Summer 2017
Summer 2017
Date | Event | Speaker | Abstract/Details |
| 07/18/2017 | Thesis defense: Natural Language Understanding: Deep Learning for Abstract Meaning Representation | Bill Foland | In the last few years there have been major improvements in the performance of hard natural language processing tasks due to the application of artificial neural network 91أغجز¸َ. These 91أغجز¸َ replace complex hand-engineered systems for extracting and representing the meaning of human language with systems which learn features based on processing examples of language. In this dissertation, I present deep neural networks for semantic role labeling, and then for Abstract Meaning Representation parsing, and a novel Distributed Abstract Meaning Representation, or DAMR. I then describe a model used to create fixed vector representations of sentence meaning from DAMR. Finally, I use natural language inference to test the quality of the meaning content of these fixed vectors. |
| 07/19/2017 | The ACL 2017 UD shared task | Jan Hajic | Universal Dependencies (UD) is a project that is developing cross-linguistically consistent treebank annotation for many languages, with the goal of facilitating multilingual parser development, cross-lingual learning, and parsing research from a language typology perspective. The annotation scheme is based on an evolution of (universal) Stanford dependencies (de Marneffe, Manning et al., 2006-2014), Google universal part-of-speech tags (Petrov et al., 2012), and the Interset interlingua for morphosyntactic tagsets (Zeman, 2008). The general philosophy is to provide a universal inventory of categories and guidelines to facilitate consistent annotation of similar constructions across languages, while allowing language-specific extensions when necessary. First release was in 2014, with bi-annual updates and additions (now at 65 treebanks). To prove the point, the UD initiative, building on the 10-year history of CoNLL Shared Tasks on parsing, organized a shared task centered around the UD treebanks (Hajiؤچ et al., 2017). The focus of the 2017 task was to develop syntactic dependency parsers that can work in a real-world setting, starting from raw text, and that can work over many typologically different languages, even surprise languages for which there is little or no training data, by exploiting a common syntactic annotation standard. For the Shared Task, the Universal Dependencies version 2 (UD v2) annotation scheme was used, which is the latest version. |