Sentiment Analysis IS AN AREA OF STUDY WITHIN NATURAL LANGUAGE PROCESSING THAT IS CONCERNED WITH IDENTIFYING THE MOOD OR OPINION OF SUBJECTIVE ELEMENTS WITHIN A TEXT. THIS PAPER FOCUSES ON THE VARIOUS METHODS USED FOR CLASSIFYING A GIVEN PIECE OF NATURAL LANGUAGE TEXT ACCORDING TO THE OPINIONS EXPRESSED IN IT I.E. WHETHER THE GENERAL ATTITUDE IS NEGATIVE OR POSITIVE. WE ALSO DISCUSS THE TWO-STEP METHOD (ASPECT CLASSIFICATION FOLLOWED BY POLARITY CLASSIFICATION) THAT WE FOLLOWED ALONG WITH THE EXPERIMENTAL SETUP.

SOCIAL NETWORKS ARE THE MAIN SOURCE OF USER OPINIONS ABOUT EVENT AND PRODUCT. EXTRACTING USER Sentiment FROM THEIR COMMENTS IN SOCIAL NETWORKS VERY HELPFUL FOR COMPANIES AND GOVERNMENTS FOR THEIR DEVELOPMENT PLAN. TWITTER CONSISTS OF BILLIONS OF USER AND THEIR OPINIONS AND IT IS A GOOD SOURCE FOR Sentiment Analysis. LOTS OF WORKS PROPOSED IN RECENT YEARS ABOUT Sentiment Analysis IN TWITTER. VARIOUS METHODS ARE USED TO DEVELOP A SA METHOD SUCH AS NLP BASED, MACHINE LEARNING BASED AND HYBRID METHODS. BUT, ALL OF THESE METHODS DON’T SATISFY ALL REQUIREMENTS OF THIS RESEARCH AREA. IN THIS REPORT WE TRY TO REVIEW THE IMPORTANT SOLUTIONS ARE PROPOSED FOR THIS PROBLEM.THIS PAPER CONSISTS OF FIVE CATEGORIES: 1) TO INTRODUCE AND TO ORIENTATE WITH THE FIELD OF Sentiment Analysis IN TWITTER SOCIAL NETWORKS 2) TO REVIEW THE WORKS DONE IN THE AREA OF SA 3) TO INTRODUCE CONFERENCES RELATED TO THE FIELD OF SA IN RECENT YEARS THAT HOLD COMPETITIONS WITH THEIR RESULTS AND THEIR BEST PRACTICES OFFERED 4) TO INTRODUCE AVAILABLE DATASETS 5) TO INTRODUCE A FEW AVAILABLE MASH-UPS.

Aspect-based Sentiment Analysis (ABSA) is a type of Sentiment Analysis that aims to identify the polarity of Sentiment for aspects in a sentence. Also according to the studies, it is an important research area that plays an important role in business intelligence, marketing and psychology. To solve this problem different methods based on dictionary, machine learning and deep learning have been used. Research shows that among the methods based on deep learning, Transformers has been able to achieve good results and help to understand the language better. In this paper we use induced trees from Fine-tuning pre-trained models (FT-PTMs). We also use dual contrastive learning and different pre-trained models such as BERT, RoBERTa and XLNet in our proposed model. The results obtained from the implementation of the model in SemEval2014 benchmarks confirm the performance of our model.

Recently, interests in the appliance of deep learning techniques in natural language processing tasks considerably increased. Sentiment Analysis is one of the most difficult tasks in natural language processing, mostly in the Persian Language. Thousands of websites, blogs, social networks like Telegram, Instagram and Twitter update, and modify by Persian users around the world that contains millions of contexts. To extract knowledge of these huge amounts of raw data, Deep Learning techniques became increasingly popular but there is a number of challenges that the novel models encounter with them. In this research, a hybrid deep learning-based Sentiment Analysis model proposed and implemented on customer reviews data of Digikala Online Retailer website. We already applied the classifier based on various deep learning networks and regularization techniques. Finally, by utilizing a hybrid approach, we achieved the best performance of 78. 3 of F1 score on three different classes: positive, negative, and neutral.

THE INCREASING GROWTH OF WEB HAS GIVEN PEOPLE THE ABILITY TO SIMPLY EXPRESS THEIR OPINION AND KNOW OTHERS’ OPINION. MINING VIEWPOINTS AND OPINION OR Sentiment Analysis IS CONSIDERED AS A SUBFIELD OF TEXT MINING AND ITS MAIN GOAL IS TO FIND WRITER’S OPINION ABOUT A TOPIC. MEETING THIS GOAL IS NOT A SIMPLE TASK SINCE EMOTIONS IN A SENTENCE OR A PHRASE ARE USUALLY RECOGNIZED BY COMBINING EMOTIONS OF ITS WORDS. IN THIS PAPER, WE CONCENTRATE ON BIPOLAR TERMS WHICH ARE THOSE PHRASES CONTAINING AT LEAST ONE POSITIVE AND ONE NEGATIVE WORD. IN ORDER TO CONSIDER BIPOLAR TERMS, PHRASES WITH OPPOSING POLARITY ARE FIRST EXTRACTED FROM PERSENT DATASET THEN, BASED ON THE WORDS OF THESE PHRASES AND THEIR POLARITY IN THE SENTENCE THE FINAL SCORE IS COMPUTED. THEN, THE SCORE OF EACH SENTENCE IS CALCULATED USING CNRC LEXICON AND MAXIMUM OF ABSOLUTE VALUES, DIFFERENCE, AND AVERAGE METHODS WITH AND WITHOUT CONSIDERING BIPOLAR TERMS. THE RESULTS OF IMPLEMENTATION OF THE PROPOSED METHOD SHOW THAT EMPLOYING BIPOLAR TERMS IMPROVES THE LEXICON-BASED APPROACH FOR BOTH POLARITY DETECTION AND SCORE PREDICTION PROBLEMS.