Important Features Extraction by using POS Tag Pattern of Dependency Relation for Online Movie Reviews

With the emerging of web technologies, online reviews become trending and lead to the progression of sentiment analysis research. Feature extraction is the most important task in sentiment analysis of online reviews. This task is enable to help researcher to differentiate between relevant and irrelevant features in sentiments. The relevant features will kept for later process while the irrelevant features will be discarded. Dependency relation is one technique that can be used in features extraction process. By manipulating the dependency relation between words in reviews, the most important features can be identified. However, not all types of dependency relation are able to extract the important features. In this paper, the most useful of specific types of dependency relation with POS tag pattern for extracting the important features from movie reviews dataset is proposed. The proposed pattern will be used to filter out the relevant features that later will send to the classification process. By doing this, it will increase the classification accuracy as well as decrease the time taken for classification process.