аннотация7 (Аннотации)

Ulanov A. V., Sapozhnikov G. A.:Context-dependent opinion lexicon translation with the use of a parallel corpus.The paper deals with multilingual sentiment analysis. It’s one of the mostpopular information extraction tasks both for business and research. It has numerousbusiness applications, such as evaluation of a product or company perception insocial media. Opinion words are used in a number of sentiment analysis tasks. Theyinclude document and sentence sentiment classification, product features extraction,subjectivity detection etc.The authors propose a method for projecting an opinion lexicon from a sourcelanguage to a target language with the use of a parallel corpus. They make sentimentclassification in a target language using an opinion lexicon even without labeleddataset.

The advantage of this method is that it captures the context of a word and,thus, produces a correct translation of it. The algorithm of the proposed methodconsists of three steps: collecting a corpus of parallel reviews and sentencesalignment; computation word lexical translation probabilities; collecting opinionwords translations and their normalizing.The method was applied to the language pair English-Russian and conductsentiment classification experiments. They show that the suggested method allowscreating high-quality opinion lexicon set..