A Procedural DTD Project for Dictionary Entry Parsing Described with Parameterized Grammars
Neculai Curteanu, Mihai Alex Moruz · 2012
The present paper continues the successful parsing experiments with the method of Segmentation-Cohesion-Dependency (SCD) configurations, a breadth-first, formal grammar-free, and optimal approach to dictionary entry parsing, proposed in the previous CogALex Workshops and applied to the following five very large thesaurus-dictionaries: DLR (The Romanian Thesaurus – new format), DAR (The Romanian Thesaurus – old format), TLF (Le Tresor de la Langue Francaise), DWB (Deutsches Worterbuch – GRIMM), and GWB (Gothe-Worterbuch). In this work we report new results: (a) The lexicographic modeling and parsing experiments of the sixth large DMLRL (Dictionary of Modern Literary Russian Language); (b) Outlining the Enumeration Closing Condition (ECC) for solving the recursive calls between sense marker classes situated on different nodes of a sense dependency hypergraph (SCD-configuration, i.e. parsing level); (c) The central result we report here is the project of a new, procedural DTD (Document Type Description) for dictionaries, based on the formalization of the SCD parsing method, providing parameterized grammars to describe the dependency hypergraphs that correspond to the main parsing levels in a dictionary entry. Here we give two parameterized grammars for DLR, as a small sample from a larger package of combined grammars for the above mentioned dictionaries. This package is constructed as the “least common multiple” of the parameterized grammars written for the parsed dictionaries; it represents the DTD description of a general parser for large dictionary entries, and thoroughly extends the current DTD in the XCES TEI P5 standard.