The MOLOKO grammar
Version 2.2 (December 2005)
Introduction
The MOLOKO grammar defines a natural language grammar for a range of phenomena in the domain of "talking about things to make and do." MOLOKO is specified as a Combinatory Categorial Grammar, which can be processed with the OpenCCG development platform. OpenCCG enables you to use MOLOKO both for parsing (analysis) and for realization (production). For brief documentation of the grammar, see here
Download
The MOLOKO grammar is provided as open-source software, under the GNU Lesser General Public License.
| Required: | Java 1.5 (or higher), Apache Ant, OpenCCG 0.8.6 (or higher) |
| Author: | Geert-Jan M. Kruijff, Sabrina Wilske (mail for comments, questions, suggestions) |
| Download: | moloko.dec05.zip (922KB. December 12 2005) |
| Install: | Unzip the downloaded archive. Run build.sh to start the Antishield installation shield. |
| Platforms: | Platform independent. Tested on Mac OS.X 10.4.3, Linux (SuSe, RedHat, CentOS 4.1) |
Brief documentation for the MOLOKO grammar
The grammar distribution consists of two parts: (1) a core grammar [ grammars/openccg/core/ ], and (2) the MOLOKO grammar [ grammars/openccg/moloko/ ]. The MOLOKO grammar primarily focuses on expressions that are used in the context of talking about objects and their properties, spatial organization of objects in a local visual scene, as well as rooms. The grammar handles basic forms of assertions, commands, and questions (factive as well as polar). Sample interactions for the use of the MOLOKO grammar in the setting of supervised spatial exploration can be found in this PDF. Most of the families for the MOLOKO grammar are described in this PDF (draft!), though please note that since the creation of that draft document some structures have slightly changed. For the latest documentation, please refer to the documentation in the grammar specifications.The CORE grammar
The core grammar defines basic atomic categories, lexical families, and macros for SVO languages:
| cats.xsl | Atomic categories |
| dict.xsl | Lexical macros |
| adj.xsl | Lexical families: adjectives |
| adv.xsl | Lexical families: adverbials |
| coord.xsl | Lexical families: coordination |
| cue.xsl | Lexical families: cue words |
| det.xsl | Lexical families: determiners |
| expl.xsl | Lexical families: expletives |
| n.xsl | Lexical families: nouns |
| neg.xsl | Lexical families: negation |
| q.xsl | Lexical families: questions |
| v.X.Y.xsl | Lexical families: verbs (modal, copula, auxiliary, infinitival; intransitive, transitive) |
The file core/lexicon.xsl includes a full list of the kinds of lexical families. The MOLOKO grammar references this file, to (re)use the families specified in the core.
The MOLOKO grammar
In addition, the MOLOKO grammar specifes lexical families for the following types:
| ctxtref.xsl | Lexical families: contextual reference |
| du.xsl | Lexical families: discourse units |
| pp.xsl | Lexical families: prepositional phrases |
| ppspat.xsl | Lexical families: spatial prepositions |
| ctxtref.xsl | Lexical families: punctuation (comma categories) |
The file moloko/dict.xsl specifies the complete dictionary for the MOLOKO grammar. Example expressions are provided in the moloko/testbed.xml test suite. The (flat) ontology currently being used in the grammar is given in moloko/types.xsl.
