The purpose of dissertation is to utilize connotative messages for enhancing image retrieval and browsing. By adopting semiotics as a theoretical tool, this study explores problems of image retrieval and proposes an image retrieval model. The semiotics approach conceptually demonstrates that: 1) a fundamental reason for the dissonance between retrieved images and user needs is representation of connotative messages, and 2) the image retrieval model which makes use of denotative index terms is able to facilitate users to browse connotatively related images effectively even when the users' needs are potentially expressed in the form of denotative query. Two experiments are performed for verifying the semiotic-based image retrieval model and evaluating the effectiveness of the model. As data sources, 5,199 records are collected from Artefacts Canada: Humanities by Canadian Heritage Information Network, and the candidate terms of connotation and denotation are extracted from Art & Architecture Thesaurus. The first experiment, by applying term association measures, verifies that the connotative messages of an image can be derived from denotative messages of the image. The second experiment reveals that the association thesaurus which is constructed based on the associations between connotation and denotation facilitates assigning connotative terms to image documents. In addition, the result of relevant judgments presents that the association thesaurus improves the relative recall of retrieved image documents as well as the relative recall of browsing sets. This study concludes that the association thesaurus indicating associations between connotation and denotation is able to improve the accessibility of the connotative messages. The results of the study are hoped to contribute to the conceptual knowledge of image retrieval by providing understandings of connotative messages within an image and to the practical design of image retrieval system by proposing an association thesaurus which can supplement the limitations of the current content-based image retrieval systems (CBIR).

The main goal of this project has been to investigate how a message queue can be used to handle error codes in log files more actively. The project has followed the Design Science Research Methodology for development and implementation of the solution. A model of the transaction system was developed and emulated in newly developed applications. Two experiments were performed, the first of which tested a longer run time with intervals between messages and the second a time measurement of how long it takes to send 20 000 messages. The first experiment showed that the message queue was able to handle all messages which gave a high throughput of 22.5 messages per second without any messages being lost. The implemented consumer application received all messages and successfully counted the number of error codes in the received data. The experiments that have been carried out have proven that a message queue can be implemented to handle error codes in log files more actively. The future work that can be performed may include an evaluation of the security of the system, comparisons of performance compared to other message queues, performing the experiments on more powerful computers and implementation of machine learning to classify the log data. / Målet med det här projektet har varit att undersöka hur en meddelandekö kan användas för att felhantera felkoder i loggfiler mer aktivt. Projektet har följt Design Science Research Methodology för utveckling och implementering av lösningen. En modell av transaktionssystemet togs fram och emulerades i nyutvecklade applikationer. Två experiment utfördes varav det första testade en längre körning med intervall mellan meddelanden och det andra en tidmätning för hur lång tid det tar att skicka 20 000 meddelanden. Det första experimentet visade att meddelandekön klarade av att hantera meddelanden som skickades över två timmar. Det andra experimentet visade att systemet tog 14 minuter och 45 sekunder att skicka och hantera alla meddelanden, vilket gav en hög genomströmning av 22.5 meddelanden per sekund utan att några meddelanden gick förlorade. Den implementerade mottagarapplikationen tog emot alla meddelanden och lyckades räkna upp antalet felkoder som presenterades i den inkomna datan. De experiment som har utförts har bevisat att en meddelandekö kan implementeras för att felhantera felkoder i loggfiler mer aktivt. De framtida arbeten som kan utföras omfattar en utvärdering av säkerheten av systemet, jämförelser av prestanda jämfört med andra meddelandeköer, utföra experimenten på kraftfullare datorer och en implementering av maskininlärning för att klassificera loggdatan.

