Skip to content

donahut/simple-parser

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

3 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

simple-parser

Author: Thomas Donahue

Email: [email protected]

This was built as part of an assignment for a course, Situated Natural Language Dialogs with Robots, that I took while studying at Tufts University -- Spring 2013.

Description

A Clojure library designed to parse sentences using combinatory categorical grammars. I have utilized a naive breadth first approach, which is guaranteed to find a valid parse if one is possible (given the lexicon), though it gets extremely large very fast. For more infomation on CCG Parsing, see:

http://en.wikipedia.org/wiki/Combinatory_categorial_grammar

Usage

java -jar simple-parser-0.1.0-standalone.jar [args]

arg1 - Sentence

Choose which sentence to parse. Options below:

1 - "the dog bit John"

2 - "Andie saw Steve"

3 - "the doctor sent for the patient arrived"

arg2 - Vision

Choose whether the "vision system" constrains the parse. This only works for sentence 2. In the example case, an actual "saw" has been identified in vision, and thus constrains the lexicon to remove the verb possibility for "saw." You will see no effect from turning on vision for sentences 1 and 3.

0 - Off 1 - On

Examples:

java -jar simple-parser-1.0-standalone.jar 1 0

java -jar simple-parser-1.0-standalone.jar 2 1

Notes:

What is printed is the entire workspace after each shift and reduce cycle. For sentence 3, it is virtually impossible to follow what is going on, but sentences 1 and 2 are manageable. After all of the workspaces have printed, the total number of stacks on the workspace is printed. This helps see how the situated "vision" example with sentence 2 constrains the parse. After this number, a list of the valid parses is printed. There is an (S) printed for every valid possible parse, if empty, there were no valid parses.

About

Simple CCG parser for NLU course

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published