Hide menu

Examensarbeten och uppsatser / Final Theses

Framläggningar på IDA / Presentations at IDA


Se även framläggningar annonserade hos ISY och ITN i Norrköping / See also presentations announced at ISY and ITN in Norrköping (in Swedish)

If nothing is stated about the presentation language then the presentation is in Swedish.


WExUpp - kommande framläggningar
2017-05-29 - ADIT
Creating a Database and a User Interface for Storing Test Data for Radar Equipment
Robert Habrman
Grundnivå (16hp)
kl 08:15, Muhammad al-Khwarizmi (In English)
[Abstract]
Försvarets materielverk- test och evaluering (FMV T&E) are using an Excel application to store test data for radar equipment. This storing solution has some flaws as for accessibility, the way that the data is handled, and the way that the data is displayed. FMV T&E would like a new solution of displaying, storing, and accessing the data using a web browser.
This thesis will therefore describe how a database and its user interface can be designed and implemented. This thesis also describes how this can be done in the best way regarding usability. The new solution can be accessed from anywhere due to the fact that it is going to be placed on a server which then can be accessed from a web browser. If the design of the user interface and the structure of the database turns out to work as intended, then the new solution will be able to exclude the flaws that the current solution hold.
2017-05-30 - AIICS
Designing an Artificial Neural Network for state evaluation in Arimaa
Simon Keisala
Grundnivå (16hp)
kl 10:15, Alan Turing (På svenska)
2017-05-31 - HCS
Sound as feedback on motion in active games - A sound preference experiment
Janne Väisänen
Grundnivå (16hp)
kl 10:15, Alan Turing (In English)
[Abstract]
This thesis is a pilot study on sound as feedback to motion
in active video games. The subject addressed herein has been
approached from a multidisciplinary perspective drawing from
several fields as audio, film and computer science. The study
was set up as a user experience study. Participants where exposed
to two videos of identical gameplay, one with feedback sound
turned on and a second with feedback sound turned off. Each
video was followed by a questionnaire and the study was ended
by a semi-structured interview. The main aim was to answer
whether users preferred gameplay with feedback sound turned on or
off. Results of main question were slightly in favour of sound
scenario with feedback sound off even thou not statistically
significant. The study also highlights several issues with
implemented feedback sound (a secondary goal of the study). Main
finding here was problem with perceiving sounds intended function.
Even thou results can be somewhat discouraging one should see this
study in the light of it being a pilot study.The main value and
contribution of this study can be found in lessons learned for
future implementations
2017-05-31 - HCS
Algoritmer för semantisk dependensparsning baserade på dynamisk programmering
Nils Axelsson
Avancerad (30hp)
kl 13:15, Alan Turing (På svenska)
[Abstract]
Dependency parsing can be a useful tool to allow computers to parse text. In 2015,
Kuhlmann and Jonsson proposed a logical deduction system that parsed to non-crossing
dependency graphs with an asymptotic time complexity of O(n^3) , where "n" is the length
of the sentence to parse. This thesis extends the deduction system by Kuhlmann and
Jonsson; the extended deduction system introduces certain crossing edges, while main-
taining an asymptotic time complexity of O(n^4). In order to extend the deduction sys-
tem by Kuhlmann and Jonsson, fifteen logical item types are added to the five proposed
by Kuhlmann and Jonsson. These item types allow the deduction system to introduce
crossing edges while acyclicity can be guaranteed. The number of inference rules in the
deduction system is increased from the 19 proposed by Kuhlmann and Jonsson to 172,
mainly because of the larger number of combinations of the 20 item types.
The results are a modest increase in coverage on test data (by roughly 10% absolutely, i.e.
approx. 70% -> 80%), and a comparable placement to that of Kuhlmann and Jonsson by
the SemEval 2015 task 18 metrics. By the method employed to introduce crossing edges,
derivational uniqueness is impossible to maintain. It is hard to define the graph class
to which the extended algorithm, TNC, parses, and it is therefore empirically compared
to 1-endpoint crossing and graphs with a page number of two or less, compared to which it
achieves lower coverage on test data. The TNC graph class is not limited by page number
or crossings. The takeaway of the thesis is that extending a very minimal deduction system is not
necessarily the best approach, and that it may be better to start off with a strong idea of to
which graph class the extended algorithm should parse. Additionally, several alternative
ways of extending Kuhlmann and Jonsson are proposed.
2017-06-01 - ADIT
An Approach to Achieve DBMS Vendor Independence for Ides AB's Platform
Niklas Blomqvist, Philip Johansson
Grundnivå (16hp)
kl 10:15, Muhammad al-Khwarizmi (In English)
[Abstract]
Software that’s developed with few integration capabilities to different user interface or database vendors might lose market share in the long run. To stay competitive, companies that find themselves in situations like this might need to look at options to increase their alternatives.
This thesis aims to present and evaluate how Ides AB could achieve vendor independence as it relates to database integration.
The proposed solution is based on pre-existing code for a different product and thus includes theory about the methods and details how one can read, understand and analyse the code.
The outcome is presented with code examples to give the reader a clear and concise understanding. In the evaluation phase we take other related work into consideration as it relates to our thesis focus.
2017-06-05 - AIICS
Software developers' performance awareness
Henrik Phung
Avancerad (30hp)
kl 09:00, Alan Turing (In English)
[Abstract]
Automated tests and non-functional requirements are 2 widely used terms in the software development sector. Both are essential for software development teams but rarely mentioned together.

