A novel approach to sentiment analysis in Persian using discourse and external semantic information

Sentiment analysis attempts to identify, extract and quantify affective\nstates and subjective information from various types of data such as text,\naudio, and video. Many approaches have been proposed to extract the sentiment\nof individuals from documents written in natural languages in recent years. The\nmajority of these approaches have focused on English, while resource-lean\nlanguages such as Persian suffer from the lack of research work and language\nresources. Due to this gap in Persian, the current work is accomplished to\nintroduce new methods for sentiment analysis which have been applied on\nPersian. The proposed approach in this paper is two-fold: The first one is\nbased on classifier combination, and the second one is based on deep neural\nnetworks which benefits from word embedding vectors. Both approaches takes\nadvantage of local discourse information and external knowledge bases, and also\ncover several language issues such as negation and intensification,\nandaddresses different granularity levels, namely word, aspect, sentence,\nphrase and document-levels. To evaluate the performance of the proposed\napproach, a Persian dataset is collected from Persian hotel reviews referred as\nhotel reviews. The proposed approach has been compared to counterpart methods\nbased on the benchmark dataset. The experimental results approve the\neffectiveness of the proposed approach when compared to related works.\n