balance and alignment of suffix trees. In music, alphabet size of music is large. Therefore, we present clustering of music sequence. Therefore each sequence cluster can be indexed by a separate suffix tree to balance the tree.Both our melody extraction and suffix tree construction approaches are tested in detail and discussed. Our evaluation metrics are based on cognition, mathematical proofs and simulations. Experimental results showed that our approaches outperforms.
Distributed computing on beowulf clusters
Building low cost Beowulf style clusters by using tens or hundreds of PCs is a popular method to achieve higher computing capacities. To gain advantages of such a computing platform, a load balancing scheme is needed for transparent distribution of loads of individual computers throughout the whole cluster in a scalable and efficient manner.In this thesis a scalable cluster architecture and a load balancing model for heterogeneous Beowulf cluster environments are presented. For scalability issues the system relies on a hierarchically centralized architecture. To have general purpose characteristics the proposed load balancing model offers some dynamic and customizable properties in its design. For this purpose, multiple user defined load indices are considered in load calculations like CPU utilization, memory usage, network bandwidth capacity, etc. along with their combinations. In addition, the load distribution policy is based on customizable adaptive load threshold values that dynamically adjust the load distribution decisions according to the system state.The thesis details the design, implementation and performance evaluations of proposed models.
A multi agent system for urban traffic control
Traffic signal control is a system for synchronizing the timing of any number of traffic signals in a target road domain. These systems automate the process of adjusting signals to optimize traffic flow by reducing stops, overall vehicle delay and thus maximizing throughput. The study presented in this thesis proposes a new intelligent traffic light control that is quickly adaptive to changing environment. The new controller focuses on urban intersections and road lanes that are incoming to this intersection. The system inputs are traffic volumes on road lanes and the outputs are continuously changing light periods for the traffic light units in target intersections. The proposed system is based on a hierarchical multi agent model and a fuzzy controller is executed through this agent hierarchy. In addition to this local traffic light control, a reasoning engine is also integrated into the system to evaluate neighbor intersection situations. The outputs of the reasoning engine are also used for updating traffic light periods. All these work has been implemented on software basis and the results are given according to some sample scenarios. The obtained results show that the proposed dynamic signalization system outperforms the fixed time-plan based controllers and generate better vehicle flows through intersections.
3G wireless multicasting service description discovery and transport
Wireless Multicasting is a technology that enables data and multimedia services to be delivered from a single source to a group of mobile receivers particularly for the actors in the broadcasting and telecommunication world. Although multicasting has been extensively researched in the past, the wired IP Multicast model has not picked up due to various limitations. The new generation wireless counterpart of this technology is receiving tremendous interest from all over the world.In this work, first we have provided a survey of recent technological improvements for wireless multicasting in both cellular and broadcast world. Then, one of the 3G wireless multicasting architecture, 3GPP?s MBMS (Multimedia Broadcast Multicast Services) in UMTS (Universal Mobile Telecommunication System) networks, is investigated with a focus on reliable download mechanism. We have provided an end to end download prototype for MBMS. Our prototype, called MBMS legacy download, also covers an implementation of a Service Discovery Architecture. As a unique contribution the thesis provides the gain of using progressive download instead of legacy download and proposes ways to increase the gain for streamable multimedia files for MBMS. With progressive download, downloadable media can be streamed earlier after some waiting time, while the downloading still continues in the background. First we provide optimizations of the parameters for an efficient MBMS legacy download. Then based on these optimizations, we provide experimental analyses to show the gain in using progressive download in MBMS. Finally in order to further increase the progressive download performance, we apply our application layer interleaving strategy to our MBMS download systems and give a performance comparison of the legacy, interleaved and progressive download delivery. This work has has been fully funded by TUBITAK and Vidiator Technology US under the project EEEAG 104E163.
Multimodal emotion recognition in video
This thesis proposes new methods to recognize emotions in video considering visual, aural, and textual modalities.In visual modality, we proposed a new facial expression recognition algorithm based on curve fitting method for frontal upright faces in still images. Proposed algorithm considers the shape of mouth region to recognize happy, sad and surprise emotions. According to our experiments, our method achieves 89% average accuracy. In addition, we proposed a skip frame based approach for video segmentation.In aural modality, we present an approach to emotion recognition of speech utterances that is based on ensembles of Support Vector Machine classifiers. In addition, we proposed a new approach for Voice Activity Detection in audio signal, and presented a new emotional dataset called Emotional Finding Nemo based on a popular animation film, Finding Nemo.In textual modality, we proposed an emotion classification method based on Vector Space Model (VSM). Experiments showed that VSM based emotion classification on short sentences can be as good as other well-known methods including Naïve Bayes, SVM, and ConceptNet on predicting emotional class of a given sentence.Finally, we use late fusion technique with a web-based interface for emotional browsing of TRECVID dataset, and we developed an emotion-aware video player to demonstrate the system performance.
