Design and Development of a Recommender System for Job Recommendations

Innovation Development
Thesis Code: 

Thesis Type: Thesis in Computer Science, Data Engineering, Computer Engineering, Mathematical Engineering

• Experience with Python and/or Java
• Basic knowledge of modular development
• Basic knowledge of REST API implementation
• Beginner (at least -- or willing to learn shortly) of machine learning.

The undergraduate will design and develop a deep learning recommender system that selects those users who are (a) interested in being notified about a given job position, and (b) appropriate candidates for the given job. The innovation being researched will output a solution that balances both user interests and requirements for a given job as well as dealing with the cold-start situation (i.e. not enough personal data to perform the recommendation task).

The thesis will be structured as follows:
• problem formulation: objective function, data structures and resources to be used
• algorithm design (covering all the value chain from the collection of the required data to the generation of the recommendation) and approach implementation according to software engineering best practices
• In-lab testing verification with real data and measurement of the goodness of the approach.

The implementation will process real data from a well-known business- and employment-oriented social networking service. The goodness of the solution will be tested in an offline-setting scenario with quantitative and qualitative analyses. The undergraduate will benefit from being immersed in a research environment. It is a unique setting to get into a research mindset with a strong push for the innovation. At the end of the thesis, the undergraduate will be able to master recommender systems, semantics and neural networks. He/she will use proficiently control version systems, continuous integration systems, remote deploying and monitoring techniques.

Contact: send a resume with attached the list of exams to specifying the thesis code and title.