Difference: WritingCLMSThesisOnTheGrammarMatrix ( vs. 1)

Revision 12014-12-20 - trimblet

Line: 1 to 1
META TOPICPARENT name="GrammarEngineeringFAQ"
The following is meant to be a rough guide of what to expect when writing a thesis on the Grammar Matrix along with some helpful tips and hints to get started on the right things in the right order to help ease friction.

Read this page

Define your question

  • What are you doing?

Define the scope of research

  • What is the end goal?

Get set up with LaTeX and the UW thesis file

  • If not familiar with LaTeX, practice!
  • Analyze previous Grammar Matrix theses for figures/code desired
    • Look at Drellishak, Saleem, Crowgey, Song, Trimble, etc.

Seek out thorough understanding of basic grammatical facts

  • What is the phenomenon you are studying? What is the terminology?
  • What is the range of phenomena? What differences are there between languages? How does the phenomena differ in syntax, morphology, and semantics, and do these intersect?

Conduct typology review

  • What is the range of variation in the phenomenon across the world’s languages?
  • Look at data from various languages: look for things of interest that typologists haven’t seen
  • Look at existing HPSG grammars for interesting phenomena

Write typology review section

  • Keep lots of extra notes on interesting languages, facts, phenomena, etc.

Define target semantic representations

  • What does the output MRS look like?
  • What does the output syntax tree look like?

Define scope of project

  • New library? Update existing libraries? Both?
  • New page? Update existing page? Both?
  • New HPSG features? New HPSG types (with existing features)? Both? Neither? (see below)

Set up testing ubuntu environment

Set up coding environments

  • Set up to modify matrix SVN
  • Run regression tests (python matrix.py r)
  • Create new branch
  • Explore existing libraries
    • Understand pipeline

Create some test suites

  • Create test suites for pseudo languages and development languages

Do HPSG analysis

  • Look at ERG, jacy, gCLIMB, and CoLLAGE grammars for various analyses
  • Understand what already exists in matrix.tdl and mylanguage.tdl
  • Define types to be implemented
    • Discern whether these types will be language specific in language.tdl or in matrix.tdl
  • Find out what relevant code (if any) already exists (for instance, in morphotactics.py)

Implement customization system changes

  • Update matrixdef file

Implement HPSG analysis

  • Create grammars by hand first, testing against test suites
  • Add or change existing library modules
    • Lots and lots of comments!
  • Use constructor to build TDL

Write introduction, analysis, implementation sections

  • Follow convention

Finish HPSG analysis

  • Create choices files to verify correct output
    • Compare to test suites
  • Repeat previous step…
  • Test all regression tests!
  • Ask for feedback from Matrix developers
  • Create held_out languages test suites / choices files
    • Compare to test suites and get evaluation numbers

Finish paper

  • Write evaluation and conclusion
  • Revise
  • Leave plenty of time for revisions from advisor and reader

Write matrixDoc pages!

  • Flesh out old matrixDoc pages

Update "Writing a CLMS thesis on the Grammar Matrix”

  • :-)
This site is powered by the TWiki collaboration platformCopyright & by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback
Privacy Statement Terms & Conditions