A Recommender System Based On Collaborative Filtering Using Polarity Improvement in Sentiment Analysis

Sentiment Analysis, which is a new subfield of the processing of natural language and text mining, categorizes the texts based on the sentiment expressed in them. Sentiment plays a significant role in decision-making. So sentiment analysis technology has a broad scope for scientific applications. On the other hand, a huge amount of information in the world today is in the form of text. Therefore, text mining techniques are important. Exploring comments or analyzing sentiment as a branch of text mining, means finding the author's perspective on a specific subject. The Internet allows users to easily express their opinions and get informed about the opinions of others. The high volume and the lack of proper structure for the text of the comments provided on the web, make it difficult to use hidden knowledge within them. Therefore, it is important to provide methods that can prepare and provide this knowledge in a summarized and structured way. In this research, it has been tried to provide a fuzzy method for analyzing the following comments on news sites according to the text of the report. In this regard, it has been tried to investigate the relationship with the author's commentary and opinion in light of the subject of the text using the grammatical features of texts such as noun and verb, as well as sentimental load analysis of sentences. Subsequently, the method is evaluated by implementing it on the dataset collected from news and comments. The proposed method has 87% diagnosis accuracy.