Jekyll2026-02-12T01:09:00+00:00http://dmacjam.github.io/feed.xmljakub@macina.skJakub Macina is a Machine Learning Researcher.Jakub Macinajakub@macina.sk[Award] Forbes 30 under 30: Science & Education2023-05-02T00:00:00+00:002023-05-02T00:00:00+00:00http://dmacjam.github.io/awards/forbes-30u30Selected for the Forbes 30 Under 30 list in the category Science & Education.

]]>Jakub Macinajakub@macina.sk[Paper] EACL2023 Opportunities and Challenges in Neural Dialog Tutoring2023-04-01T00:00:00+00:002023-04-01T00:00:00+00:00http://dmacjam.github.io/research/dialog-tutoring-paperJakub Macina*, Nico Daheim*, Lingzhi Wang, Tanmay Sinha, Manu Kapur, Iryna Gurevych, Mrinmaya Sachan. In Proceedings of the 17th Conference of the European Chapter of the Association for Computational Linguistics, pages 2357–2372, Dubrovnik, Croatia. Association for Computational Linguistics.

Designing dialog tutors has been challenging as it involves modeling the diverse and complex pedagogical strategies employed by human tutors. Although there have been significant recent advances in neural conversational systems using large language models and growth in available dialog corpora, dialog tutoring has largely remained unaffected by these advances. In this paper, we rigorously analyze various generative language models on two dialog tutoring datasets for language learning using automatic and human evaluations to understand the new opportunities brought by these advances as well as the challenges we must overcome to build models that would be usable in real educational settings.We find that although current approaches can model tutoring in constrained learning scenarios when the number of concepts to be taught and possible teacher strategies are small, they perform poorly in less constrained scenarios.Our human quality evaluation shows that both models and ground-truth annotations exhibit low performance in terms of equitable tutoring, which measures learning opportunities for students and how engaging the dialog is.To understand the behavior of our models in a real tutoring setting, we conduct a user study using expert annotators and find a significantly large number of model reasoning errors in 45% of conversations. Finally, we connect our findings to outline future work.

Oral presentation at EACL 2023 in Dubrovnik as a main conference paper:

]]>Jakub Macinajakub@macina.sk[Paper] 2nd Math-AI Workshop at NeurIPS22: Automatic Generation of Socratic Questions for Learning to Solve Math Word Problems2022-12-01T00:00:00+00:002022-12-01T00:00:00+00:00http://dmacjam.github.io/research/neurips22-mathaiJakub Macina*, Kumar Shridhar*, Mennatallah El-Assady, Tanmay Sinha, Manu Kapur, and Mrinmaya Sachan.

Presented at NeurIPS 2022 in New Orleans at 2nd MATH-AI Workshop:

]]>Jakub Macinajakub@macina.sk[Paper] EMNLP22 Automatic Generation of Socratic Subquestions for Teaching Math Word Problems2022-10-01T00:00:00+00:002022-10-01T00:00:00+00:00http://dmacjam.github.io/research/socratic-questions-paperJakub Macina*, Kumar Shridhar*, Mennatallah El-Assady, Tanmay Sinha, Manu Kapur, and Mrinmaya Sachan. In Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing, pages 4136–4149, Abu Dhabi, United Arab Emirates. Association for Computational Linguistics.

We explore the ability of large language models (LMs) in generating sequential questions for guiding math word problem-solving. We propose various guided question generation schemes based on input conditioning and reinforcement learning (RL) and found that on both automatic and human quality evaluations, LMs constrained with desirable question properties generate superior questions and improve the overall performance of a math word problem solver.

Presented at EMNLP 2022 in Abu Dhabi as a main conference paper:

]]>Jakub Macinajakub@macina.sk[New position] Doctoral Fellow at ETH AI Center2021-09-01T00:00:00+00:002021-09-01T00:00:00+00:00http://dmacjam.github.io/research/eth-ai-centerI’m starting my Doctoral Fellowship at the ETH AI Center working on the intersection of Natural Language Processing (NLP) and Learning Sciences. As ETH’s central hub for artificial intelligence, the ETH AI Center fosters research excellence, industry innovation, and AI entrepreneurship to promote trustworthy, accessible, and inclusive AI systems.

]]>Jakub Macinajakub@macina.sk[Attending] NeurIPS 20202020-12-06T00:00:00+00:002020-12-06T00:00:00+00:00http://dmacjam.github.io/research/neurips20I attended NeurIPS 2020 research conference which was held virtually.

]]>Jakub Macinajakub@macina.sk[Attending] RecSys 20202020-09-22T00:00:00+00:002020-09-22T00:00:00+00:00http://dmacjam.github.io/research/recsys20I attended RecSys 2020 research conference which was held virtually.

]]>Jakub Macinajakub@macina.sk[Talk] Vienna Deep Learning Meetup2019-09-24T00:00:00+00:002019-09-24T00:00:00+00:00http://dmacjam.github.io/talk/deep-learning-viennaTalk: Deep Learning for Recommender Systems by Jakub Mačina, Machine Learning Engineer, Exponea

Recommender systems are driving business value through personalisation for customers of Amazon, Netflix or Spotify. This talk will provide an overview of traditional and deep learning recommender system approaches and highlight the challenges encountered by industry practitioners such as extreme data sparsity. Real-world case study will show how to capture users varying tastes and products into a dense (latent) embeddings representation in order to design a scalable recommender system architecture.

Find out more about the event here:

]]>Jakub Macinajakub@macina.sk[Attending] RecSys 20192019-09-16T00:00:00+00:002019-09-16T00:00:00+00:00http://dmacjam.github.io/research/recsys19I attended RecSys 2019 research conference in Copenhagen, Denmark.

]]>Jakub Macinajakub@macina.sk[Attending] RecSys 20182018-10-02T00:00:00+00:002018-10-02T00:00:00+00:00http://dmacjam.github.io/research/recsys18I attended RecSys 2018 research conference in Vancouver, Canada.

]]>Jakub Macinajakub@macina.sk