Ashutosh Modi

This webpage is deprecated, check the latest webpage here

Researcher at Department of Computational Linguistics and MMCI Cluster of Excellence, University of Saarland, Computer Science Department.
Office Building C7 4, Room 3.17 phone: + 49 (681) 302 70029 e-mail: ashutosh-at-coli.uni-saarland.de Postal address Saarland University Building C7 4, Room 3.17 66123 Saarbrücken, Germany Curriculum Vitae CV

Research

I research on modeling common sense knowledge about the world in form of Scripts. By scripts, I do not mean python scripts or movie scripts but Scripts are sequence of events representing a stereotypical activity for e.g. going to restaurant or making a coffee.

I use natural language processing and machine learning techniques (including Deep Learning) for modeling scripts.

My research focuses on modeling scripts in terms of events governing the script. I try to model good representations for events, which I refer as event embeddings (event vectors). The learned event embeddings capture the semantics of the event and should be find applications beyond scripts as well.

I finished my PhD under the supevision of Dr. Ivan Titov and Prof. Dr. Manfred Pinkal

I am associated with SFB 1102.

Publications

• Modeling Common Sense Knowledge via Scripts,
  Ashutosh Modi,
  PhD Thesis 
  Link   Bibtex
  
• Modeling Semantic Expectations: Using Script Knowledge for Referent Prediction,
  Ashutosh Modi, Ivan Titov, Vera Demberg, Asad Sayeed and Manfred Pinkal,
  Transactions of Association for Computational Linguistics (TACL) 2017 
  PDF   Bibtex
  
• Does discourse predictability affect rate of pronominalization?,
  Ekaterina Kravtchenko, Ashutosh Modi, Ivan Titov, Vera Demberg, Asad Sayeed and Manfred Pinkal,
  30th CUNY Conference on Human Sentence Processing, MIT, Cambridge MA, 2017 
  Bibtex
  
• A Mixture Model for Learning Multi-Sense Word Embeddings,
  Dai Quoc Nguyen, Dat Quoc Nguyen, Ashutosh Modi, Stefan Thater and Manfred Pinkal,
  Lexical and Computational Semantics (* SEM 2017) 
  PDF   Bibtex
  
• Event Embeddings for Semantic Script Modeling,
  Ashutosh Modi,
  Eighteenth Conference on Computational Natural Language Learning (CoNLL-2016). Berlin, Germany 
  PDF   Bibtex
  
• InScript: Narrative texts annotated with script information,
  Ashutosh Modi, Tatjana Anikina, Simon Ostermann and Manfred Pinkal,
  Proceedings of the Tenth International Conference on Language Resources and Evaluation (LREC 2016) 
  PDF   Bibtex   Corpus
  
• Learning to predict script events from domain-specific text,
  Rachel Rudinger, Vera Demberg, Ashutosh Modi, Benjamin Van Durme and Manfred Pinkal,
  Lexical and Computational Semantics (* SEM 2015) 
  PDF   Bibtex
  
• Inducing Neural Models of Script Knowledge,
  Ashutosh Modi and Ivan Titov,
  Eighteenth Conference on Computational Natural Language Learning (CoNLL-2014). Baltimore, U. S. A 
  PDF   Bibtex
  
• Learning Semantic Script Knowledge with Event Embeddings,
  Ashutosh Modi and Ivan Titov,
  International Conference on Learning Representations (ICLR 2014), workshop track, selected for oral presentation. 
  PDF   Bibtex   Talk
• Unsupervised Induction of Frame-Semantic Representations,
  Ashutosh Modi, Ivan Titov and Alexandre Klementiev,
  NAACL-HLT 2012 Workshop on Inducing Linguistic Structure, Montreal, Canada 2012
  PDF   Bibtex   Poster