Design and implementation of Turkish speech recognition engine
In this thesis, we have designed and implemented syllable based Turkish speech recognition systems based on Linear Time Alignment (LTA), Dynamic Time Warping (DTW), Artificial Neural Network (ANN), Hidden Markov Model (HMM) and Support Vector Machine (SVM). These speaker dependent and isolated word recognition systems consist of five main parts: Preprocessing, feature extraction, training, recognition and postprocessing. Preprocessing includes some operations such as speech signal smoothing, windowing and syllable end-point detection. In feature extraction, we have used speech features as mel frequency cepstral coefficients, linear predictive coefficients, parcor, cepstrum and rasta coefficients. In training stage for HMM, SVM and ANN, every syllable of the words in the dictionary is trained, and the syllable models are generated. In recognition stage, every syllable in the word utterence is compared with the syllable models. So, the recognized syllables are determined and ordered. Then, the recognized syllables are concatenated with each other. In postprocessing operation, we have developed the system which is based on Turkish syllable n-gram frequencies. The system decides whether the recognized word is Turkish or not. If the word is Turkish, then it is new recognized word.The system is middle scaled speech recognition because the system dictionary has 200 different Turkish words. After the system is tested on 2000 spoken words, we have seen that the word error rate of the system is about 5.8% for DTW, 12% for ANN, 8.8% for LTA, 17.4% for HMM and 9.2% for SVM with postprocessing. System recognition rate increased approximately 14% using postprocessing.
Simulation of wave propagation in anisotropic media
Analyzing complex structures such as electromagnetic waves requires interactive visualization techniques for better interpretation. Based on deep mathematical knowledge, model of the structure and the medium is defined with computationally expensive explicit formulas. Without using computer resources, it is difficult to make robust analyses over the model. In such cases using computers are necessary for rapid and reliable data computation and visualization. Moreover, computers are necessary for disseminating the resulting information to other researchers. To fulfill these requirements, interdisciplinary research between mathematics and computer science is needed.Considering the above requirements, in this thesis we studied the simulation of wave propagation in anisotropic media. Firstly the explicit formulas are constructed as a solution of the problem. Secondly appropriate parallel computation and visualization technique is implemented. For this purpose we used graphic card processing unit that is capable of executing hundreds of instructions in milliseconds. Using this approach makes it possible to access the computation result directly in the graphic card. By this way we eliminate the data transmission between main memory and the graphic card memory. Immediately after computation, resulting data in the graphic card?s memory is directly visualized. Finally we developed a web based prototype platform that makes it possible to share the experiment results with other scientists.
Reliable transport for wireless sensor and actor networks
Wireless Sensor and Actor Networks (WSANs) are used for monitoring the physical world, processing data, making decisions and performing appropriate actions. Reliable transport of information in these networks is necessary for the correctness of an appropriate action, for obtaining the exact picture of phenomenon and for updating the modules of sensor nodes.A scalable, energy-aware and flexible transport solution for WSANs is presented in this study. The proposed transport solution is divided into two major parts sensors-to-actors and actor-to-sensors reliable transport. In order to fulfill different reliability requirements of events, the sensors-to-actors transport is further sub-divided into different transport modes; simple, fair, prioritized and real-time.Since the sudden impulse of event information from the sensors to the actor results in congestion, a novel congestion control scheme based on packet delivery time and buffer size of nodes is also presented in this study. In order to decrease the affect of interference, a novel schedule based packet forwarding scheme is introduced at the transport layer for orderly delivery of event packets to underlying layers. The actor to sensors reliable transport is aimed to provide successful transport of all data packets from the source to sensor nodes. In this study it is shown that, the rate at which lost packets should be recovered depends on the arrangement of nodes in the network.
An intelligent agent application for buyers in electronic commerce
One of the mostly used areas of information agents is electronic commerce. At business to customer commerce which is one of the parts of electronic commerce, web pages including information about properties of products such as price, color, size, etc. enable customers to be more conscious to shop. Because of browsing thousands of web pages will be impossible for customers, intelligent information agents play essential role for collecting, processing data in Internet and producing meaningful results for customers.This thesis includes a new information extraction algorithm which is based on frequencies of sequential word groups, and design and application of an intelligent agent that acquire knowledge from Internet to automatically build a knowledge base.To evaluate the algorithm, an intelligent agent which is called FERBot (Feature Extractor Recognizer Bot) is designed. FERbot collects the labeled or unlabeled information from structured or unstructured web pages to inform customers about the features of products and value range of these features for reasonable and cost efficient shopping.This new algorithm is an alternative for the current information extraction algorithms that extract information from web. It plays an important role for text miners and natural language processors to process the extracted information.
