Wednesday, May 30, 2012

Theses 2012 - first part

the academic year is slowly ending. It was very busy, I have been supervising twelve MSc. and BSc. theses. My students are preparing to defend their work and take the state exams. This was the main reason, we got together and rehearsed the presentations. Let me give you an idea about the first four projects in this blog. 

As in all my work at the university I try to split my attention between education, research and business. This is reflected in the diploma problems formulation. Every project has to solve some real problem. Some of the theses turned out more research oriented and some led to pilots of real products.

In total I was leading twelve theses and it is a lot of work. I tried to spend at least a hour with each of the students every week. To be effective I have followed a simple development process. The work on the diploma problem has been always started with an essay. Half of A4 page with high-level description of the problem. Students were not allowed to describe the technology the key is the added value. They had to conclude with an explanation of what is being developed and what will be the benefit for users. I think, the full understanding of the project benefits is the key for motivating students.

Next, they had to set a blog for weekly progress reports. Some inclined more to a shared Google Docs document, some used github, but each of them was keeping track about the development. I used the blog during the weekly meetings. The first step was to review what has been done since the last meeting. Second I tried to understand what needs to be done next and set the follow up tasks.

CTU is offering a dissertation Latex template, with a standard theses structure. All of the students were using it and we have shared the pdf version in Dropbox. Blog, Google Docs, Dropbox and other tools enabled me to be constantly in touch and help when required. What are the result? Let me describe the first batch of theses.

Martin Prokš has designed a helpdesk application. The dissertation title is RESTful Helpdesk Application (in Czech). It sounds simple there are many implementations of helpdesk in shareware, but he has taken a very difficult path. The app is implemented in Java and Spring. As part of the solution he has developed simple templates for implementing REST API. The server part is hosted on Google App Engine. For client he has chosen Google Web Toolkit. The development was not exactly smooth. Martin had to solve many little problems because of documentation details etc., but he is a real programmer spending major part of the day next to his machine. Since we have been sharing the source, when I touched some of mine computers I have been observing new files being updated to my Dropbox. The application is almost ready.

Petr Michalička was also dealing with the web. His dissertation “Automatic Deployment to PaaS Cloud” helps users install and configure popular open source PHP packages to PaaS environment. Currently his application supports Cloud Foundry, AWS, and Eucalyptus. The UpCf SaaS application enables simple deployment of several popular shareware applications to cloud. It is designed for users with no programming skills. The thesis also summarizes in general the main steps in migrating PHP apps to PaaS cloud. It suggests solutions for future integration of more packages.

Ondřej Šťastný has submitted the “Rapid Application Development of RESTful Servers” theses. His main idea was to help developers rapidly create server part of web applications. His approach allows designers to use only declarative client-side code. Ondrej has formalized the process introducing a novelty approach Schema-Permissions-Data (SPD) schema for defining data model and server interface. His approach is also minimizing number of protocols used in order to simplify the design and development. There are two reference applications in his theses.

Tomáš Tunys wrote the “Gestures detection for Android” theses. He has put together a complete system for collecting mobile devices gestures including a management system. The gesture is defined as a device movement, hand-drawn shape on a touch screen, or a movement captured by a camera. They all generate time series characterizing the movement. The main part was design and implementation of all basic algorithms for training and recognition. The gesture recognition deploys vector quantization and Hidden Markov Models. To collect the data and test the reco he had to design and implement an Android application. Nice job. Tomas applied for a PhD program. Even better!

This is the short description of the first four theses. I'll describe all theses and I will try to put the pdfs on web, wait for the part two of this blog. Keep the fingers crossed for my students, while defending the theses next week.

No comments:

Post a Comment