Today, most software development teams are utilizing the development practice continuous integration. A method where software are built in iterations and in each iteration small chunks of code are merged into the main repository. Continuous integration requires automated tests to verify that each chunk of code are compatible with the main chunk. Automated test are essential for continuous integration to detect anomalies in each chunk of code.

Customer satisfaction is a result of how well the delivered product performs in terms of non-functional requirements. Although the term “non-functional requirement” has not been formally defined and the existing definitions are diverse.

In this thesis, we define the non-functional requirement, response time with help from a user-centered evaluation of responsiveness study. We create a testsuite that can be ran on an automated build with focus on user-action-response. Based on the test result and a conducted survey, we evaluate how aware developers are when it comes to causes to performance issues.
2017-06-05 - AIICS
Clustered Data Management in Virtual Docker Networks Spanning Geo-Redundant Data Centers
Hayder Alansari
Avancerad (30hp)
kl 10:30, Alan Turing (In English)
[Abstract]
Containerization in general and Docker in particular is potentially changing how we develop and operate software at its core. Not only can we build and package the container with the application and its dependencies in a local development environment but also define and test virtual software-defined networks. In these networks, clusters of containers deliver a high-availability service and ultimately deploy it all into a production environment as is.
Docker is the world’s leading software container technology. One of the main advantages of using Docker is eliminating the problem of the same program not working on a different system due to different software environment, missing dependencies, etc. This is especially the case when moving from the development environment to the production environment. Containers technologies solve this issue by encapsulating the application its dependencies into a container.
The idea behind software containers is based on creating virtual isolation. However, in many cases these containers that reside on the same host or multi-hosts need to communicate together. Therefore, Docker provides different types of networking modes to choose from. Although Docker seems to be a promising technology, nothing comes without a cost. In this thesis, we study the effect of different Docker networking modes on various performance aspects. This is done by measuring different performance metrics natively (without using Docker) and in each Docker networking mode. Then, these measurements are analyzed and compared to find the cost of using Docker containers in general and the cost of using each Docker networking mode.
2017-06-05 - AIICS
Muting pattern strategy for positioning in cellular networks
Elena Moral Lopez
Avancerad (30hp)
kl 13:00, Alan Turing (In English)
[Abstract]
Location Based Services (LBS) calculate the position of the user for different purposes like advertising and navigation. Most importantly, these services are also used to help emergency services by calculating the position of the person that places the emergency phone call. This has introduced a number of requirements on the accuracy of the measurements of the position. Observed Time Difference of Arrival (OTDOA) is the method used to estimate the position of the user due to its high accuracy. Nevertheless, this method relies on the correct reception of so called positioning signals, and therefore the calculations can suffer from errors due to interference between the signals. To lower the probability of interference, muting patterns can be used. These methods can selectively mute certain signals to increase the signal to interference and noise ratio (SINR) of others and therefore the number of signals detected. In this thesis, a simulation environment for the comparison of the different muting patterns has been developed. The already existing muting patterns have been simulated and compared in terms of number of detected nodes and SINR values achieved. A new muting pattern has been proposed and compared to the others. The results obtained have been presented and an initial conclusion on which of the muting patterns offers the best performance has been drawn.
2017-06-05 - SaS
Analysis of GPU accelerated OpenCL applications on the Intel HD 4600 GPU
Arvid Johnsson
Avancerad (30hp)
kl 14:30, Alan Turing (In English)
[Abstract]
GPU acceleration is the concept of accelerating the execution speed of an application by running it on the GPU. Researchers and developers have always wanted to achieve greater speed for their applications and GPU acceleration is a very common way of doing so. This has been done a long time for highly graphical applications using powerful dedicated GPUs. However, researchers have become more and more interested in using GPU acceleration on everyday applications. Moreover now a days more or less every computer has some sort of integrated GPU which often is underutilized. The integrated GPUs are not as powerful as dedicated ones but they have other benefits such as a lower power consumption and faster data transfer. Therefore this thesis purpose was to examine whether the integrated GPU Intel HD 4600 can be used to accelerate the two applications image convolution and sparse matrix vector multiplication (SpMV). This was done by analysing the code from a previous thesis that produced some unexpected results as well as a benchmark from the OpenDwarfs benchmark suite. The Intel HD 4600 was able to speedup both image convolution and SpMV by about two times compared to running them on the Intel i7-4790. However, the SpMV implementation was not that well suited for the GPU which meant that the speedup was only observed on ideal input configurations.
