Extending persian sentiment lexicon with idiomatic expressions for sentiment analysis

Abstract Nowadays, it is important for buyers to know other customer opinions make informed decisions on buying a product or service. In addition, companies and organizations can exploit improve their products services. However, the Quintilian bytes of generated every day cannot be manually read summarized. Sentiment analysis opinion mining techniques offer solution automatically classify summarize user opinions. current sentiment research mostly focused English, with much fewer resources available languages like Persian. our previous work, we developed PerSent, publicly lexicon facilitate lexicon-based texts in Persian language. PerSent-based approach fails real-world sentences consisting idiomatic expressions. Therefore, this paper, describe an extension PerSent more than 1000 expressions, along polarity, propose algorithm accurately text. Comparative experimental results reveal usefulness extended as compared well Persian-to-English translation-based approaches. The version will made available.