As the era of digitalization dawns, the need to integrate separate silos into a synchronized connected system is becoming of ever greater significance. This thesis focuses on the Canvas Learning Management System (LMS) and the Digitala vetenskapliga arkive (DiVA) as examples of separate silos. The thesis presents several methods of automating document handling associated with a degree project. It exploits the fact that students will submit their thesis to their examiner via Canvas. Canvas is the LMS platform used by students to submit all their coursework. When the examiner approves the thesis, it will be archived in DiVA and optionally published on DiVA. DiVA is an institutional repository used for research publications and student theses. When manually archiving and publishing student theses on DiVA several fields need to be filled in. These fields provide meta data for the thesis itself. The content of these fields (author, title, keywords, abstract, …) can be used when searching via the DiVA portal. It might not seem like a massive task to enter this meta data for an individual thesis; however, given the number of theses that are submitted every year, this process takes a large amount of time and effort. Moreover, it is important to enter this data correctly, which is difficult when manually doing this task. Therefore, this thesis project seeks to automate this process for future theses. The proposed solution parses PDF documents and uses information from the LMS in order to automatically generate a cover for the thesis and fill in the required DiVA meta data. Additionally, information for inserting an announcement of the student's oral thesis presentation into a calendar system will be provided. Moreover, the data in each case will be checked for correctness and consistency. Manually filling in DiVA fields in order to publish theses has been a quite demanding and time-consuming process. Thus, there is often a delay before a thesis is published on DiVA. Therefore, this thesis project’s goal is to provide KTH with an automated means to handle thesis archiving and publication on DiVA, while doing so more efficiently, and with fewer errors. The correctness of the extracted meta data will be evaluated by comparing the results to the previously entered meta data for theses that have previously been achieved in DiVA. The automated process has been calculated to take roughly 50 seconds to prepare the information needed to publish a thesis to DiVA with ~71% accuracy, compared with 1 hour and 34% accuracy in the previous manual method. / När digitaliseringens tid uppstår, så blir behovet av att integrera separata silor i ett synkroniserat anslutet system större. Denna avhandling fokuserar på Canvas Learning Management System (LMS) och Digitala vetenskapliga arkivet (DiVA) som är exempel på separata silor. Avhandlingen presenterar flera metoder för automatisering av dokumenthantering för examensarbeten. Projektet utnyttjar det faktum att eleverna kommer att skicka sin avhandling till sin examinator via Canvas. Canvas är den LMS-plattform som används av eleverna för att lämna in sitt kursarbete. När examinatorn godkänner avhandlingen kommer den att arkiveras i DiVA och eventuellt publiceras på DiVA. DiVA är ett institutionellt arkiv som används för forskningspublikationer och studentavhandlingar. När man manuellt arkiverar och publicerar studentuppsatser på DiVA måste flera fält fyllas i. Dessa fält ger metadata för själva avhandlingen. Innehållet i dessa fält (författare, titel, nyckelord, abstrakt, ...) kan användas vid sökning via DiVA-portalen. Även om det inte är en stor uppgift att skriva in denna metadata för en individuell uppsats så blir det en mycket tidskrävande process för många examensarbeten. Dessutom är det viktigt att ange dessa uppgifter korrekt, vilket är svårt när man manuellt utför den här uppgiften. Därför syftar detta avhandlingsprojekt till att automatisera denna process för framtida avhandlingar. Lösningen som presenteras i denna avhandling kommer att analysera PDF-dokument och använda annan information från LMS för att automatiskt skapa en fram- och baksida för avhandlingen och fylla i de nödvändiga DiVA-metadata. Grunden för införandet av denna data i ett kalendersystem för att ge ett meddelande om studentens presentation kommer också att ges. Dessutom kontrolleras uppgifterna för korrekthet. Manuell fyllning av DiVA-fält för att publicera avhandlingar har varit en ganska arbetsam och tidskrävande process. Således är det ofta en fördröjning innan en avhandling publiceras på DiVA. Därför ska detta projektet ge KTH ett automatiserat system att hantera avhandlingar och publicering på DiVA, samtidigt som det gör det mer effektivt och med färre fel. Korrektheten hos de extraherade metadatan kommer att utvärderas genom att jämföra resultaten med de tidigare inmatade metadatan för examensarbeten som redan ligger uppe på DiVA.  Den automatiska processen tar ungefär 50 sekunder att förbereda information för att publicera en avhandling till DiVA med ~ 71% noggrannhet jämfört med 1 timme och 34% noggrannhet i tidigare manuell metod.

REST (Representational state transfer) is an architectural style for distributed hypermedia systems. The simplicity of REST allows straightforward communication between HTTP clients and servers using URIs (Uniform Resource Identifiers) and HTTP methods, e.g., GET, POST, PUT, and DELETE. To do the communication effectively between clients and servers, there is a set of best design practices (design and linguistic patterns) shall be followed, and a set of poor design practices (design and linguistic antipatterns) shall be avoided. This study aims to determine whether there is a relationship between design and linguistic quality in Google RESTful APIs. To find this relation, a tool is developed to detect patterns and antipatterns in REST APIs both in terms of design and linguistic quality. The input of this tool is qualitative data (Google APIs) and its output is quantitative data. Using this quantitative data, a statistical study is then performed to detect the relation. The tests that are conducted to obtain the final results are Chi-squared and Phi Coefficient tests. The result of Chi-squared that considered all the groups of patterns and antipatterns shows that there is a statistically significant relation between design and linguistic quality. However, when we assess the individual pair of patterns and antipatterns, our Phi Coefficient tests show that for most of the cases, there is no or negligible relationship between linguistic and design patterns and antipatterns.

REST (REpresentational State Transfer) is commonly used for designing APIs. Two main categories of REST API quality have been identified in previous research: linguistic and design quality. Linguistic quality revolves around the design of the URIs. Design quality revolves around the metadata and body in HTTP requests and responses. For enabling and simplifying communications with REST, both linguistic and design quality are important, however, previous research has shown that even major APIs using REST are not always following best practices for linguistic and design quality. This study investigates if there is a statistical relation between linguistic and design quality. We selected 326 API endpoints from ten public APIs for this study. This study has reused and improved a Java-based tool in previous research for detecting aspects of linguistic quality in the APIs endpoints. For this study, we also developed a tool based on Node.js for detecting aspects of design quality in the API endpoints. These two tools are applied on the same API endpoints to be able to study the statistical relation. A Chi-Square test, implemented with R, showed that there is a significant statistical relation in our findings between linguistic and design quality. Pairwise phi-coefficient comparisons, implemented with Python, between each combination of the linguistic and design aspects used in this study identified eight weak and two moderate relations among the linguistic and design quality aspects. However, sample tests showed that the Java-based tool for detecting linguistic quality were not accurate, which made us fail to answer our problem formulation.