SlpTk library 0.6.0

SlpTk0.6.0

SlpTk is a tool for natural language processing which is composed of an ANSI C library and a set of command line utilities and scripts. The library provides data structures and treatments related to lexical (for example lexical correction and tokenization) and grammatical levels (for example probabilistic parsing). It works on any platforms that has an ANSI C compiler. SlpTk is released under the terms of the GNU General Public License (GPL) and is developped at:

• Swiss Federal Institute Of Technology Lausanne (EPFL) - http://www.epfl.ch ;
  
• Computer Sciences & Communication Faculty (I&C);
  
• Articifial Intelligence Laboratory (LIA) - http://liawww.epfl.ch ;
  

This documentation describes the API of SlpTk library which constists of 4 hierarchical levels:

• nospectools: it gathers low-level modules, which are general tools widely used inside the library. These modules are not domain specific and may be used by other applications.

• spectools: it gathers low-level modules that are specific to natural language processing, like character class recognition functions.

• lexicaltools: it gathers modules dedicated to data structures and treatments related to lexical level. Two kinds of lexical memories are defined here. Access table allows unidirectional access (sequence -> index) and may concretely be implemented by the lexical tree or by the lexical automaton structure. Lexical correction operations can be performed from the content of this type of memory. Associative memory allows bidirectional access (sequence <-> index) and may concretely be implemented by the bidirectional lexical tree or by the lexical automaton data structure. This type of lexical memory makes up the basis of a lexicon. A lexicon can however not be used with a lexical automaton strucutre due to it's import mechanism.

• syntactictools: it gathers modules dedicated to data structures and treatments of the syntactical level. The parser module allows to perform a syntactical analysis from a given grammar . All availables interpretations are stored in a parsing chart that can be display under various formats. Parsing can also performed on speech lattice.

---