goodmami.org

I'm now back in Keihanna, after going back to Seattle for about two months.

I'm taking an internship at NICT, working on generation for the Jacy Japanese grammar. I will be here for approximately 3 months, and hope to have a paper about the research submitted to a conference (ACL 2009, perhaps) or journal by that time.

After a very long (but relatively pleasant) flight, I'm here in Keihanna, Japan for the HPSG 2008 conference and Delph-In 2008 summit. I don't give my talk until August 1st, but I'm excited to hear from (and meet) other researchers.

I knew that I would be dealing with the generation of incorrect lexical items, but I'm starting to realize that there may be an unwieldy number of them being generated. Instead of initially creating as many new items as possible and filtering later, it is looking like a better idea to be selective from the beginning and only translate words that fit some criteria.

In order for my project to be able to handle non-Matrix derived grammars, I must provide a typemap that compares the types of the grammar to the Matrix types. The Jacy grammar is a kind of variant of the latest version of the Matrix, so it shouldn't be too hard. However I run into a problem when mapping things like verbs.

The title of this post is basically the working title of my thesis project for University of Washington's CLMA (Computational Linguistics Master of Arts) program. My adviser is Dr. Emily Bender (hereafter Emily), and the research period is from now (spring quarter, 2008) to the end of summer quarter, 2008.