AspCcgTk


This page is about AspCcgTk - the Answer Set Programming Combinatory Categorial Grammar Toolkit. AspCcgTk is a parser based on Combinatory Categorial Grammar (CGC) developed using declarative programming paradigm -- Answer Set Programming. It implements wide-coverage CCG parsing by utilizing the CCG postagger and supertagger of the C&C tool. AspCcgTk aims at producing all semantically distinct parse trees for a given sentence. Compared to typical approaches to CCG parsing, no parsing algorithm is implemented within this tool. Instead, the task of CCG parsing is formulated as a planning problem and answer set programming systems are used to compute solutions that correspond to valid parses. AspCcgTk can also be seen as a basis for a generic tool for encoding different CCG category and rule sets in a declarative and straightforward manner. In other words, AspCcgTk provides a test-bed for experimenting with different theoretical CCG frameworks without the need to craft specific parsing algorithms. AspCcgTk also contains a module for visualizing CCG derivations found by a parser.

Table of Contents

Publications

[LPNMR13] Flexible Combinatory Categorial Grammar Parsing using the CYK Algorithm and Answer Set Programming (PDF), Peter Schüller, 12th International Conference on Logic Programming and Nonmonotonic Reasoning (LPNMR), 2013, to appear.

[IWCS13] Towards a Tight Integration of Syntactic Parsing with Semantic Disambiguation by means of Declarative Programming (PDF), Yuliya Lierler and Peter Schüller, 10th International Conference for Computational Semantics (IWCS), 2013, to appear.

[CR12] Parsing Combinatory Categorial Grammar via Planning in Answer Set Programming (PDF), Yuliya Lierler and Peter Schueller, Correct Reasoning - Essays on Logic-Based AI in Honour of Vladimir Lifschitz, Esra Erdem, Joohyung Lee, Yuliya Lierler and David Pearce (editors), volume 7265 of LNCS, pgs. 436-453, 2012.

[WLP11] Parsing Combinatory Categorial Grammar with Answer Set Programming: Preliminary Report (PDF), Yuliya Lierler and Peter Schueller, 25th Workshop on Logic Programming (WLP), 2011.

Download (full toolkit package)

If you want to try AspCcgTk, please download the full package containing all related libraries and other tools (IDPDraw has been patched to display strings, BioASP has been extended for better parsing and optimization capabilities).

(Deprecated package versions do not include C&C, but you can extract it from the most recent package).

Package downloads:

Overview of package contents

The following is a brief list of some important files in the package:

README
A file that should be read.
parse_and_display.py
Main entry point to AspCcgTk python toolkit for CCG parsing and visualizing. Can be configured from commandline and within the source.
aspccgtk.py
The AspCcgTk toolkit core library.
encodings/cyk{1,2,3,4,5,6,7}*.asp
ASP encoding of the CCG parser based on CYK, see [LPNMR13]
encodings/cykccg.asp
Deprecated ASP encoding of the CCG parser based on CYK, see [IWCS13]
encodings/cykccg2idpdraw.asp
ASP encoding for visualizing answer sets of CYK parser encodings using IDPDraw. (Note that the answer set of cykccg2idpdraw.asp are postprocessed by aspccgtk.py to create human readable CCG category labels from function symbols.)
encodings/ccg.asp
ASP encoding of the CCG parser based on planning, see [CR12]
encodings/ccg2idpdraw.asp
ASP encoding for visualizing answer sets of ccg.asp using IDPDraw. (Note that the answer set of ccg2idpdraw.asp are postprocessed by aspccgtk.py to create human readable CCG category labels from function symbols.)

How to use the package

Please have a look at the README file in the package which contains step-by-step instructions for building IDPDraw and experimenting with AspCcgTk, including example files contained in the package.

In case of problems please contact the authors via email (addresses can be found in the COPYRIGHT file in the package).

Example visualizations

I file without reading the book.

Single derivation:
derivation image

Ann saw yesterday and proved today completeness.

Derivation 1:
derivation image

Derivation 2:
derivation image

Derivation 3:
derivation image

Third Party Software Used

AspCcgTk uses the following libraries and/or tools developed by others:

Authors

AspCcgTk has been developed by

$Id: index.html 1233 2013-06-24 10:25:23Z ps $