BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Natural Language Understanding and Generation with Abstract Meanin
 g Representation - Marco Damonte\, the University of Edinburgh
DTSTART:20191121T110000Z
DTEND:20191121T120000Z
UID:TALK135109@talks.cam.ac.uk
CONTACT:Qianchu Liu
DESCRIPTION:In this talk\, I will discuss my recent work on parsing and ge
 neration with Abstract Meaning Representation (AMR). AMR is a semantic rep
 resentation for natural language that represents sentences as graphs\, whe
 re nodes represent concepts and edges represent semantic relations between
  them. Sentences are represented as graphs and not trees because nodes can
  have multiple incoming edges\, called reentrancies. These are due to seve
 ral linguistic phenomena such as control\, coreference\, and coordination.
  \n\nI will present my work on AMR parsing (from text to AMR) and AMR-to-t
 ext generation (from AMR to text). For the parsing task\, we showed that i
 t is possible to use techniques from tree parsing and adapt them to parse 
 AMR graphs. To better analyze the quality of AMR parsers\, we developed a 
 set of fine-grained metrics to better assess parsers\, including a metric 
 for reentrancy prediction. Hence\, we performed a study of the main causes
  of reentrancies in AMR and their impact on performance. For the generatio
 n task\, we showed that neural encoders that have access to reentrancies o
 utperform those who do not\, demonstrating the importance of reentrancies 
 also for generation.\n\nI will also discuss the problem of using AMR for l
 anguages other than English. Annotating new AMR datasets for other languag
 es is an expensive process and requires defining ad-hoc annotation guideli
 nes for each new language. It is therefore reasonable to ask whether we ca
 n share AMR annotations across languages.
LOCATION:Board room\, Faculty of English\, 9 West Rd (Sidgwick Site)
END:VEVENT
END:VCALENDAR