Supervised techniques in data mining
Usage of Data Mining techniques is very common for reaching info on huge database. Techniques especially canalized by the user are used in this study. Theory of Data Mining is shortly described in first 6 chapters. Subjects are: learning and reaching info methods, Database Operational System types and selection, organizing data, removal of problems related data and presentation of obtained results.Data mining application is very common on especially commercial and medical areas. However, known application has not been encountered in earth sciences. Therefore, data is being used which obtained from Seyitömer Coal Basin in this application. When data examined: it is noted that there is no standardization for material naming. First of all, it is tried to hinder to name material in different ways at the stage of forming database. Summarized info is being represented after entering the data. Even if summary is not canalized by the user, it is added to the application because it may help to searcher. User chooses the material. Finds the first layer met for the chosen material in bore-hole. Therefore, reaches the material list takes place above this layer. Besides finds the last layer met. And obtains the material list takes place under this layer.User may wish to group some material under same name. And can re-organize the database according to this. The above described studies can be applied on this new database. This application also obtains vertical cross-section diagram drawing. At last, user can classify bore-holes according to code of layer which chosen material first met. The result of this procedure is represented on a plane by using different colored points to the researcher.Keywords : Data mining, database, Seyitömer Coal Basin, application for coal beds.
Developing an automated software defect management system and effects of system on software development process
There has always been major interest in the area of software development is ?quality?. Today, all kinds of software companies and some departments of universities develop new software?s. This products may be very successful and meet all of customer?s satisfaction. Across to this, there is impossible to develop software without error practically. Therefore, number of errors and defects can help us to identify software quality.While defects affect software quality, they must be controlled and managed. Otherwise, we cannot find the exact problems that cause a defect or an error. If we analyze the real problems on our software, we can develop a project more reliable with fewer errors and we can improve our project?s quality to better quality standards.In this approach, a new defect management system that called UHISYS was developed for software development team of a company. The main purpose of this defect management system is controlling the errors and defects on development period. After enough data that has information about the real problem, collected, an analyzing period that includes all parts of software production lifecycle starts. New solutions find on software or new software methodologies practice by development team according to these analyzes. This thesis will show us how a defect management system affects the overall product performance and product quality of a software development company.
Executing codes for IA-32 (x86) architecture on java virtual machines
Today, the IA-32 architecture is ubiquitous among PCs. Much more softwares have been developed for this architecture than for any of others. To benefit from these softwares on the platforms which use other architectures, the softwares must be ported to that platforms or must be redeveloped.A Java Virtual Machine (JVM), is a set of computer software programs that enables running Java applications. JVM isn't specific to any processor or operating system. It's implemented for various hardware and operating systems so that Java programs can run identically on all of them.Being able to execute IA-32 specific softwares on JVMs, makes those softwares available for platforms that a JVM is implemented for. This is an easy way to make use of IA-32 softwares on other platforms.There are many methods to run an IA-32 application on a JVM: the application can be redeveloped for Java, the source code of the application can be converted by automatic tools, the application binary can be executed on an emulator and others. Regardless of which method is used, software services that are needed by the application in question must be somehow supplied. Software services are provided by other softwares. These softwares can be ported to the Java platform with the methods that are mentioned before or the needed services can be supplied by using existent Java services.Generally most of the needed software services are provided by operating systems (OS). It's hard and very resource consuming to execute a complete IA-32 OS on a Java platform. It affects all of the system performance while running most of the IA-32 applications. To get rid of this bottleneck, OS services may be provided by a Java software.In this study, a Linux compatibility layer for Java was developed. This layer provides Linux operating system services under Java platform. It was shown that the Linux compatibility layer facilitates executing codes for IA-32 architecture on Java platforms.
Outlier detection with K nearest neighbor clustering
A server which serves wireless network needs strong security systems. For this aim, a new perspective to network security is won by using data mining paradigms like outlier detection, clustering and classification. This study uses k-Nearest Neighbor algorithm for both firstly clustering and then classification. K- NearestNeighbor algorithm needs data warehouse which impersonates user profiles to cluster. Therefore, requested time intervals and requested IPs with text mining are used for user profiles. Users in the network are clustered by calculating optimum k and threshold parameters of k-Nearest Neighbor algorithm with a new approach. Finally, over these clusters, new requests are separated as outlier or normal bydifferent threshold values with different priority weight values and average similarities with different priority weight values.
Turkish language characteristics and author identification
Models of natural languages and language characteristics are widely used in many computer science applications such as data security, language identification, spell checking, data compression, authorship attribution and speech recognition. In the scope of this study, a large scale corpus is created and used to discover language characteristics of Turkish. Word and letter based analyses are made on this corpus to build a base for several NLP studies.In the next step of the study, we used two different methods based on word n-grams to identify author of an anonymous text. For 16 authors, training and test set articles are collected, and mentioned two methods are applied on these article sets. Finally, obtained results are compared and most successful method is determined.
