The paper presents a system for identifying and classifying sentiments in opinions, focusing on the affective aspects of text. The system aims to find people who hold opinions about a given topic and determine the sentiment of each opinion. It consists of two main components: a word sentiment classifier and a sentence sentiment classifier. The word sentiment classifier uses a small set of manually selected positive and negative words, expanded with synonyms and antonyms from WordNet, to classify the sentiment of individual words. The sentence sentiment classifier combines the sentiments of words within a defined region to determine the overall sentiment of the sentence. The system also includes a mechanism for identifying opinion holders and determining the sentiment region within a sentence. Experiments with various models and datasets show promising results, with the best performance achieved by a model that considers the polarity of words rather than their sentiment strength. The paper discusses challenges and future directions, including improving the identification of opinion holders and handling sentences with multiple opinions or weak-opinion-bearing words.The paper presents a system for identifying and classifying sentiments in opinions, focusing on the affective aspects of text. The system aims to find people who hold opinions about a given topic and determine the sentiment of each opinion. It consists of two main components: a word sentiment classifier and a sentence sentiment classifier. The word sentiment classifier uses a small set of manually selected positive and negative words, expanded with synonyms and antonyms from WordNet, to classify the sentiment of individual words. The sentence sentiment classifier combines the sentiments of words within a defined region to determine the overall sentiment of the sentence. The system also includes a mechanism for identifying opinion holders and determining the sentiment region within a sentence. Experiments with various models and datasets show promising results, with the best performance achieved by a model that considers the polarity of words rather than their sentiment strength. The paper discusses challenges and future directions, including improving the identification of opinion holders and handling sentences with multiple opinions or weak-opinion-bearing words.