Resources
This is the first lexical resource ready for download. We presented it in
Sokolova M., Bobicev V. "Classification of Emotion Words in Russian and Romanian Languages". Proceedings of RANLP-2009 conference, Borovets, Bulgaria, pp. 415-419, 2009. pdf
We plan to continue our work. An updated version of our resource will be posted as soon as it is ready.
WordNet-Affect translated in Romanian and Russian.
WordNet-Affect is a lexical resource that contains information about emotions the words convey. It has been developed from the lexical knowledge base WordNet, through a selection and labelling of the affective concepts represented by sets of synonyms. Affective labels (a-labels) were manually assigned to Word Net synsets of nouns, adjectives, verbs and adverbs which convey affective meaning. Words labelled with the Emotion tag were further reannotated into six emotional categories: joy, fear, anger, sadness, disgust, surprise. Word Net-Affect is freely available for research purposes at http://wndomains.itc.it. The collection of WORDNET-AFFECT synsets used in our work was provided as a resource in SemEval-2007 Affective Text task focused on text annotation with affective tags.
Word Net-Affect is organised in six files: anger.txt, disgust.txt, fear.txt, joy.txt, sadness.txt, surprise.txt. We keep the same data organisation. Data files are arranged as follows:
- each row contains one synset;
- the first letter in the row indicates part of speech: n-noun; a-adjective; v-verb; r-adverb;
- the letter is followed by the synset number;
- English synset definition follows the number and is separated by a tab;
- English synset words are listed separated by spaces;
- Romanian synset words are listed separated by spaces;
- Russian synset words are listed separated by spaces;
- English, Romanian and Russian sets of words are separated by tabs.
Please cite the following reference in the publications or presentations containing research results obtained through the use of this resource:
Sokolova M., Bobicev V. "Classification of Emotion Words in Russian and Romanian Languages". Proceedings of RANLP-2009 conference, Borovets, Bulgaria, pp. 415-419, 2009.
The second version of Romanian and Russian WordNet-Affect
The next version is aligned with Romanian WordNet accessed at MultiWordNet site. The aligning process was not trivial because WordNet-Affect was created on the base of WordNet 1.6 and MultiWordNet is based on WordNet 2.0. Some synsets were not exactly the same and we could not establish the correspondence. However, most of synsets found in Romanian WordNet were attached to Romanian part of WordNet-Affect.
Data files are arranged slightly in another order:
- English synset definition follows the number and is separated by a tab;
- English synset words are listed separated by spaces;
- Russian synset words are listed separated by spaces;
- Romanian synset words are listed separated by spaces;
- For Romanian synsets found in MultiWordNet definition in Romanian follows Romanian synset separated by tab.
The process of multilingual WordNet-Affect creation is described in the paper: "Emotions in words: developing a multilingual WordNet-Affect". CICLING 2010, Iasi, Romania, 2010. pdf