An incremental genetic algorithm and neural network for classification and sensitivity analysis of their parameters
This study proposes classification by using algorithms that are inspired by computation in biological systems and to compare of them. These are genetic algorithm and neural network. New incremental genetic algorithm and new incremental neural network algorithm are developed for classification for efficiently handling new transactions. To achieve incremental classification, a specific model that includes all information about a train operation, rules for each class for genetic algorithm and weight values for neural network is created after each training operation. Later, these models are used for testing, correctness test, comparing models and incremental classification. With that new incremental method, training time gets smaller for new dataset. Experimental results proof that assumption. This paper introduces that new method and importance of that method. This study also includes the sensitivity analysis of the incremental and traditional genetic algorithm parameters and neural network parameters. In this analysis, many specific models were created using the same training dataset but with different parameter values, and then the performances of the models were compared. To achieve these operations two tools are developed for both genetic algorithm and neural network and all of these investigations are done by using these tools.
Archaeological imaging and visualization
The main goal of this thesis is to develop a sophisticated visualization application that can be better used by geophysicists, geologists and archaeologists to explore and understand the spatial visualization of data collected from the historical site. Application has functionalities such as rotating, zooming and cutting the 3D visualization of data from the historical site. Additionally, it is also possible to make some geophysical values transparent for getting a better understanding about the historical site. Application also has capability such as taking snop-shot and recording video abilities.Visualization process can be repeated by using different interpolation and colormap settings. Maximum geophysical data limit value can be defined and changed in application, so users can adjust the insignificant data values and get start the visualization process. System was developed by using Java programming language in Eclipse Platform and VTK (Visualization ToolKit) is used to generate three dimensional graphics.
Computer aided lipreading training tool
Worldwide oral language education is becoming widespread for hearing impaired children. Nowadays hearing loss in babies can be diagnosed with newborn hearing screening tests and cochlear implantation can be performed when required. Implanted children can not start to hear and talk all by themselves as it is in normal hearing children. After implantation, every child should receive adequate speech training sessions. But, this special supplemental education is tiring work for both children and educator. A computer aided instructional tool may help to lessen the hardship and struggle that children and educators experience during the educational sessions.In these respect, AURIS is designed as a computer aided instructional tool combining both visual and audio technology for assisting teachers of hearing impaired in their practice. AURIS is developed to improve verbal communication skills of hearing impaired children. AURIS stands on residual hearing of hearing impaired children. It is necessary to emphasize that AURIS which is introduced in this thesis is only a prototype.Additionally, the result of the effectiveness of the AURIS which has been tested on preschool hearing impaired children (ages 2-6) as a case study, is presented in this thesis. These results show that using AURIS in hearing impaired education, children can learn concepts and nouns quickly and easily with the help of moving images and visual effects. At the end of the sessions, it is observed that children could even comprehend three word sentences and improve their speaking and pronunciation skills with the help of voice analysis methods inserted in AURIS.
An infrastructure model for collecting electronic data to develop large scale corpus
In the Dokuz Eylül University Computer Engineering Department, different studies on Natural Language Processing (NLP) have been carried out. For NLP research grammatical rules of the language must be determined and a text sample of that language, which is called as corpus, must be prepared. These sample texts should satisfy the grammar rules of language.In this study, an infrastructure for a large scale corpus is designed and implemented. A database model, which supports 6 different document type such as newspaper, report, magazine, book, parliamentary report and official gazette, is designed.By implementing the developed application depending on the database model, 195256 articles were downloaded from 5 newspapers, and their metadata was stored for future use.
Personal data protection in Turkey: An information technology framework indented for privacy risk management
In this study, it is shown that, technology originated threats on privacy can also be avoided by privacy enhancing technologies with a risk management approach.
Embedded firewall software package
Firewalls became an indispensable part of computer networks. Corporations buycommercial firewalls or use open source firewall software packages to meet their need.Most of these are Linux or FreeBSD based and few of them are OpenBSD based, whichhas few features or are command line tools. In this thesis, creating an OpenBSD basedfirewall software package, which has a graphical user interface, an extensible infrastructureand targets embedded systems, has been aimed.In this thesis, firstly firewall has been described and then previous studies about firewallsoftware packages have been discussed. Since this study aims OpenBSD operating systemand embedded hardware, the basics of OpenBSD and the design decisions for an embeddedsystem have been discussed. Finally the infrastructure of the implementation of this studyhas been explained.
Extraction of named entities from Turkish document collections
This thesis aims to develop a model improving Hidden Markov Model (HMM) and Conditional Random Field (CRF), which are two common sequence classifier techniques, for Named Entity Recognition (NER) task on Turkish documents. So, we first examined for the best values of parameters used as input in these models. In HMM, we represented each token with multi features. Next, we used CRF model to determine most effective parameters values that are used as input in this model such as window size, output encoding format and features extracted from tokens. After detailed examination of both HMM and CRF models, we applied a linear-chain CRF model, for NER in Turkish documents. Besides, we proposed 41 different features in four categories: rule based, lexical, dictionary lookup and morphological based features. First, we performed a set of experiments using this feature set on publically available NER datasets. We achieved the best performance with a linear-chain CRF model using [-3, +3] as a window size, BIO encoding as an output encoding format and extended feature set. In terms of F1 measure, we obtained the 91.83 percent, 91.2 and 88.62 for person names, location names and organization names respectively. Furthermore, this thesis also presents METU-NER corpus, which is based on annotation METU corpus for NER. We evaluated our a linear-chain CRF model with the same parameters used in the previous dataset. In terms of F1-measure, we achieved 73.26 percent, 70.12, 63.83, 63.83 and 69.14 for person, location, organization, temporal names and overall, respectively.