2017-06-05 - SaS
En testprocess för webbutvecklingsprojekt med små team
Ludwig Wikblad, Mikael Ögren
Avancerad (30hp)
kl 16:00, Alan Turing (På svenska)
[Abstract]
Att hitta ett lämpligt tillvägagångssätt för att utföra testning i små utvecklingsteam är en utmaning. Många små företag upplever traditionella testprocesser och testförbättringsprocesser som alltför resurskrävande. Målet med denna studie var att undersöka hur en testprocess kan anpassas till små utvecklingsteam utan att den medför en för stor tidsinvestering. Studien utfördes på avdelningen Webb & Mobilt i företaget Exsitec där team om 2-6 personer utvecklar webbapplikationer till företagskunder. I nära samarbete med utvecklarna på avdelningen togs en testprocess fram med målet att anpassas till verksamheten i så stor utsträckning som möjligt. Studien genomfördes som aktionsforskning i tre faser utefter Cooperative Method Development i ett projekt med två utvecklare. Under studiens första fas intervjuades alla utvecklare på avdelningen för att skapa en grundförståelse för verksamheten. Under den andra fasen togs ett antal förbättringsförslag fram tillsammans med utvecklarna. Under den tredje fasen infördes och utvärderades dessa förbättringar. Genom att fokusera på enhetstestning av central affärslogik i applikationen uppnåddes en testprocess som gav utvecklarna ett ökat förtroende för kodens kvalitet utan att upplevas som för resursintensiv.
2017-06-07 - SaS
Prioritizing Tests with Spotify's Test & Build Data using History-based, Modification-based & Machine Learning Approaches
Petra Öhlin
Avancerad (30hp)
kl 09:00, Alan Turing (In English)
[Abstract]
This thesis intends to determine the extent to which machine learning can be used to solve the regression test prioritization (RTP) problem. RTP is used to order tests with respect to probability of failure. This will optimize for a fast failure, which is desirable if a test suite takes a long time to run or uses a significant amount of computational resources. A common machine learning task is to predict probabilities; this makes RTP an interesting application of machine learning. A supervised learning method is investigated to train a model to predict probabilities of failure, given a test and a code change. The features investigated are chosen based on previous research of history-based and modification-based RTP. The motivation for looking at that research is that it resembles the data provided by Spotify. The result of the report shows that it is possible to improve how tests run with RTP using machine learning. Nevertheless, one of the features is performing better than the combined machine learning model. The best performing approach is looking at the history of test results, the more failures recorded for the test over time, the higher priority it gets. Less is sometimes more.
2017-06-07 - SaS
Tuning of machine learning algorithms for automatic bug assignment
Daniel Artchounin
Avancerad (30hp)
kl 10:30, Alan Turing (In English)
[Abstract]
In software development projects, bug triage consists mainly of assigning bug reports to software developers or teams (depending on the project). The partial or total automation of this task would have a positive economic impact on many software projects. This thesis introduces a systematic four-step method to find some of the best configurations of several machine learning algorithms intending to solve the automatic bug assignment problem. These four steps are respectively used to select a combination of pre-processing techniques, a bug report representation, a potential feature selection technique and to tune several classifiers. The aforementioned method has been applied on three software projects: 66 066 bug reports of a proprietary project, 24 315 bug reports of Eclipse JDT and 29 857 bug reports of Mozilla Firefox. In production, using the approach introduced in this work on the bug reports of the proprietary project would have increased the accuracy by up to 16.64 %.
2017-06-08 - HCS
The feasibility and practicality of a generic social media library
Fredrik Jonsén, Alexander Stolpe
Grundnivå (16hp)
kl 10:15, A37 (På svenska)
[Abstract]
Many people today use social media in one way or another, and many of these platforms have released APIs developers can use to integrate social media in their applications. As many of these platforms share a lot of functionality we see a need for developing a library, to contain these, and ease the development process when working with the platforms. The purpose of this paper is to find common functionality and explore the possibility of generalization in this regard. We first look for common denominators between the top social media networks, and using this information we attempt to make an implementation to evaluate the practicality. After the development process we analyze our findings and discuss the usability and maintainability of such a library.
2017-06-08 - SaS
Method for optimisation of WCET in code generated from models
Joakim Kvarnström
Grundnivå (16hp)
kl 10:15, A36 (På svenska)
[Abstract]
In real-time systems, it is important that no task exceeds its budget, the longest execution time allocated to the task. Since failure to execute within budget can have catastrophic consequences, it is important that the WCET, Worst-Case execution time, of the task is known and accounted for.
