讲座题目:Building Spoken Dependency Treebanks
讲座学者:巴黎三大 Kim Gerdes
讲座时间:2017年4月27日(周四)10:00-11:30
讲座地点:浙江大学紫金港校区东5-201(东)
欢迎广大师生积极参加!
Abstract:
The talk will give an overview of the development of oral dependency treebanks. We will attempt to address practical and theoretical considerations that have to be tackled during the treebank creation process: How to define the hypotheses that we want to test by means of the treebank? What to transcribe? How to transcribe the spoken text? What is a dependency annotation and why treebanks should be dependency-based, in particular for spoken texts? What are the choices of dependency annotation? How to fix the granularity of POS tags and functions? How to get started with the annotation and how to set up a bootstrapping approach to tagging and parsing? How well does this work for different languages, in particular French and Chinese.
Kim Gerdes, an associate professor at the Sorbonne Nouvelle in Paris, works on dependency syntax, corpus tools, and text generation. He is a cofounder of the conference cycle “Dependency Linguistics” (since 2011, depling.org) and has developed dependency annotations of spoken texts in various national and international projects (Rhapsodie, Orféo, Naijasyncor, and Hubert Curien partnership with the City University of Hong Kong). He also works on the development of tools for dependency annotation (arborator.ilpga.fr) and for web-corpus collection (gromoteur.ilpga.fr). He also collaborates with cloem.com on Artificial Creativity and Natural Language Generation of juridical texts.
外语学院
2017年4月20日