Hi there, I'm Thiemo.

Picture of Thiemo

I am a PostDoc and project leader at the Machine Learning for Education Laboratory (ML4ED) at the École Polytechnique Fédérale de Lausanne (EPFL).

My research lies in the general area of Human-Computer Interaction with influences from Natural Language Processing, and Machine Learning. I strive to understand how humans perceive, interact, and learn with intelligent tools. Based on these insights, I build adaptive user interfaces that go beyond static and rule-based interaction to advance the capabilities when working and learning in the digital world.

Specifically, I am driven by the vast opportunities to enhance and improve pedagogical scenarios based on recent advantages in Natural Language Processing and Machine Learning to enable humans to learn self-reliant and individually independent of an educator or their background. To do so, I use techniques from Artificial Intelligence (AI) to build AI-powered education tools. My research follows three connected main lines of work: 1) create and study pedagogical conversational agents or intelligent writing support systems to improve educational scenarios, 2) explore methods and techniques to model student performance in numerical or textual data, and 3) build computational approaches that control when, where and how to provide students intelligent feedback and self-evaluation.

As part of a research team, I have developed adaptive argumentation writing support systems which helps individuals to develop better writing and reasoning skills (Link to CHI 2020 or ACL 2022 paper). Also, I built an intelligent conversational agent that provides students with adaptive and individual argumentation tutoring while writing persuasive essays independent of an instructor, time or location (Link to CHI 2021 paper). Moreover, I designed an empathy writing support tool by modeling the cognitive and emotional empathy structures in student-written texts (Link to ACL 2021 or CHI 2022 paper).

I believe in the power of collaboration for building better AI-based systems. If you are interested in my work, please contact me via LinkedIn.

I co-organized the Second Workshop on Intelligent and Interactive Writing Assistants at CHI 2023! More information on our website! Stay tuned! 🎉

Selected Publications

Here you find an overview of my publications at Google Scholar and my academic profile at ReserachGate.

Education

Natively Developed Prototypes

I have designed, developed and scientifically evaluated various prototypes based on different technologies such as Flask, TensorFlow, FastText, PyTorch, Keras, SpaCy or Scikit-Learn.

ArgueLearn


an adaptive argumentation writing support tool based on artificial intelligence (Info)

ArgueTutor


a chatbot for individual argumentation tutoring (Info)

ArgumentFeedback


a Google Docs Add-on for argumentative writing support on business or project pitches (Info)

Elea - an empathy feedback tool


an adaptive empathy writing support tool based on artificial intelligence for students to foster empathic writing independent of an instructor, time and location (Info)

Eva - a course evaluation bot


a chatbot for personal course evaluations in large-scale or distance-learning scenarios (Info)

Hermine - a Q&A-Bot designed for educational settings


a chatbot for frequently-asked questions in large-scale or distance-learning scenarios (Info)

Corpora and Data Sets

In order to build HCI innovations, I collected, annotated and scientifically evaluated different corpora and data sets to train novel machine learning or deep learning models. Here is an overview of my work:

  • German User-Generated Repair Instruction Corpus: 3,157 user-generated posts, with 682 tagged “Instruction” and 2,475 tagged “NoInstruction”, representing different characteristics of the platform motor-talk. Access
  • Argumentation Annotated Student Peer Reviews Corpus: 1000 persuasive student peer reviews about business model feedbacks annotated for their argumentative components and argumentative relations. Access
  • Argumentation Annotated Student Business Model Pitch Corpus: 200 persuasive business model pitches annotated for argumentative components, argumentative quality score and argumentative relations. Access
  • Cognitive and Emotional Empathy Annotated Student Peer Reviews Corpus: 500 student peer revies about business model feedbacks annotated for their cognitive and emotional empathy level according to key constructs. Access

    • Awards and Honors

    Invited Talks