I'm trying to find out how to make a sentiment dictionary for non-English languages like Farsi & Arabic, because as I know there is no SD available so far for many languages. By the way, is using weights of translated words from another dictionary (from English for instance) a valid method of attributing (positive, negative) emotions to words, for another language? Any suggestions would be appreciated.
You can check: "Creating sentiment dictionaries via triangulation" link to its sciendirect page: http://www.sciencedirect.com/science/article/pii/S0167923612001406
Good luck!
Yves Bestgren and Nadja Vincze have tried various methods to develop a dictionary of sentiment terms for French. If you read French, see their article "Identification de mots germes pour la construction d'un lexique de valence au moyen d'une procedure supervisee." They tried different methods and then compared them for accuracy. Should be able to find this article on google. Bibliography is still good since their article just appeared in 2011.
Please check this dicionary for Spanish colocations: http://www.dicesp.com/paginas/index/1
"La nomenclatura del DiCE se limita, por el momento, al campo semántico de los nombres de sentimiento.."
http://www.cl.uzh.ch/research/opinionmining_en.html
http://sentimental.li/german.lex
There are two general approaches: dictionary based and corpus based. In dic based, relations such as synonymy and antonymy are used to expand a small seed word list; in corpus based methods other semantic relations such as PMI between words, or conjunctions of adjectives by "and" and "or" are used to extract new polar terms. For more information, refer to the book opinion mining and sentiment analysis by Bing Liu.
With a colleague, Annie PIolat, we have designed Emotaix, a data base of emotional words in French It has more than 45000 items,words and locution, with literal and figurative meanings. Items are organize in positive and negative editions, with three levels of specificity ion emotions. The database has already been used fearfully in studies. A Portuguese translation is almost finished, an English one is on-going. Emotaix runs with Tropes, which allows analyzing emotional lexicon semi-automatically. Do not hesitate to contact me or Annie Piolat
https://www.researchgate.net/profile/Annie_Piolat
thank you everybody; @Thierry_Olive could you please elaborate a little about the database you mentioned in your answer to this post, i mean could you please enlighten me in detail how you made Emotaix? any particular method? how can i have a peek into it(the database)? and still i'm doubtful about the translation part; is there a standard routine for that? what are the usual assumptions you made when you translate your dictionary into another language?
@Hassan Abedi.
Constructing the database has en done by putting together all terms that were already used in other emotional databases, next searching in languages dictionary for all terms that contained words related to emotion in their definition. Second step was to organize the database. This was done by referring to the affected sciences, particularly Robert Plutchnik's works. Translation of the data base canot be don automatically. It is a very long processes, fastidious with a word to word translation, verifying in dictionaries, etc..... checkin for the meanings of figurative idioms, etc.
Check the attached file which show the organization of the database.
three approaches: (i) manual annotation, (ii) corpus based, and (iii) lexicon/dictionary based.
All of the above have merits and demerits, however, as far as your case is concerned, I recommend corpus-based or hybrid (corpus+lexicon), as it can produce promising results for creating non-English lexicons.
you can take idea from below work (although, its for English)
1. https://www.researchgate.net/publication/283068376_A_Unified_Framework_for_Creating_Domain_Dependent_Polarity_Lexicons_from_User_Generated_Reviews?ev=prf_pub
2. https://www.researchgate.net/publication/273727879_Medical_opinion_lexicon_an_incremental_model_for_mining_health_reviews?ev=prf_pub
3. https://www.researchgate.net/publication/283318837_SUBJECTIVITY_LEXICON_CONSTRUCTION_FOR_MINING_DRUG_REVIEWS?ev=prf_pub
Article Medical opinion lexicon: An incremental model for mining hea...
Article A Unified Framework for Creating Domain Dependent Polarity L...
Article Subjectivity lexicon construction for mining drug reviews
- Badges
- Science topic
- Similar topics
- Linguistics