Estimating the WCET of a task can be done in numerous ways and can be quite cumbersome. Since changes to the code may alter the path of the WCET, it is important that this estimation can be done as easily as possible to ensure that it is in phase with development.
2017-06-08 - HCS
Implementation and evaluation of Bluetooth Low Energy as a communication technology for wireless sensor networks
Tommy Lindman, Emil Nilsson
Grundnivå (16hp)
kl 10:15, A38 (På svenska)
[Abstract]
As the growth and enthusiasm for IoT increases, energy consumption and efficiency is of immense importance in order to develop maintainable and dependable sensor networks. In this thesis we implement and evaluate a wireless sensor network using BLE (Bluetooth Low Energy) as the communication standard with regards to energy consumption, range capabilities and transfer rate. We found BLE to be a highly viable option for such systems, yielding long battery life for systems where long range capabilities are not needed.
2017-06-08 - HCS
Visualizing atmospheric data on a mobile platform
Maximilian Bragazzi Ihrén, Henrik Ingbrant Björs
Grundnivå (16hp)
kl 10:30, A37 Tid:10:35 (På svenska)
[Abstract]
Weather data is important for almost everyone today. The daily weather report, home thermometers, and a lot of other things affect our every day life. In order to develop better and more efficient equipment, tools and algorithms, the people working with this data need to be able to access it in an easily accessible and easy to read format. In this research, methods of visualizing data on mobile platforms are evaluated based on what researchers in the field wants, since their respective fields might want to use very specific visualizations. The implementability of these visualizations are then evaluated, based on the implementations made throughout this paper. The results show that the researchers know what they want, and that what they want is implementable on mobile platforms given some limitations caused by performance.
2017-06-08 - SaS
Enhetstestning inom PL/SQL – en fallstudie
Harald Vogel
Grundnivå (16hp)
kl 10:30, A36 Tid: 10:35 (På svenska)
[Abstract]
Att ha ett väl fungerande automatiskt testsystem är allt viktigare inom mjukvaruutveckling. Tyvärr är det fortfarande vanligt med företag som framförallt förlitar sig på manuell testning i olika former vilket för med sig en rad problem. Extra svårt blir det för de ovanligare programmeringsspråken, som inte har samma historia av automatisk testning som andra språk. Genom att implementera enhetstestning på liten skala på ett riktigt system i PL/SQL och mäta resultaten, kan flera slutsatser dras om enhetstestnings duglighet. Enhetstestning tar betydligt kortare tid att exekvera än motsvarande tester manuellt och kan lätt återupprepas. De kan även testa kombinationer som är omöjliga eller mycket svåra att utföra manuellt, och de minskar tiden det tar att rätta buggar. Att skriva enhetstester tar initialt mycket tid och kräver en viss kunskap. Det kan även dröja innan enhetstester börjar ge resultat och de behöver underhållas allt eftersom att koden ändras.
2017-06-08 - HCS
Internet of Things - Does Particle Photon rely too much on its own cloud solution?
Joel Karlsson
Grundnivå (16hp)
kl 10:30, A38 Tid: 10:35 (På svenska)
[Abstract]
Internet of Things is rapidly growing and there are many devices and cloud solutions on the market. This paper addresses the usage of Particle Photon alongside Microsoft Azure and intends to determine its suitability as an IoT solution. Another intention is to determine if Photon rely too much on its own cloud by aswering what will happen if the service provision ceases? Which parts that can be replaced? What functionality will stop working? The research method includes a Microsoft Azure implementation and device testing without the Particle Cloud. This paper mainly focuses on the communication and software layers between the IoT device and the cloud platform.
2017-06-08 - HCS
Algorithmic Design of Graphical Resources for Games Using Genetic Algorithms
Daniel Eriksson
Grundnivå (16hp)
kl 11:00, A38 Tid 11:05 (På svenska)
[Abstract]
Producing many varying instances of the same type of graphical resources for games can be of interest, such as trees or foilage. But when randomly generating graphical resources, you can often end up with many similar looking results or results that doesn't look like what it is meant to look like. This work investigates whether genetic algorithms can be applied to produce greater varying results when generating graphical resources by basing the fitness of each individual for each genetic generation on how similar the graphical resource is to previously generated resources.
2017-06-08 - SaS
Implementing WebRTC to VISIARC Coffee
Adrian Hedman, Warhell Nasim
Grundnivå (16hp)
kl 11:00, A36 Tid: 11:05 (På svenska)
[Abstract]
WebRTC, Web Real-Time Communications, is a number of protocols and formats for secure real-time communication with buit-in real-time audio and video capabilities. Today this technology is built into leading web browser, for example Google Chrome. The technology is license free and there are several implementations that are open source. This study aims to evaluate some of these open source implementations and apply it to VISIARCs framework Coffee to test WebRTC on a platform-free mobile application.
2017-06-08 - AIICS
Transition-Based Dependency Parsing with Neural Networks
Joakim Gylling
Grundnivå (16hp)
kl 11:00, A37 TID:11:05 (På svenska)
[Abstract]
