This is a great day. Over the last approximately 7 months I have worked hard on my Master’s Thesis. The goal was to contribute to the community by creating an open source app for an open source platform. And thus, I started to work on a Recommender App at the company behind Nextcloud, an open source file sync and share software.
The Recommender App consists of two recommendation techniques: Collaborative Filtering and Content-Based Recommendation. The app knows which files was last viewed and searches for similar files that could possibly interest the user. At the same time, the contents of the files are parsed and compared with those available in the user’s Nextcloud.
The main challenge of creating such an recommendation system was the lack of test and training data. Due to the nature of Nextcloud, it is not possible to collect/store training data centrally in order to train the system. But on the other hand, this fact means that data never leaves the users Nextcloud!
Parallel to the software I was working on my thesis document which profounds the requirements, State of the Art, implementation and evaluation of the app.
“RecommendationAssistant” is unfortunately not finished yet. Tests on my development machine went fine as well as tests on my private Nextcloud. But other instances had some issues that have to be fixed in the future.
The app is published on GitHub (see here: https://github.com/doganoo/RecommendationAssistant). My thesis document is available here: http://dogan-ucar.de/wp-content/uploads/2018/04/Master’s-Thesis-Doğan-Can-Uçar.pdf. The final presentation is also available here: http://dogan-ucar.de/wp-content/uploads/2018/04/Presentation-Master’s-Thesis-Doğan-Can-Uçar.pdf.pdf.
I hope I have lived up to my own claim to make a contribution to community to some extent. Please feel free to let me know your opinion about the software and thesis.