Survival time prediction of cancer patients
In recent years, in order to reduce noise in experimental data and to add the common role of genes in biological processes into diagnostic and prognostic prediction models, researchers entegrates more than one data type. In this context, many studies have shown that protein interaction networks increase the success of scientific diagnosis. This study aims to find biomarkers that successfully predict the potential survival time of cancer patiens by merging gene transcriptome and protein level data belonging to kidney renal clear cell carcinoma (KIRC) and glioblastoma multiforme (GBM). For this purpose, expression level of mRNA (RNA-seq) and protein (RPPA) data entegrated a with network modelling protein interactions in the human genome. Survival time of patients will be predicted by selecting certain amount of biomarkers and feeding those as inputs to the supervied learning method. For both cancer types, this study showed that our new entegrated method, RPBioNet, outperforms both "only protein" and "only mRNA" methods.
Developing process mining algorithms for finding meaningful patterns
Process mining is a technique for extracting knowledge from event logs recorded by an information system. In the process discovery phase of process mining, a process model is constructed to represent the business processes systematically and to give a general opinion about the progressive of processes in the event log. Considering in advance the trend and different features of running process is important. Especially, time management is crucial in designing and conducting business processes. Every day information systems collect different kind of process instances of a business flow. As time goes on, size of collected data builds up speedily and constitutes a huge volume of data. It is a very challenging task to obtain valuable information and features of processes from such a large volume of data. This thesis proposes a novel algorithm, Interactive Process Miner (IPM), to create process model based on event logs and, also a new approach that contains three different features; including activity deletion, aggregation and addition operations on the existing process model. The proposed algorithm, IPM, is enhanced by introducing time perspective. Time-oriented IPM algorithm, T-IPM, is capable of predicting the remaining and completion time of each process in a flow. This thesis also includes the development of a new process mining tool, ProLab, in order to work on large volume of event logs and to handle the execution records of running process instances. Experimental studies demonstrate the capability of IPM and T-IPM algorithms and, also ProLab tool on both real-life and experimental datasets, including low memory usage, modification opportunity and improvement in performance compared to the existing algorithms.
Modified stacking ensemble machine learning method for network intrusion detection
Machine learning (ML) methods became highly popular since the amount of the data produced on the Internet started increasing exponentially, although it was a known topic in academic studies before that era. It started becoming highly hard to extract rules from this huge amount of data or find patterns. ML started playing an important role in the area of finding patterns and extracting rules. With increasing number of people accessing the Internet and having smart mobile phones, the possible threats in the Internet became important topic in network security studies. The conventional way of detection network intrusion is to use signature based rules (pre-defined rules), where this type of rules can't detect unknown signatures even though the type of the attack is the same. In last decade, ML started to be used in network security studies more often. This study is based on using stacking ensemble machine learning method for the purpose of detecting network intrusion. In this study, we propose two different methods to improve the performance of ML methods to detect network intrusion. Firstly, we used different combination algorithms and different base model selection methods to improve the performance of stacking ensemble method which provided significant results when it is compared to conventional machine learning methods. Secondly, we used genetic algorithm to for the base model selection phase of the first study.
Enhancing authentication in radio frequency identification systems by designing a fully fledged class protocol
The main topic of this thesis is the security problems of Radio Frequency Identification (RFID) system, one of the most important technologies of recent years. Despite the numerous advantages of RFID technology, it faces a major security and privacy threats. Because of the wireless communication nature between reader and tag, RFID technology is vulnerable to many attacks. Since, authentication adds trust to the identifing process, authentication protocols are first step in protection against wireless attacks and only the authenticated reader can access the contents of the authenticated tags. As a first step in this thesis, a class related comprehensive survey, review and comparison of the most recent and considerable RFID mutual authentication protocols are made in detail. The significant points of the compare are presented in two tables. The outcome of the comparison revealed that the investigated authentication protocols have adopted various methods to deal with attacks and ensure security and privacy. Due to hardware restriction of the low cost tags, most of the investigated authentication protocols are lie under fully fledged class. Besides, every examined protocols has a particular capability to handle the security and privacy issues. Secondly, in this thesis an efficient and powerful RFID mutual authentication protocol is proposed named AERMAP-W5. AERMAP-W5 uses both private and public key algorithms, AES and ECC. Two methods are used during the authentication process. Dissimilar the existing schemes, AERMAP-W5 is coded, tested and proven on real devices and could send tag ID and valuable data as well. Moreover, in AERMAP-W5, mutual authentication has been realized in only 2 steps. Finally, the security and performance of AERMAP-W5 is thoroughly analyzed and the results show that it can stand out against almost all common attacks and satisfies the essential security requirements of RFID-based healthcare systems.
