RussNet as a monolingual resource
- RussNet was primarily designed to serve the needs of
RussNet data can be used for modifying user's query effectively:
Further IR-related subtasks are:
due to the paradigmatic relations presented, e.g., a
query may be expanded by members of the synset,
corresponding to the word meaning under consideration, as
well as by its co-hyponyms and hyperonym, while all the
words from other synsets including the given word, should
be excluded (taken as stop-words).
- due to the syntagmatic relations presented.
- Context Information Retrieval;
- Part-of-Speech Tagging;
- Word Sense Disambiguation;
- Text Similarity Measurement (doc, 43 Kb).
- RussNet can be used as a semantic component of different
formal grammars (e.g. AGFL).
- RussNet is a useful formalism to represent a vocabulary of
sublanguages (e.g. special terminology).
RussNet as a multilingual resource
At present we are developing mostly the language-internal
structures of RussNet and not focusing on Inter-Lingual-Index,
connecting semantically related words in different languages.
But in future it is planned to adjust RussNet to the common ILI
and, thus, make it useful for the following tasks as well:
- Cross-Language and Multilingual Information Retrieval;
- Machine Translation. (It is believed to be unreasonable to use
RussNet for MT directly. Yet, as ILI allows to extract not
only equivalent but semantically related words as well in
different languages, RussNet may be used a subsidiary
component for lexical analysis-synthesis accompanied by
morphological and syntactic components.)