Jelle Teijema
PhD Candidate @ Utrecht University, Applied Data Scientist, Researcher @ ASReview.
General Info
- Date of BirthDec 02, 1996
- Mailjelle@teije.ma
- University Mailj.j.teijema@uu.nl
- Twitter @JJTeijema
- Telegram @JetJailMelee ← it's an anagram 🤯
Online presence
PhD Candidate @ Utrecht University, Applied Data Scientist, Researcher @ ASReview.
I am the first Ph.D. candidate following the Applied Data Science master's at Universiteit Utrecht. I will combine applied data science, open science, AI, and behavioral science in many different projects for the next four years with Rens van de Schoot and Lars Tummers as doctoral supervisors and Ayoub Bagheri and Matthieu Brinkhuis as co-supervisors.
Inbetween my master's and Ph.D., I have worked as a researcher for ASReview. I remain a maintainer for the project during my Ph.D.
During the Summer of 2021, I worked for the University's Summer School as a teaching assistant for the course Applied Text Mining. The course teaches the basic and advanced concepts and ideas in text mining and natural language processing.
In this project, I joined a Utrecht University-based research team in exploring the implementation and use of complex neural networks as text classification models for open-source software used by tens of thousands of researchers worldwide.
Deep learning is often used in text classification tasks for its efficiency and proficiency in modelling nonlinear processes. However, using this type of machine learning takes more time and processing power than using shallow learning algorithms. This study aims to determine if a combination of shallow and deep learning techniques can be used in increasing the classification performance for automated systematic review software. Since deep learning networks excel at finding difficult connections, this study aims to find situations in which they outperform shallow networks. To find these situations, simulations were run on a prepared dataset using different classifiers, switching from shallow to deep networks. While at first results showed no improvement, continued exploratory research provided results that support the hypothesis, as situations were found in which the neural network outperforms the shallow network classifier.
The thesis can be HERE,
and the code accompanying my thesis can be found here: 
My research thesis on Process Mining, Architecture Mining and Visualisation as part of my bachelor's degree.
The process mining technique contains powerful tools to extract processes from big event logs. However, real world applications are more likely to produce snapshots of the process over only a smaller period, keeping the absolute process a secret. This research supplies a framework that can create a controlled environment from a absolute event log in which these snapshots, called samples, can be mined and analyzed, while having full control over the degree of variation in sampling. Having the absolute process and the samples derived from it provides a new perspective into the relations between sampling rates and techniques, and the assessments of the samples created by them. The framework is created using readily available tools such as ProM and RapidMiner. The sampler techniques used are derived from the ''Measuring the Behavioral Quality of Log Sampling''. The result is a flexible framework, split up into 3 components. Each component, i.e. the Sampler, Miner, and Analyzer, applies algorithms related to sampling and process mining to transform the original log into samples, models and measurements, ready to used for further research.
Apple Authorised Service Provider as a parttime occupation next to my study. I stopped here in 2021 after moving to ASReview as Researcher
During my bachelor's, I spent a year rowing for ORCA's Men's eight team, following an intense training and dietary schedule.
Data are everywhere. In today’s society it is becoming more and more important to put these data effectively and safely to use. The new Master’s programme Applied Data Science enables you to become a data science professional with excellent analytic capabilities.
You are able to apply these capabilities across a wide number of domains, such as health science, geo science, social and behavioural science and media studies. The programme is taught in English and was set up in co-creation with the industry. This Master’s programme is intended for students who have a clear and ambitious interest in Applied Data Science. Applicants can come from a wide range of fields. You must have an academic understanding of statistics and programming.
Information Science offers an interesting mix of psychology, communication studies, organisational studies and computing sciences. Through this combination of aspects students will be uniquely qualified to build bridges between people and technology, and between the needs of companies and IT solutions.
My research thesis is on Process Mining, Architecture Mining and Visualisation as part of this degree. Subjects I enjoyed during this bachelor were Data Analytics, Information systems and Architecture modelling.
For my pre-Uni education I chose for the Nature and Technology branch. The curriculum I picked was a course in Latin and Music history and theory.