An analysis of pesticide use for cotton production through data mining:the case of Nazilli
Data mining involves certain methods of obtaining or inferring meaningful information unknown within the data. Besides of the fact that it is used in the fields of health, marketing, banking; it is also utilized in agriculture. With the increasing significance of precision agriculture practices, farmers have become inclined to be engaged in a more conscious agriculture. Farmers use pesticides to destroy a disease or hazard on plants. Nevertheless, in some researches, it has been revealed that pesticides have harmful effects on human health, environment and plant. Although many farmers are aware of the risks of excessive use of agricultural pesticides, they still use them to get a faster yield and avoid financial loss. The goal of this study is to create a model for cotton growers that will indicate the optimum amount of pesticides to apply for the maximum yield rate. The ultimate aim is the modeling decision tree based classification algorithms on the data and the observation of the results. The cotton planted fields and used pesticides data received from Aydın Nazilli District Directorate of Agriculture was organized, it was evaluated with the help of classification algorithms in SPSS Clementine software. C5.0, Classification And Regression Tree (C&RT) and Chaid decision tree algorithms that are the most preferred data mining methods are employed in this study. Thereby, it was observed that there exist some certain suggestive differences between the fertility obtained from the product and the pesticide used.
Agricultural decision support system using data mining for farmers
The estimation of agricultural yield is a challenging and essential task for every farmer. Since the very old times, agriculture has always been the most important means of livelihood both in Turkey and all around the world. There are many factors that directly affect the efficiency in agriculture such as climatic features, use of water resources, proper and timely use of pesticides and fertilizers. Computer-based systems are needed to transform agriculture data into tangible information. Data mining involves certain methods of obtaining or inferring meaningful and otherwise-unknown information from the data. With the increasing significance of precision agricultural practices, farmers have become inclined to be engaged in a more conscious strategy of agriculture. In this study, barley crop data received from İzmir Menemen Provincial Directorate of Agriculture was carefully organized and evaluated with the classification algorithms in the SPSS Clementine software. CHAID and CR&T algorithms were employed and major factors that affect crop yield were defined. Based on these, a decision support system has been developed for farmers to forecast both harvest season and crop yield.
Object oriented application frameworks compare and select the appropriate design technique
Object oriented frameworks are defined in many ways. The most popular definition: ?aframework is a partial design and implementation from an application in a given domain?[Bosch]. In my opinion frameworks are a set of abstract and concrate classes that togethercomprise a generic solution to similar problems in a specific domain. The core of theframework is made up of abstract classes.Object-oriented frameworks have been used since the early eighties and now they arebecaming increasingly popular. They provide software developers with the means to build aninfrastructure for their applications. Also they decrease the time of developing application. Agood framework has several properties such as ease of use, extensibility, flexibility, andcompleteness, which can help to make it more reusable.The aim of this study is to examine the details of the frameworks and their designtechniques. Therefore, I studied basic concepts related with frameworks, design techniquesused for frameworks recently and selected an object-oriented technique, which is the mostpowerful technique in developing framework. Some of the frameworks have been chosen tocompare because of the large number of different applications. These frameworks are ACE(Adaptive Communication Enviroment), MET++ (Multimedia Application Framework) andSMA (State Maneger Interface). In addition, more general framework .NET Framework isalso selected to be examined. As a result, the most appropriate technique from inside of thesetechniques is suggested for developing object oriented application frameworks. Also selectedframeworks are compared.
