This paper presents SILEX, a multi-functional software system which makes it possible to solve a rather wide range of problems of computational linguistics concerning the Romanian language and the processing of Romanian texts.
The main functions of SILEX are:
- lemmatization, i.e. for each occurrence of a word, the system provides its lexico-grammatical class, together with its grammatical categories materialized in both the given word form and its lemma;
- paradigm generation (for inflectional words); being given a lemma and using the attributes associated with the lemma in the morphological dictionary, the system generates all the inflected forms of a word.
Recently, the SILEX system was enriched with several other functions: the automatic generation of lexical homographs, of homonyms correlated with morphological classes etc.
In performing these functions, SILEX uses an internal machine dictionary, which contains, in a structured form, a Romanian vocabulary of approximately 40,000 basic entries with all the morphological information needed for defining the systemic and (grammatical-) textual status of any lexical unit. These entries, as will be shown later, cover a Romanian vocabulary of approximately 60,000 words.
This dictionary is used exclusively
for performing the two main functions of SILEX. For building
this internal dictionary and for other tasks related to the Romanian
vocabulary, there are a number of auxiliary dictionaries,
one for each main lexico-morphologic class. These dictionaries
contain specific information for the morphological class they
refer to and will be presented in section 2.3. The internal dictionary
used by SILEX is created through a fusion process, which
consists of selectively picking up information from all these
auxiliary dictionaries and in putting it together in a structured,
uniform manner. Henceforward, the internal dictionary of
SILEX will be referred to as the SILEX dictionary,
or simply dictionary.
63
