Séminaire de Recherche en Linguistique

In this talk I will highlight some of the key technologies at Google that rely on the automatic syntactic analysis of text, including search quality, question answering and machine translation. While the use of syntactic analyzers has now become common place in many user facing language technologies, this has not always been the case for a variety of reasons. Key amongst them, is the lack of resources for languages outside of English. Furthermore, even for languages that had sufficient resources to build analyzers, the annotation schemes employed were often drastically different, making it hard for downstream technologies to adapt. This has motivated the creation of the Universal Dependency Treebank project, which is a consortium of industrial and academic researchers that aim to build comparable syntactic treebanks across a variety of languages. I will describe our annotation schema and progress, highlighting key challenges and decisions to make annotations consistent across typologically and morphologically divergent languages.