This report describes a case-study evaluation of two integration strategies with particular focus on performance. The study is motivated by integration challengers within a company we have cooperated with, and our wish to gain insight into systems integration. The goal of the study has been to evaluate the performance of two message-based system integration strategies. We have evaluated this by implementing several applications which are integrated using either Web services technologies or an integration technology provided by our cooperator. Our research questions have been as follows: Q1: Which integration solution has best performance in a publish-subscribe scenario? Q2: Which integration solution has best performance in a request-response scenario? The results show that the Web service applications has best performance when sending small messages (up to 160kB). For large messages, the applications based on the integration technology from the cooperating company perform better. he contributions of the study may be split in two. The contribution to the company is the performance evaluation of their technology. Collected data for response time and throughput, and performance models for our test applications are also contributed. For a broader context, we contribute with performance evaluation of Web services technologies. Data is collected for response time and throughput on test applications, and performance models are made. The comparisons of integrations based on Web services and MIP also serves as example of the performance of Web services versus other middleware.

In this thesis we propose a novel approach were 3 days of raw Yahoo! News search query logs are analyzed to find semantic relations among queries. The analysis is based on two independent contributions. The first use session data extracted from the query logs. By finding the term best describing each session, we get a vocabulary of queries related to that term. Sessions with similar terms are merged to create larger groups of queries with one common term or phrase. The second contribution is the use of temporal correlation to give a measure of frequency variation similarity. Queries that show a similar variation over time have a high chance of either being semantically related or appear in the same situations. These two contributions are then merged into related term groups, based on their session group label and the most prominent term or phrase of the correlation query. With the use of non strict parameter settings on the contribution calculations, a great number of queries are found. With the intersection of the results this leaves high accuracy groups of related queries with a term or phrase as group label. A prototype search application was developed to use the created term groups in a search environment. The groups of queries were converted into a tree structure with their group label as the main node. This navigation tree structure let the user navigate up and down in the tree or click directly on a tree node to view its results. When a users search match one of the generated groups, he or she is presented with the first search results of the trees main node together with its children.

Software design and architecture specify how a system should be implemented to achieve the required quality attributes. Being able to automatically verify the design adherence during implementation will continuously assure that the system realizes the quality attributes, as well as over time does not drift away from them. This thesis investigates how a software design can be used to automatically verify and enforce rules for implementation. The current tool support for automatic design enforcement is assessed and reviewed. In addition, a prototype contribution to this practice, a plug-in for the Maven software project management system, is presented.

Software security is an increasingly important part of software development as the risk from attackers is constantly evolving through increased exposure, threats and economic impact of security breaches. Emerging security literature describes expert knowledge such as secure development best practices. This knowledge is often not applied by software developers because they lack security awareness, security training and secure development methods and tools. Existing methods and tools require too much effort and security is often given less priority in the trade-off between functionality and security. This thesis defines a tool supported approach to secure software analysis and design. Possible vulnerabilities and their causes are identified through analysis of software specifications and designs, resulting in vulnerability cause graphs. The security modelling tool SeaMonster is extended to include security activity graphs; this technique is used with vulnerability cause graphs to model vulnerabilities and security improvement activities. A security activity graph is created to identify activities that keep the vulnerabilities from instantiating in the final software product. The activities in the security activity graph can be the use of security patterns. This way the above approach is used to find a security pattern as a countermeasure to a vulnerability, and can be used with the security pattern design templates implemented in a preliminary project. This is a way of providing coupling between security expertise and software developers to apply security knowledge in software development practice. The approach and tools are tested and demonstrated through a development case study of a medical patient journal system. The main contributions of this thesis are an approach to secure software analysis and design, an extension of the security modelling tool SeaMonster, a case study of the approach and tools that show how security can be incorporated in early stages of software development. The contributions are intended to improve availability of security knowledge, to increase security awareness and bridge the gap between software experts and software developers.

Large music collections are now more common than ever before. Yet, search technology for music is still in its infancy. Audio fingerprinting is one method that allows searching for music. In this thesis several audio fingerprinting solutions are combined into a single solution to determine if such a combination can yield better results than any of the solutions can separately. The solution is used to find duplicate music files in a personal collection. The results show that applying the weighted root-mean square (WRMS) to the problem most effectively ranked the results in a satisfying manner. It was notably better than the other approaches tried. The WRMS produced 61% more correct matches than the original FDMF solution, and 49% more correct matches than libFooID.

Reading an electronic patient record (EPR) is a very challenging task because of the medical jargons, which are almost impossible to understand for the layman. This becomes a highly relevant challenge because of the more extensive use of the internet to get medical information. Also the Norwegian laws state that the patient has the right to read his or her own EPR. A master thesis executed in 2006, and a specialization project in 2007 addressed this subject and developed a prototype for adapting EPRs to a patient presentation. This thesis continues this work and aims to extend the system with more functionality and improve the translation of the EPRs. The main issues discussed in the thesis are how disambiguating between Norwegian words and medical terms, provide summaries of EPRs, and supply the patient with external information about his or her health condition. In addition the refined user interface from the specialization project was implemented. The conclusion of this thesis is that the Support Vector Machine classifier with character bigrams provides good and accurate disambiguation between Norwegian words and medical terms. The external information functionality provides correct and quality assured information from the patient hand book. There are still some issues, and possible improvements on providing only precise and relevant articles. Summarizing of EPRs is achieved through named entity extraction of ICD codes, and then presenting the codes together with their corresponding descriptions. This implementation seems to be accurate, correct, and precise.

