STEERS: SmarT rEsEarch Recommender System

Project introduction and background information

Thesis supervision is often organized ad-hoc, with students, supervisors and internship congregating in an unstructured and unguided manner. As such, students may miss out on valuable opportunities in terms of academic fulfillment and job opportunities. The STEERS project at the University of Twente (UT) developed a recommender system that enables students to orient themselves on thesis trajectories, aligning their personal objectives with appropriate topics, companies and people. By performing Natural Language Processing on the UT thesis database, the STEERS recommender system describes and visualizes a plethora of insights, enabling students to focus on appropriate thesis projects and connect to relevant contacts. In the end, the project delivers a tool that enables students to optimally prepare and focus their thesis journey, bridging the gaps between students, research topics, academic supervisors, and companies.

Objective and expected outcomes

The STEERS project had the following overarching objective:

“Creating a recommender system for suggesting research topics and relevant thesis supervisors, based on the professional ambition and educational path of the student.”

The main objective was decomposed into three concrete project goals:

·        Goal 1: To valorize and expand the use of the existing repository of past thesis projects by scraping and linguistically linking its data on the thesis topics, supervisors, companies, and/or study programs.

·        Goal 2: To create a dashboard with insights on network associations between research topics, companies, supervisors, students, and study programs.

·        Goal 3: To create a working prototype of a recommender system that helps in matching appropriate supervisors with students for bachelor and master thesis projects.

Results and learnings

The project realized the intended project goals, culminating in a working prototype that will be further professionalized during the 22-23 academic year. For this purpose, another year of WSV funding has been secured.

Recommendations

The experiences of the STEERS project led to the following recommendations:

• Standardize and improve the quality and completeness of thesis data stored in university repositories. This allows to operate the tool in the long term without requiring manual data cleaning activities. Furthermore, appropriate data standards maximize the impact of insights extracted from repositories.
• Engage a variety of stakeholders (students, teachers, educational directors, database managers, administrative staff) to holistically design a system with a broad support base and tuned to different needs.
• Have a follow-up plan in place, enabling to transition from a functioning prototype system to a fully deployed system that is embedded within the existing IT infrastructure and has a designated product owner.

Practical outcomes

 The STEERS project yielded the following concrete outcomes:

• Result 1: A database that contains relevant thesis information, extracted from the UT thesis repository and the thesis administration system
• Result 2: A mapping of required and desired data entries, serving as input for future system redesigns aiming at more complete and consistent data.
• Result 3: An overview of user requirements for a thesis recommender system, clustered by role at the university.
• Result 4: A prototype dashboard that guides students in their thesis journey, offering directed search options and insightful visualizations.