Dependency parsing is important in contemporary speech and language processing systems. Current dependency parsers typically use the structured perceptron machine learning component, which classify based on millions of sparse indicator features, making developing and maintaining these systems expensive and error-prone. This thesis aims to explore whether replacing the structured perceptron component with an artificial neural network component can alleviate this problem without hurting performance, in terms of accuracy and efficiency. A simple transition-based dependency parser using the artificial neural network as the classifier is written in Python3 and the same program with the classifier replaced by a structured perceptron component is used as a baseline. The results show that the ANN dependency parser provides slightly better accuracy with only the most basic atomic features, eliminating the need for complex feature engineering. However, it is about 3 times slower and the training time required for the ANN is significantly longer.
2017-06-08 - HCS
An analysis of microservice frameworks
Erik Edling, Emil Östergren
Grundnivå (16hp)
kl 11:30, A37 Tid:11:25 (På svenska)
[Abstract]
Microservice architecture has entered the industry to solve some of the problems with the monolithic architecture. However, this architecture comes with its own set of problems. In order to solve the microservice architecture problems while also providing additional functionalities, microservice frameworks have been developed. In this thesis, microservice frameworks were compared and thereafter two were chosen to implement a small part of a large monolithic system as microservices. This was done in order to see how well they could implement the different functionalities that the frameworks provide in relation to the benefits and the cross-cutting concerns of the microservice architecture. The results showed that the frameworks embraced the benefits of the microservice architecture in the aspects of maintainability and scalability. However, in the terms of being able to change frameworks in the pursuit of newer technologies there were problems. Some functionalities such as service discovery requires all of the new services created to use the same mechanism in order to create a unified system. There were also problems caused by the load balancing mechanism provided by the frameworks used in this thesis. The load balancing mechanism made the system unable to send large data files which was crucial for the system that was to be implemented as a microservice system.
2017-06-08 - HCS
Design av en synchronized-database till en mobil spelapplikation med hjälp av Firebase
Arvid Karlsson
Grundnivå (16hp)
kl 11:30, A38 Tid:11:25 (På svenska)
[Abstract]
Att ha välfungerande backend till ett system kan spara arbete på underhållning och utveckling. Detta arbete bygger på tidigare arbeten med utvecklingen av ett kartbaserat Android spel och en webbaserad spel editor. Genom att implementera Firebase-funktionaliteter som Firebases realtidsdatabas med en ny datastruktur som är optimerad därefter så kan man jämföra skillnaden mot det gamla systemet och på så sätt utvärdera Firebase.
2017-06-08 - SaS
Fuzzy Testing for design assurance levels
Marcus Gustafsson, Oscar Holm
Grundnivå (16hp)
kl 11:30, A36 Tid: 11:25 (På svenska)
[Abstract]
With safety critical software, the developers need to be sure that the programs are safe and stable. This is now done with extensive handwritten tests today.
Fuzzy testing is a well tested method for automatic testing. Though it has had problems with finding specific code paths in the software under test, which has made testing with fuzzy testing a very impractical testing method.
2017-06-08 - SaS
Klientbaserad GeoTIFF-rendering
Harald Grant, Marcus Lewin
Grundnivå (16hp)
kl 11:45, A36 (På svenska)
[Abstract]
När en användare idag efterfrågar rendering av en kartvy i en applikation behöver en server först rendera en bild utifrån given geografisk data och därefter skicka bilden till klientens mobila enhet. Detta kan resultera i höga responstider, speciellt för användare som befinner sig i områden med bristfällig täckning. I denna studie utvärderas en alternativ lösning där rendering istället sker direkt på klientens enhet. En prototyp av en mobil kartapplikation med stöd för lokal rendering av geografisk rådata utvecklas och utvärderas utefter en konstant för acceptabel fördröjning vid visualisering av information. Resultatet av testerna visar att prototypens prestanda är beroende av mängden information som ska visas. För högre zoomnivåer ger prototypen ett tillfredsställande resultat, men vidare åtgärder krävs för de lägre nivåerna. De främsta utmaningarna vid utvecklingen av applikationen redovisas och förbättringsförslag för fortsatt utveckling framförs.
2017-06-08 - AIICS
Presenting utilisation rates of resources based on positional tracking data
Alexander Kazen
Grundnivå (16hp)
kl 11:45, A37 (På svenska)
[Abstract]
Technological advancement of today provides many fields where companies may gain advantages by keeping up to date. One such field that is quite unexplored as of today is tracking the utilisation of (arbitrary) material resources. This paper evaluates the possibility to create a tool for visualising this using only simple positional tracking data as input. It shows that such a tool can be created, but also that it might need some technical improvements to be accurate in all contexts. At last, what kind of improvements might be needed are discussed.
2017-06-08 - SaS
Design and Analysis of a Novel Bitcoin Debit Card in a Payment System
Jakob Lövhall
Avancerad (30hp)
kl 13:15, Allan Turing (In English)
[Abstract]
This thesis introduces a debit card based payment system that aims to provide fast in store payments using Bitcoin. The system introduces physical constraints to Bitcoin payments to enable stores to accept payments before they are in a block.

