About me
I’m a PhD student at Bar-Ilan University, I’m co-supervised by Professors Ido Dagan and Yoav Goldberg. My research interests lie in Natural Language Processing, and more specifically in semantics and information decomposition.
Specificaly, I study how different events and entities mentioned in text relate to each other. I’m especially interested in relations that are not overtly stated in the text but are commonly understood by reading it. Such relations are often termed as implicit, and while they may seem quite obvious to our human perception, tend to be more difficult to comprehend for machines.
On a broader aspect, I’m interested in decomposing information expressed in text into minimal units that can be worked with and composed back for various applications. I’m exploring ways that a fine-grained information decomposition scheme can be utilized for robust machine comprehension, summarization and entailment.
My published works
- Asking It All – Generating a question based on a key piece of information that you want to find out.
- Controlled Crowdsourcing – Crowdsourcing semantic relation datasets with high coverage and precision.
- QA-Align – Representing and identifying aligned propositions within overlapping sentences.
Ongoing directions
- Robustifying reading comprehension models with implicit semantic relations
- Fine-grained alignment between documents and their summaries using propositions
- Expanding open semantic representations (QA-sem) to other lexical units.
Miscellaneous News
- Outstanding Reviewer in EMNLP 2021!
Teaching
- Introductory machine learning course for the neuroscience department
- Advanced Programming II
- Intro to Probability for Computer-Science.