Formal methods and programming tools for modeling ant colonies
Nature inspired algorithms have growing up interest in the area of optimization,and the class of ant colony optimization algorithms is one of the recently developedinstances of such algorithms. Ant Colony Optimization, ACO, algorithms rely on thebasic behavior of ants that is known as foraging behavior, which helps the antcolonies to find the shortest path among a number of possible choices. This isachieved by laying down a chemical substance, called pheromone, on the groundwhile moving; and preference of the paths with high pheromone level by thesuccessor ants. This type of behavior is also called social behavior in more abstractlevel, and covers many biological phenomena.There are two directions in dealing with the social behavior observed in antcolonies; one is transferring the idea to solve optimization problems, leading to theant colony algorithms, and the other one is formal modeling of the behavior followedwith a proper verification schema for better understanding of the relationshipbetween the local interactions of individuals in colonies, and the global dynamicalbehavior of the colony. Through formal modeling, and a proper verification approachnot only social behavior, but various aspects of ant behavior can be investigated.In this thesis, we have followed both directions. There are some applicationsdeveloped employing ACO algorithms for solving a real world problem, and aproblem from operations research area. In addition, the application developed forTraveling Salesman Problem serves for better understanding of the algorithm.However, much of the efforts have been spent for formal modeling, verification, anddeveloping an automated modeling tool.Among a variety of formal modeling languages, Weighted Synchronized Calculusof Communicating Systems, WSCCS, has been chosen which is a probabilistic statebased transition process algebra. However, modeling itself brings no insight unless itis combined with a verification schema. Verification aims to confirm the correctnessof the abstract model against its specification, and also to bring front the properties ofcolony being studied via asking some questions to the model. Model checking is atechnique of verification, and concerns to verify the model for a given property.In order to verify the correctness of the model, model checking approach has beenemployed. Since model checking can be performed via temporal logics, probabilisticComputation Tree Logic is another issue dealt with which is then extended to be ableto cover the notion of action.Combining the model checking and formal modeling by WSCCS can beaccomplished through transforming the model into a discrete state space withcorresponding transitions. Therefore, Labeled Kripke Transition Systems (LKTS) isanother formalism introduced, and extended to wrap the probability and action in itsstate transitions.The main achievements addressed in the thesis are: the ACO applicationsdeveloped to solve optimization problems, an investigation of WSCCS for modelingant colonies, extending the CTL and LKTS such that both systems allow representingprobabilistic action occurrences which is the most important property of WSCCS,designing a model checking schema that permits to query the model for probabilisticaction occurrences, and implementing a tool in order to automate the whole process.Keywords: Ant Colony Optimization (ACO), social behavior, WeightedSynchronized Calculus of Communicating Systems (WSCCS), model checking,probabilistic computation tree logic (PCTL), labeled Kripke transition systems(LKTS).
Java card based security application module
The amount of data we have is getting higher along with the growing technology.Internet and other digital media make it possible to share data faster and easier, as amanner of this, it is needed to protect some data. There are wide variety of softwareon the market specialized on data protection, but it is proven by research that themost powerful technique is using an external hardware module.The smart card is one of the most feasible hardware which can be used as suchmodule. Java cards would be the best choice, because of the ability of applicationdevelopment in it and also it is possible to get engineering samples with smallamounts.This study gives information about infrastructure of such a personal securityapplication module.
Developing a new methodology for software projects
The research presented in this thesis is an essay of a new software developmentmethodology.This methodology is prepared according to the agile manifesto and also accepts theaccreditation obligation in market. Because of this obligation, the overall system iscontrolled also for Capability Maturity Model via its checklist. Another property of thismanagement system for a software development process is that the system also advisesa natural improvement path for the company from a chaotic work-flow to a disciplinedand controlled system. This thesis includes the entire acceptance and the usage manualof the new methodology. So, the developers, who will try to apply this methodology intheir companies, will find a complete guide for a successful implementation withprocess model, role definitions and documentation requirements.
An agent design and implementation using XML and RDF technology
Firms basically meet their software requirements in two forms. First one is tomeet the software requirements as the need of software solution arises. Second one isto meet requirements by getting an overall solution proposing software in one time.First method usually causes a need of integrations among software systems. Secondmethod causes frequent customizations as the needs of firm changes.This study proposes a solution for integrations and customizations of softwaresystems. It aims to present a solution model for bettering the inter business processes.The model is based on the agent paradigm. As to this model, basically the interbusiness actors are determined. Agents, representing these actors, are created.Proposed model does not present an overall solution. The model wraps the presentapplications in firms. It aims to increase the benefits of the present systems and theirrightly usages according to the evolving needs.In this direction this model is implemented for a firm in production sector. Thesystem getting developed for the firm aims to speed up the production planningprocesses.Keywords: agent, multi-agent system, agent based workflow management system,agent based process management system.
Intelligent tutoring systems for education
Intelligent Tutoring Systems (ITS) provide the benefits of one-on-one instruction usingcomputers. ITSs can get up the performance of the students overcoming space, time,socioeconomic and environmental restrictions, according to lots of researchers. In this sense, theprogress of researches to developed more effective programs to test and enhance the learningperformance of students continues rapidly.In this thesis, which aims to examine some characteristic properties of ITSs, an intelligenttutoring system for mathematics education at undergraduate and graduate level, has beendeveloped. The developed system, which is called as MathITS, is based on conceptual mapmodeling. Hence, the thesis focuses on student modeling of system principally. Knowledgerepresentation in the system is based on LaTeX notation to represent the mathematical symbolsand notation easily. The Mathematica Kernel is used as an expert system in MathITS.
Morphological analysis in natural language processing for Turkish language and a new approach for lexicon design
The main motivation of this thesis is creating a system which can get text as input and build a knowledge structure of it. Thus the system will be able to simulate human beings? knowledge base. Additionally the system will be able to answer questions on the knowledge base. The system is designed as a shell and has user interfaces for maintaining and extending lexicon and morphological rules. Thus the system is easily extendible and in one point of view, it is language independent. A linguist capable of supplying all information on the language can develop a system specific for that language. Note that; as the design of the system is done on an agglutinative language, some extensions on the software may be needed for the execution of grammar rules on different types of languages.