The proposed system was tested in a proof of concept model to estimate the time it would take to pay with the system on the real Bitcoin network. A security analysis was performed to validate that the system is secure or, to point out the potential problems. The security analysis was done by interviewing five experts from different organizations to get different views on the system.

The analysis of the system indicates that it could potentially be fast enough to be used in stores but also that security problems exists which needs to be considered.
2017-06-09 - SaS
Visualization of Internet of Things data from a District Heating System
Erik Malmberg
Avancerad (30hp)
kl 09:00, Alan Turing (In English)
[Abstract]
One application of the Internet of Things that is currently developed is the collection of data from district heating systems. This development leads to the problem of how the collected data should be used to operate the systems in a manner that reduces the dependency on fossil fuels. This thesis proposes a characterization of this problem into visualization tasks and requirements, as well as a visualization application, based on the tool Power BI, for gaining insight into how to operate a district heating system. The tasks and requirements were elicited during collaborative meetings and prototyping together with district heating experts. The perceived effectiveness of the visualization application was validated with a usability evaluation, including the usability metric for user experience questionnaire, with four district heating experts, even though the perceived efficiency was found to be lacking compared to the other usability aspects. A redesign of one of the Power BI components, which according to the evaluation should improve the application, is also suggested.
2017-06-09 - SaS
Visualisation of data from IoT systems – A case study of a prototyping tool for data visualisations
Jonathan Andersson
Avancerad (30hp)
kl 10:30, Alan Turing (På svenska)
[Abstract]
The client in this study, Attentec, has seen an increase in the demand for services connected to Internet of things systems.
This study is therefore examining if there is a tool that can be a used to build fast prototype visualisations of data from IoT systems to use as a tool in their daily work.

