Michael Roth

+49 711 685 81367
Email

Pfaffenwaldring 5b
70569 Stuttgart
Deutschland
Room: 01.020

Office Hours

By appointment.

News

• April 2024:
  • Talita Anthonio defended her PhD thesis!
  • I am now an action editor for CL and a senior area chair for ARR. 
• March 2024:
  • I am part of the advisory committee for the UnImplicit workshop.
  • Welcome Berfin Aktaş to the MisT research group!
• February 2024:
  • I am an action editor and area chair for ARR/ACL 2024.
  • I have been certified by MHFA to offer first aid for mental health.
• January 2024: Katharina and I got an EACL workshop paper accepted at LT-EDI.
• October 2023: I have started work as the faculty's gender equality officer.
• September 2023: I am an area chair for "Lexicon and Semantics" at LREC-COLING.
• August 2023: Titus and I have started a new IRIS3D project on polarization.
• July/Aug 2023: I gave a course on misunderstandings and other NLP challenges at the European Summer School in Logic, Language and Information (ESSLLI)
• July 2023:
  • Ivan Vulic and I taught a course on Large Language Models at ESSAI.
  • Nicola, Agnieszka and I received a best paper award for our ACL-SRW paper!
  • I gave an invited talk in the Linguistischer Arbeitskreis Bochum at RUB.
• June 2023: We got one paper accepted at the ACL student research workshop!
• May 2023: I taught a practical workshop on Transformer Models for DH in Cologne
• April 2023: Nicola, Agnieszka and I got an abtract accepted at IC2S2!
• December 2022: I gave an invited talk for GSCL. For more GSCL events, see here.
• September 2022:
  • I gave the opening keynote at KONVENS. More details here.
  • We got one paper accepted (and one rejected) at AACL-IJCNLP 2022.
• July 2022:
  • We held the 2nd UnImplicit Workshop at NAACL 2022.
  • Our shared task paper at SemEval won a best paper award!

About me

As a computational linguist, I combine methods from computer science, linguistics, and the digital humanities. My research is driven by a deep fascination for natural languages and their efficiency. One focus for me is on variety, which occurs because language is used differently depending on context, and we as humans can apparently adapt to it without any effort. For computational models, a further challenge is posed by phenomena of implicit language, that is information and relations that seem intuitively obvious but are not explicitly realized and therefore difficult to capture automatically. Finally, misunderstandings can also occur between competent speakers, the linguistic causes of which also need to be taken into account by computational models.

Since March 2019, I have been leading a DFG Emmy Noether research group at the IMS, which focuses on potential causes of misunderstandings in instructional texts. We are investigating to what extent such causes can be reconstructed from revision histories, which linguistic phenomena play a role in these processes, and how computational linguistic methods can be employed to automatically detect and correct relevant text passages. Our research has resulted in several conference papers and student theses, as well as two best paper awards: Alok Debnath (now at Trinity College Dublin) received the Best Student Paper award at EACL 2021, and our shared task on clarifications received the Best Task Paper award at SemEval 2022. Since last August, I am also co-PI (with Titus von der Malsburg) of the interdisciplinary IRIS3D-project "Dynamics of Polarization and AI Remedies". My group's preliminary work for this seed-funded project received a Best Student Paper award at ACL 2023. Besides these funded projects, I am also publishing and teaching with colleagues on topics such as implicit reference in dialogue (with Thang Vu, see AACL'22), multilingual frame identification (with Sebastian Pado, see LiLT'22) and large language models (with Ivan Vulic, see ESSAI'23). Additionally, I serve as an equal opportunity officer, as a mentor for student research workshops, and as an organizer/advisor for the UnImplicit workshop series.

Prior to my current position in Stuttgart, I contributed to the Excellence Cluster MMCI in Saarbrücken, where I was also an associated member of SFB 1102, collaborating on interdisciplinary studies concerning the modeling of script knowledge and its implications for language comprehension (work with Manfred Pinkal). From 2015 to 2017, I carried out a DFG Research Fellowship in Edinburgh and Illinois, where I advanced machine-learning models for semantic tasks, specifically those that integrate contextual meaning (hosted by Mirella Lapata and Dan Roth). Before that, I was the technical lead for NLP in an EC FP7 cooperation project on software engineering (work with Ewan Klein). I received a PhD in 2013 from the University of Heidelberg, where I explored the modeling of implicit language phenomena (advised by Anette Frank and Michael Strube).