LINGUIST List 30.4415

Thu Nov 21 2019

Confs: Computational Linguistics/Germany

Editor for this issue: Everett Green <everettlinguistlist.org>

---

---

Date: 15-Nov-2019
From: Fischer Patricia <patricia.fischeruni-tuebingen.de>
Subject: Workshop on Data-driven Approaches to Parsing and Semantic Composition
E-mail this message to a friend

Workshop on Data-driven Approaches to Parsing and Semantic Composition
Short Title: DatParSem


Date: 10-Dec-2019 - 11-Dec-2019
Location: Fürstenzimmer, Schloss Hohentübingen, Tübingen, Germany
Contact: Patricia Fischer
Contact Email: < click here to access email >
Meeting URL: https://parsingsemcompworkshop.github.io/

Linguistic Field(s): Computational Linguistics

Meeting Description:

Topic:

Syntactic parsing assigns structure to phrases. Composition models, on the other hand, assign meaning vectors to phrases. Parsers are known to benefit from distributional word representations while composition models can again profit from (morpho-)syntactic information. Given this interplay of parsing and composition, we would like to investigate in this workshop whether parsers and models of distributional semantics can mutually inform each other above the word level.

Questions of Interest:

- How can distributional representations of phrases be created?
- What kind of semantic and/or syntactic information do these representations encode?
- What are possible underlying units beyond single words which can be represented via composition?
- What kind of parsing formalisms benefit composition models for phrases and what kind of phrase representations benefit parsing?
- Which criteria do corpora have to fulfill to be most suited as a data resource for distributional methods?
- Do we need semantic composition for (syntactic) parsing?

Invited Talks:

- Bernd Bohnet (Google AI, London)
Linguistic representations in neural network times

- Miryam de Lhoneux (Uppsala University)
Do we need recursive subtree composition in dependency parsing?

- Ellie Pavlick (Brown University)
What should constitute natural language `understanding’?

- Vered Shwartz (Allen Institute for Artificial Intelligence, Seattle)
Not a piece of cake: on lexical composition and implicit information

- Workshop website: https://parsingsemcompworkshop.github.io
- Workshop program: https://parsingsemcompworkshop.github.io/pdfs/workshopProgram.pdf
- Contact e-mail: parsemcompgmail.com (mailto:parsemcompgmail.com)

The workshop is free of charge. The workshop dinner will cost ~20-25€.
This workshop will be open to all who are interested. However, since seating capacity is limited, we would like you to ***register by December 01*** via our registration form:
https://docs.google.com/forms/d/1fLN-EQ_d68cEs_Q31sk4x5caQknMjATVjwKiooIjD1Y/edit

Page Updated: 21-Nov-2019