Abstarct: Emotion is an important aspect of human behavior through which people in a society affect each other. Sentiment analysis is for providing the signs and more interaction between human and computer. In addition to sentiment analysis from face, gestures and speech, we can recognize the emotions through the written texts. Sentiment analysis is a branch of computer science and Natural Language Processing (NLP) which can help find the author’s motivation. This paper entixtled "Sentiment analysis in text using artificial intelligence techniques" aims to classify the emotional texts and to discover the emotional mood of the author to develop a system which is smart enough and deals with human including emotions and can recognize the user’s emotions. The proposed model in this paper considers the extracted features in the text in two frxames of 1-gram and 2-gram. Feature filtering is used to reduce the feature scattering and consequently increases the classification accuracy and results improvement. The proposed model is tested using the dataset of emotional texts made for this purpose. The emotional texts are categorized in five groups using Bayes method. These five groups are grief, happiness, anger, hatred and fear. The obtained results demonstrate that in the best mode, sentiment analysis accuracy using 1-gram and 2-gram features without filtering are 79.1% and 27.7%, respectively. By filtering these features and therefore by reducing the scattering, accuracy of 91.6% and 43% are obtained for 1-gram and 2-gram features.