This system is a resource designed to disambiguate formally ambiguous pronominal sentences and propose the most likely semantic interpretation (passive, anticausative, reflexive, etc.).
EnTxala, EsTxala and CaTxala are open-source dependency grammars for English, Spanish and Catalan, respectively, implemented in FreeLing environment, a library of Natural Language Processing tools. These grammars have been designed as a resource for NLP applications that require a deep and robust syntax representation in natural language automatic analysis. They have been developed as broad-coverage grammars that provide a unique dependency parse tree for each input sentence, specifying attachments for any syntactic structure and assigning syntactic functions to each dependency arch. Both processes are carried out by a set of hand-writted rules based on linguistic knowledge 2,954 rules in EnTxala, 4,447 rules in EsTxala and 3,011 rules in CaTxala).
The SRG is an open-source multi-purpose large-coverage precise grammar for Spanish. It is grounded in the theoretical framework of HPSG and uses Minimal Recursion Semantics (MRS) for the semantic representation (Copestake at al., 2005). The SRG is implemented within the Linguistic Knowledge Builder (LKB), based on the basic components of the LinGO Matrix Grammar.
This resource provides a list of subcategorization frames for Spanish and Catalan extracted from the SenSem corpora. In addition to these results, the frequency and distribution of the frames according to the source (journalistic or literary) are provided as well. By default, frames are shown according to phrase type. However, users can also see the syntactic and semantic functions associated with the phrases. The list of sentences that exemplify the subcategorization frames in the corpus can be accessed from each frame.
This resource is available for download in Downloads.
ParTes. Test Suite for Parsing Evaluation
ParTes is the first test suite in Spanish and Catalan for parsing qualitative evaluation. This resource is a hierarchical test suite of the representative syntactic structure and argument order phenomena. The current version is distributed with the test suite module, where the syntactic structure phenomena and the argument order schemas are classified, and the test data sets, where both plain text and syntactical annotations (in dependency format) are available.
Semantic Representation of questions (ReSim)
ReSim is an engine that, given an interrogative sentence, proposes its semantic representation. The representation is based on Jackendoff´s lexical-conceptual structures (1990). This system has been developed for a restricted domain.