The study started with an initial phase with two parts.
The first part was to get better knowledge of Attentec and derive requirements for the tool and the second part was a comparison of prototyping tools for aiding in development of data visualisations.
Apache Zeppelin was chosen as the most versatile and suitable tool matching the criteria defined together with Attentec.
Following the initial phase a pre-study containing interviews to collect empirical data on how visualisations and IoT projects had been implemented previously at Attentec were performed.
This lead to the conclusion that geospatial data and NoSQL databases were common for IoT projects.
A technical investigation was conducted on Apache Zeppelin to answer if there were any limits in using the tool for characteristics common in IoT system.
This investigation lead to the conclusion that there was no support for plotting data on a map.

The first implementation phase implemented support for geospatial data by adding a visualisation plug-in that plotted data on a map.
The implementation phase was followed by an evaluation phase in which 5 participants performed tasks with Apache Zeppelin to evaluate the perceived usability of the tool.
The evaluation was performed using a System Usability Scale and a Summed Usability Metric as well as interviews with the participants to find where improvements could be made.
From the evaluation three main problems were discovered, the import and mapping of data, more feature on the map visualisation plug-in and the creation of database queries.
The first two were chosen for the second iteration where a script for generating the code to import data was developed as well as improvements to the geospatial visualisation plug-in.
A second evaluation was performed after the changes were made using similar tasks as in the first to see if the usability was improved between the two evaluations.
The results of the Summed Usability Metric improved on all tasks and the System Usability Scale showed no significant change.
In the interviews with the participants they all responded that the perceived usability had improved between the two evaluations suggesting some improvement.
2017-06-09 - AIICS
Organ Detection and Localization in Radiological Image Volumes
Ola Jigin, Tova Linder
Avancerad (30hp)
kl 13:00, Alan Turing (In English)
[Abstract]
Using Convolutional Neural Networks for image classification and object localization and detection in images is becoming increasingly popular. Within radiology a huge amount of image data is produced and meta data containing information of what the images depict is currently added manually by a radiologist.