Design and implementation of TV set software and hardware to solve technical problems of the set using user interface menu
Using a Graphic User Interface(GUI) we can communicate with the TV Set electronic board through infrared remote controller. Writing to and reading from the registers of IC?s on the electronic board of the TV set, we can do the remote diagnostic of technical problems, we can try possible solutions and improve software, which is embedded in the microcontroller of the board. This thesis fulfills this goal. GUI layer is a brunch of the complex software structure of embedded TV-set. What we have written for this thesis is actually an addition to this software, which is around 0,5 kB in binary, executable format. The whole software is 128 kB and resides in the rom area of SDA555 microcontroller. Our implementation covers another user menu,which we have created for the purpose of thesis and connections to the main software, which makes the menu work compatible with the main software. This project is mainly aimed to be used at TV television sets. This is not a restriction, but we recommend it being used in this market with a strict selftrust, since this area is our profession in industry, as an embedded software development engineer. We thereby think that this project can be used at any consumer electronic device, produced on line in high volumes.
Comparison of 3D segmentation algorithms for medical imaging
In this thesis we implemented four different 3D segmentation algoritms, and we compared their reults on three different CT Data Sets. These segmentation algoritms are; Seeded Region Growing, Volumetric Segmentation Using Weibull E-SD Fields , Automatic Multilevel Thresholding by using OTSU Method and Unseeded Region Growing. The main results gained from our application as follows; Seeded Region Growing Algorithm produced good result on unnoised datasets with suitable threshold value. Volumetric Segmentation Using Weibull E-SD Fields Algorithm produced good result on our sample dataset which has high amount of contrast difference.However, The results on medical datasets which has low ammont of contrast difference. Automatic Multilevel Thresholding by using OTSU Method Algorithm which takes the segment count as an input by user interaction, produced sufficient results. And lastly, the number of segments produced by the Unseeded Region Growing Algorithm are over the expectations. But they can be considered as sufficent.
Research and implementation of unified smart adaptive remote control protocol for consumer electronic equipments
Continuously growing technology and changes in customer needs has brought together a vast increase in the number of consumer electronic devices. With each new device, new features are introduced in parallel to this growing technology. New features mean more buttons on the devices? remote controllers. Remote controllers were unable to show the same technological improvements as the new technology consumer electronic devices. Remote controller concepts in use today enable us to design only controllers with fixed number of buttons. Nevertheless some remote controllers that exist in the market today can define their own buttons and even macros by their properties or by aid of the computers. But all remote controllers basically control the devices with single directional communication by sending key codes with the help of protocols they use. To realize the same technological development achieved in consumer electronic devices a new concept is needed for their remote controllers. This approach is based on the principle of changing the number of keys on the remote controller that can be used according to the state of the controlled device. The aim of this study is to design a new protocol in order to develop more flexible applications and provide a framework for new kind of applications. In this approach bidirectional communication is targeted. The main aim in this thesis is making the communication is bidirectional. By doing this the number of key can be reduced or limited. Thus users will deal with smaller number of keys consequently the system will be less likely to crash on any given condition as the user will not be able to press unsuitable keys.
Modeling and analyzing marine data using data mining techniques
The research presented in this thesis is an interdisciplinary work that combines computer science and marine science. It provides new computer based approaches, techniques and technologies for (i) modeling, collecting, archiving marine data, (ii) analyzing and mining marine data by using data mining techniques and (iii) visualizing marine data. It presents my efforts on the collecting physical, biological, chemical marine data, some explanations about the visualization of marine data on the map, my works on the construction of decision trees to classify physical marine data. This thesis introduces two new data mining algorithms: one is for clustering spatio-temporal data and the other is for spatio-temporal outlier detection in data warehouses. It also proposes a new approach: web service-based parallel clustering which includes the parallel execution of web services for discovering clusters in large data warehouses. In addition to new clustering algorithm, this thesis also presents the validation and evaluation of the clustering results of this clustering algorithm. It shows the mathematical quality and reliability of the clustering results by using a cluster validation technique. It also presents the sensitivity analysis of the new clustering algorithm to the input parameters.
Paperless office automation
In this study, a system has been designed and implemented that allows to serve PDF forms over a web site, enable users to fill forms and send form data over internet. The system gives advantage of saving and serving forms in PDF, makes office system paperless, and also allows users to digitally sign PDF documents and signature control. On the server side Windows 2000 server installed with Active Directory, IIS (Internet Information Server), Certificate Server (Root CA) used to create and serve digital signatures. PHP (Hypertext Preprocessor) used to develop a web site which consists of administrator side and user side. Data is stored in a MySQL database. PDF templates stored in the web directory. Adobe Acrobat 7 Professional used to create PDF forms and to convert existing forms in other formats to PDF. php-fdf functions used to manage PDF forms data. Keywords: Paperless Office Automation, Portable Document Format (PDF), Fillable PDF, Forms Data Format (FDF), Digital Signature, internet based paperless office system