It is not always that easy to accurately interpret a parking signs The driver is expected to keep track of what every road sign, direction, prohibition, and amendment means, both by themselves and in combination with each others In addition, the driver must also keep track of the time, date, if there is a holiday, week number, etcs This can make the driver unsure of the rules, or interpret the rules incorrectly, which can lead to hefty fnes or even a towed vehicles By developing a mobile application that can analyze a photograph of a parking sign and quickly give the driver the verdict, the interpretation process can be made easys The purpose of this study has been to examine available technology within image and text analysis and then develop a prototype of an Android application that can interpret a photograph of a parking sign and quickly give the correct verdict, with the help of said technologys The constructed prototype will be evaluated partly by user tests to evaluate the application’s usability, and partly by functionality tests to evaluate the accuracy of the analysis processs Based on the results from the tests, a conclusion was drawn that the application gave a very informative and clear verdict, which was correct most of the time, but ran into problems with certain signs and under more demanding environmental circumstancess The tests also showed that the interface was perceived as easy to understand and use, though less interaction needed from the user was desireds There is a great potential for future development of ParCam, where the focus will be on increasing the automation of the processs / Att tolka en parkeringsskylt korrekt är inte alltid så  enkelt. Föraren förväntas ha koll på vad alla vägmärken, anvisningar, förbud, och tillägg betyder, både för sig själva och i kombination med varandra. Dessutom måste föraren även ha koll på  tid, datum, ev. helgdag, veckonummer m.m. Detta kan leda till att föraren blir osäker på vad som gäller eller tolkar reglerna felaktigt, vilket kan leda till dryga böter och även bortbogserat fordon. Genom att utveckla en mobilapplikation som kan analysera ett fotografi av en parkeringsskylt och snabbt ge svar kan denna tolkningsprocess underlättas för föraren. Syftet med denna studie har varit att utforska befintliga teknologier inom bild- och textanalys och därefter konstruera en prototyp av en Android-app som med hjälp av denna teknologi samt användarens mobilkamera kunna tolka fotografier av en parkeringsskylt och snabbt ge en korrekt utvärdering. Den konstruerade prototypen kommer att utvärderas dels genom användartester för att testa applikationens användbarhet och dels genom analys av utdata för att mäta analysens träffsäkerhet. Från testerna drogs slutsatsen att applikationen gav ett väldigt tydligt och informativt svar där analysen var korrekt de allra flesta gångerna, men stötte på problem med vissa skyltar och under svårare miljöförhållanden. Testerna visade också att gränssnittet upplevdes lätt att använda, men skulle helst kräva mindre inblandning från användaren. Det finns stor utvecklingspotential för ParCam, där fokus kommer att läggas på utökad automatisering av processen.

Over the past years, many software industries have adapted the Software Product Line(SPL) as a paradigm that empowers software reuse by exploiting software similarities and managing variabil- ities to enable high-quality deliverables with a shorter time to market. Nevertheless, the lifecycle of SPL development often faces complex tasks. Creating a specific product from the product family is the main challenge. One way towards product realization is through configuration files. Still, manually creating configuration files for each product is an error-prone and time-consuming activ- ity. Therefore, this thesis proposes a variability modeling approach that shall enable an automatic generation of the configuration files for a single product. We conduct a thorough investigation on how to model variability to support automatic generation of the configuration files, introduce four essential decisions related to variability expression, features, constraints and configuration transformation, and present a number of alternative solutions to these decisions. Moreover, we identify evolution scenarios of SPL and evaluate the decisions concerning the scenarios. Finally, a validation of the approach in an industrial case study provided by Bombardier Transportation is presented.

Logs generated by the applications, devices, and servers contain information that can be used to determine the health of the system. Manual inspection of logs is important, for example during upgrades, to determine whether the upgrade and data migration were successful. However, manual testing is not reliable enough, and manual inspection of logs is tedious and time-­consuming. In this thesis, we propose to use the machine learning techniques K­means and DBSCAN to find anomaly sequences in log files. This research also investigated two different kinds of data representation techniques, feature vector representation, and IDF representation. Evaluation metrics such as F1 score, recall, and precision were used to analyze the performance of the applied machine learning algorithms. The study found that the algorithms have large differences regarding detection of anomalies, in which the algorithms performed better in finding the different kinds of anomalous sequences, rather than finding the total amount of them. The result of the study could help the user to find anomalous sequences, without manually inspecting the log file.

