Avirup (Avi) Sil  Avirup (Avi) Sil photo       

contact information

Research Staff Member & Chair, NLP Professional Community
IBM Research AI, New York, USA


Professional Associations

Professional Associations:  Association for Computational Linguistics


About me: 
Avi is a Research Scientist in the Information Extraction and NLP group at IBM Research AI. He is also the Chair of the NLP professional community of IBM. His research focusses on making AI systems understand the meaning of natural language text. Most of his work has been on making sense of large unstructured web data applying machine learning algorithms: Extracting entities (names of persons, organizations, etc.), resolving them to big knowledge-bases (like Wikipedia) and extracting relationships between the entities.

Here's a recent article about his research on Entity Linking.

Avi completed his PhD in Computer Science under the supervision of his thesis advisor Alexander Yates. He also worked on Temporal Information Extraction in the Machine Learning Group at Microsoft Research, Redmond managed by Chris Burges and John Platt. His mentor was Silviu Cucerzan. 

Latest News:

  1. (new) Area chair for COLING 2018
  2. (new) Area chair for Information Extraction at NAACL 2018
  3. (new) AAAI 2018 Paper accepted: Neural Cross-lingual Entity Linking
  4. (new) Organizing ACL 2018 Workshop on "Relevance of Linguistic Structures in Neural NLP".
  5. (new) Best Score: English Entity Discovery and Linking. TAC 2017.
  6. EMNLP 2017: Slot Filling with Neural Attentive models
  7. ACL 2016 Papers: 1. Language Independent Entity Linking 2. Liberal Information Extraction 

Research Interests: 

Information Extraction from the Web                         Natural Language Processing
Machine Learning                         Information Retrieval

Some of My Research Areas:

  • Entity Extraction and Disambiguation: Open-Database techniques for Entity Extraction and Disambiguation (CIKM'13 & EMNLP'12 papers): My NER-EL system outperforms 2 state-of-the-art NER systems and 6 EL systems.
  • Temporal Information Extraction: Algorithms for temporal scoping of relations between entities (TAC and CoNLL paper). Our system achieves state-of-the-art results by outperforming 4 other systems.
  • Learning Action Representations, relation extraction from text (papers on STRIPS extractions): Perhaps, the first to extract preconditions, add and delete effects from text automatically.
  • Education Data Mining: Automated scoring of explanations, responses and essays, automatic graders of scientific inquiry, using statistical NLP in education data mining (NAACL-BEApaper)

 Previous Internships/Collaboration:

  1. Summer 2013: Research Intern, Microsoft Research, Redmond.
    • Worked with the Machine Learning Group at MSR, Redmond.
    • Research project: Performed research on Temporal Slot Filling (TSF). Obtained Best Score at TAC 2013.
  2. Spring 2012: Research Assistant, Temple University (in collaboration with Yahoo! Research).
    • Worked with members of Yahoo! Labs at Sunnyvale (supported by a gift from Yahoo!).
    • Research project: Open-Database Entity Linking. Project resulted in a paper at EMNLP'12.
  3. Summer 2011: Data and Technology Analyst Intern, Morgan Stanley.
    • Worked with the Investment Management (MSIM FI) team on the Trade Acknowledgement Processor.
    • Project: Developed a system which is currently running live to parse the financial data.

Academic Service:

  • Conference Program Committee Member and/or Reviewer:
    • ACL
    • NAACL
    • EMNLP
    • RANLP
    • AAAI
    • IJCAI
    • CONLL
    • COLING
    • LREC