The advances in technology over recent years have opened up a lot of opportunities in the field of wearable health monitoring. Technology equipment that was once reserved for hospital use may now be used in the home of a consumer; this could ease the life of many long term patients and their next of kin. We have been in contact with the case of a child, who suffers from a number of rare conditions and complications. She has to be monitored almost all the time in her home by her parents. She has been in and out of the hospital a number of times, without ever figuring out what causes her problems. Her parents have to use a lot of time and effort to monitor her. Our goal in this thesis was two-fold, automate and ease the monitoring of her as well as logging all the data of vital signs so that it may later be used for diagnosing. We made a prototype system using the hardware of a wearable monitoring shirt called the LifeShirt. Our main focus was to create a system that would allow for discussion around potential usage areas of the LifeShirt. We did live testing on the patient and evaluated the solution with the family and the patient’s physician. We found that our prototype concept fulfills a need that is currently unmet. Their monitoring can be simplified, and the physician can get more data to use for diagnosing purposes. Based on our result we see a great potential for using wearable health monitoring technology in the home. We envisage many areas that could benefit from automated monitoring with the LifeShirt, both in home as well as in hospital settings.

The overall aim of this Master Thesis was to achieve a long life-time for the Miles Platform by enabling loose coupling between its various components and tools. The Miles Platform is a software development platform consisting of several interconnected tools and components. For this platform to survive future changes in technology it must be possible to replace the various components without requiring large changes to the surrounding dependent components. Based on a preceding prestudy, a layered Proof of Concept was implemented and evaluated with respect to the success of the implementation, coupling (modifiability) and amount of work. Simple prototypes involving the two first layers, namely the RESTful and the Atom Publishing Protocol based layers were implemented. The final prototype included all three layers, which involved using Semantic Web technology in addition the RESTful Atom Publishing Protocol. Only a few non-blocking issues are unresolved for this final prototype. Both the RESTful, Atom Publishing Protocol and Semantic layer contributed to maintaining a stable interface on top of the issue tracker component. By enforcing stable interfaces and serving as wrappers or intermediares between the issue trackers and their dependent components, each layer contributed to achieving looser coupling, if not in all dimensions. If the issue trackers have many dependent components that needs customization in order to communicate with the issue trackers, and the issue trackers are expected to be replaced one or preferably several times, a positive return on investment are expected from this Proof of Concept.

Ensuring robustness in software is as important as ever, with the increasing significance of information technology in our lives. Users of any IT system expect and require a high level of uptime. The earlier a threat to robustness is discovered in the development of a system, the cheaper it is to handle. Allowing robustness to come into focus at an early stage of development has been the objective of this master thesis. This thesis is a continuation of the work that we did in the autumn 2007 (Skjervold and Haga 2007), where we created a requirement specification for a tool that can aid system developers in realizing robustness during their design. This requirement specification was based upon interviews with software development companies in Trondheim, one which we performed usability testing with in this thesis. We have developed a tool based upon the requirements from our previous work, along with some additional requirements in the early phases of this thesis. After developing a first version of the tool, a usability test was performed on 11 students. The feedback we got was evaluated and the implementation was updated correspondingly. A second test was performed, focusing on both the usability and the value of the tool, with four system developers from a software development company in Trondheim. The responses to the usability were divided, but mostly positive, and helpful. Some of the changes suggested were implemented, and the rest was inserted as further work. For the value of the tool, there were strong opinions amongst the four developers, as expected, followed by a constructive discussion. The consensus was that the tool had good potential, but the professional developers felt it needed some improvements and changes. Most of these suggested improvements were too time consuming to address in this thesis, and are therefore inserted as further work. This report consists of four parts. Part I describes the state of the art and requirement specification. In Part II the experiments that was run are described and evaluated, and finally the implementation of the tool is shown in Part III. In Part IV the bibliography and appendices are shown.

The World Wide Web has become a vast and often chaotic source of all sorts of information and services. Computers have become the tool of the modern knowledge worker, and the Web constitutes both office and library. The current tools of the Web are unable to utilize its full potential, which limits the efficiency of the knowledge workers. The goal of the Semantic Web, and Semantic Web Services, is to solve this problem by introducing an evolution of the Web that is understandable for machines and humans. The objective of this master thesis was to extend the evaluation in my project on Semantic Web Services frameworks from fall 2007, by evaluating a framework implementation. The evaluation was to focus on the framework implementation's ability to realize the goals of Semantic Web Services, how easy a Semantic Web Service could be created for the given implementation, and to evaluate the available documentation and tools. The goal of the evaluation was to come up with suggestions for improvements for the framework implementation. The WSMO framework and its WSMX implementation was chosen as subject of the evaluation, based on the evaluation from the first project. The framework was evaluated by implementing a Semantic Web Service from a constructed scenario. The development of this service resulted in positive and negative experiences with the WSMX implementation, experiences that were used in the evaluation. The evaluation focused on WSMX, the available documentation, and the two tools WSMT and WSMO Studio. The results of the evaluation were suggestions for improvements for WSMX, the documentation, and the tools. By making the changes and additions that were suggested, I believe that the development of Semantic Web Services for the WSMX implementation of the WSMO framework could be made easier.