Cette thèse porte sur l'efficacité des algorithmes distribués de partage des ressources et des algorithmes de découvert de chemin en ligne. Dans la première partie de la thèse, nous analysons un jeu dans lequel les utilisateurs paient pour utiliser une ressource partagée. La ressource allouée à un utilisateur est directement proportionnel à son paiement. Chaque utilisateur veut minimiser son paiement en assurant une certaine qualité de service. Ce problème est modélisé comme un jeu non-coopératif de partage des ressources. A cause du manque des expressions analytiques de la discipline de file d'attente sous-jacente, nous pouvons résoudre le jeu que sous certaines hypothèses. Pour le cas général, nous développons une approximation basée sur un résultat fort trafic et nous validons la précision de l'approximation numériquement.Dans la deuxième partie, nous étudions l'efficacité des jeux de balance de charge, c'est à dire, nous comparons la perte de performance de routage non coopératif décentralisé avec un routage centralisé. Nous montrons que le PoA est une mesure très pessimiste car il est atteint que dans des cas pathologiques. Dans la plupart des scénarios, les implémentations distribués de balance de charge effectuent presque aussi bien que la mise en œuvre centralisée optimale.Dans la dernière partie de la thèse, nous analysons problème de découverte chemin optimal dans les graphes complets. En ce problème, les valeurs des arêtes sont inconnues, mais peuvent être interrogés. Pour une fonction donnée qui est appliquée à des chemins, l'objectif est de trouver un meilleur chemin de valeur à partir d'une source vers une destination donnée interrogation le plus petit nombre de bords. Nous vous proposons le rapport de requête en tant que mesure de l'efficacité des algorithmes qui permettent de résoudre ce problème. Nous prouvons une limite inférieure pour ne importe quel algorithme qui résout ce problème et nous avons proposé un algorithme avec un rapport de requête strictement inférieure à 2. / This thesis deals with the efficiency of distributed resource sharing algorithms and of online path discovery algorithms. In the first part of the thesis, we analyse a game in which users pay for using a shared resource. The allocated resource to a user is directly proportional to its payment. Each user wants to minimize its payment while ensuring a certain quality of service. This problem is modelled as a non-cooperative resource-sharing game. Due to lack of analytical expressions for the underlying queuing discipline, we are able to give the solution of the game only under some assumptions. For the general case, we develop an approximation based on a heavy-traffic result and we validate the accuracy of the approximation numerically. In the second part, we study the efficiency of load balancing games, i.e., we compare the loss in performance of noncooperative decentralized routing with a centralized routing. We show that the PoA is very pessimistic measure since it is achieved in only pathological cases. In most scenarios, distributed implementations of load-balancing perform nearly as well as the optimal centralized implementation. In the last part of the thesis, we analyse the optimal path discovery problem in complete graphs. In this problem, the values of the edges are unknown but can be queried. For a given function that is applied to paths, the goal is to find a best value path from a source to a given destination querying the least number of edges. We propose the query ratio as efficiency measure of algorithms that solve this problem. We prove a lower-bound for any algorithm that solves this problem and we proposed an algorithm with query ratio strictly less than 2.

Context: Log files are produced in most larger computer systems today which contain highly valuable information about the behavior of the system and thus they are consulted fairly often in order to analyze behavioral aspects of the system. Because of the very high number of log entries produced in some systems, it is however extremely difficult to seek out relevant information in these files. Computer-based log analysis techniques are therefore indispensable for the method of finding relevant data in log files. Objectives: The major problem is to find important events in log files. Events in the test suite such as connections error or disruption are not considered abnormal events. Rather the events which cause system interruption must be considered abnormal events. The goal is to use machine learning techniques to "learn" what an"expected" behavior of a particular test suite is. This means that the system must be able to learn to distinguish between a log file that has an anomaly, and which does not have an anomaly based on the previous sequences. Methods: Various algorithms are implemented and compared to other existing algorithms based on their performance. The algorithms are executed on a parsed set of labeled log files and are evaluated by analyzing the anomalous events contained in the log files by conducting an experiment using the algorithms. The algorithms used were Local Outlier Factor, Random Forest, and Term Frequency Inverse DocumentFrequency. We then use clustering using KMeans and PCA to gain some valuable insights from the data by observing groups of data points to find the anomalous events. Results: The results show that the Term Frequency Inverse Document Frequency method works better in finding the anomalous events in the data compared to the other two approaches after conducting an experiment which is discussed in detail. Conclusions: The results will help developers to find the anomalous events without manually looking at the log file row by row. The model provides the events which are behaving differently compared to the rest of the event in the log and that causes the system to interrupt.

Context: The volume of log files is massive as they contain vital information about the application’s behavior; they map out broad parts of the application, allowing us to understand how every component behaves, whether normally or abnormally. As a result, it is critical to examine the log files to see if the system is deviating from its usual path. Because they are so large, it is difficult for the developer to identify each and every error. So, to overcome this problem we developed a machine-learning model to detect types of errors in log files with minimal manual effort.  Objectives: The main objective is to discover errors in log files throughout the testing and production phases so that the application behaves properly. We intend to detect errors by training the module with relevant datasets and teaching the model to differentiate between the types of errors like error, debug, info, fail, etc. caused when the application is tested or operated during the production phase.  Methods: We employ machine learning techniques like SVM and multinomial naive Bayes as well as long-short-term memory (LSTM) networks, which are a sort of re-current neural network capable of learning order dependency in the prediction of sequences, which is appropriate for our use case. These techniques are used to de- termine whether errors such as assert, fail, error, and warning were generated. Then we used verdict generation machine learning techniques to generate the verdict from the error log messages.  Results: The results indicated that, instead of manually detecting errors, we can easily discover and fix them by integrating machine learning and classification methods, making it easier to move the application to production.  Conclusion: The results will assist developers in identifying the errors without having to manually examine the log file row by row. This approach has the potential to reduce the need for additional human efforts to examine log files for errors and can determine the type of error that occurred in the specific row that caused the application to diverge from its typical flow.

Thèse numérisée par la Direction des bibliothèques de l'Université de Montréal.

Thèse numérisée par la Direction des bibliothèques de l'Université de Montréal.