To aid in streamlining physician's workflow this study has investigated the possibility to use Convolutional Neural Networks (CNNs) that are pre-trained on natural images to automatically detect the presence and location of multiple anatomies in medical CT-images.

The results show promise for multiclass classification with an average precision 89.41% and average recall 86.40%. This also confirms that a CNN that is pre-trained on natural images can be succesfully transferred to solve a different task. It was also found that adding additional data to the dataset does not necessarily result in increased precision and recall and decreased error rate. It is rather the type of data and used preprocessing techniques that matter.
2017-06-09 - SaS
Comparing modifiability of React Native and two native code bases
Robin Abrahamsson, David Berntsen
Avancerad (30hp)
kl 14:30, Alan Turing (In English)
[Abstract]
Creating native mobile application on multiple platforms generates a lot of duplicate code. This thesis have evaluated if the code metric modifiability improves when migrating to React Native. One Android and one iOS codebase existed for an application and a third was created with React Native. The measurements of the codebases were based on the SQMMA-model. The metrics for the model were collected with static analyzers created specifically for this project. The result created consists of graphs that show the modifiability for some specific components over time and graphs that show the stability of the platforms. These graphs shows that when measuring code metrics on applications over time it is better to do this on a large codebase that has been developed for some time. When calculating a modifiability value the sum of the metrics and the average value of the metrics between files should be used. It has also been shown that the React Native seems to be more stable.
2017-06-09 - SaS
Automating Text Categorization with Machine Learning: Error responsibility routing in a multi-layer hierarchy
Ludvig Helén
Avancerad (30hp)
kl 16:00, Alan Turing (In English)
[Abstract]
The company Ericsson is taking steps towards embracing automating techniques and applying them to their product development cycle. More specifically, they would like to use machine learning to automate the evaluation of a text categorization problem of error reports, or trouble reports (TRs). An excess of 100,000 TRs are handled annually.
A TR consists of a free text description of an issue or bug together with log data regarding intellectual property of Ericsson. The current method of evaluating TRs is an arduous, bureaucratic and costly process where experts manually analyze the logs to identify the product team responsible for correcting the issue, resulting in long lead times between meetings at great cost for the company.

The challenge posed is to accurately predict which company product team a TR belongs to, based on predefined fields and error descriptions. Two different preprocessed data sets of 66 066 and 10 218 historical TRs has been used in this work.
This thesis presents two possible solutions for solving the routing problems where one technique uses traditional classifiers for deciding the route through the company hierarchy where a specific TR belongs. The other solution utilizes a deep learning technique with a Convolutional Neural Network for translating the TRs into word vectors, or word embeddings, in order to decide what group within the company should be responsible for the handling of the TR. The traditional classifiers achieve up to 83% accuracy and the Convolutional Neural Network achieve up to 63% accuracy in the task of predicting the correct class for a specific TR.


Page responsible: Ola Leifler
Last updated: 2017-04-27