Natural Language Inference for Humans

Profile photo of Valaria de Paiva

Event details

Wednesday 15 July 2020
Online event,
Access to the session will open 30 minutes prior to the seminar
Join the session


Speaker Bio:

Valeria de Paiva is a mathematician and AI scientist based on the Bay Area, CA, interested in NLP and Inference and all kinds of semantics. She works for a not very-stealthy start-up project, called the Topos Institute. Before that she was at Samsung Research America, Nuance, Deem, Cuil and for many years at Xerox PARC. She is very keen on making sure that women are not short-changed in their professional lives. For that she maintains a facebook group Women in Logic, a blog Women in Logic, helps with the ACM-W Scholarship program ACM-W Scholarships and initiated and helps to organize the Workshop Women in Logic, now in its 4th year.


One hears much about the incredible results of recent neural nets methods in NLP. In particular much has been made of the results on the Natural Language Inference task using the huge new corpora SNLI, MultiNLI, SciTail, etc, constructed since 2015. Wanting to join in the fun, we decided to check the results on the corpus SICK (Sentences Involving Compositional Knowledge), which is two orders of magnitude smaller than SLNI and presumably easier to deal with.

We discovered that there were many results that did not agree with our intuitions. As a result, we have written so far five papers on the subject (with another one submitted to COLING2020). I want to show you a potted summary of this work, to explain why we think this work is not near completion yet and how we're planning to tackle it. This is work with Katerina Kalouli, Livy Real, Annebeth Buis and Martha Palmer. The papers are:

  • Explaining Simple Natural Language Inference. Proceedings of the 13th Linguistic Annotation Workshop (LAW 2019), 01 August 2019. ACL 2019,
  • WordNet for “Easy” Textual Inferences. Proceedings of the Globalex Workshop, associated with LREC 2018
  • Graph Knowledge Representations for SICK. informal Proc of 5th Workshop on Natural Language and Computer Science, Oxford, UK, 08 July 2018
  • Textual Inference: getting logic from humans. Proc of the 12th International Conference on Computational Semantics (IWCS), 22 September 2017
  • Correcting Contradictions. Proc of Computing Natural Language Inference Workshop (CONLI 2017) @IWCS 2017


A world top-100 university

We're a world top-100 university renowned for the excellence, impact and distinctiveness of our research-led learning and teaching.