Abstract: Now a days, social media is hot topic on research. Millions of the peoples express their views on social media. This huge data will beneficial for better product marketing. But, because of massive of volume of reviews, end-users can’t read all reviews in order to solve this problem lot of researchers has been carried out sentiment analysis.
Sentiment analysis is the automated process of understanding and opinion about a given subject from return or written language. Most of sentiment analysis & opinion mining work focuses on binary classification & ternary classification of texts. But our novel idea is to classify the text or sentences into multiple classes. Using Hotel Reviews datasets we classify the sentences into multiple classes like happy, sad, hungry, love etc. In dataset various sentences contains the hashtags, URL’s, operators it cannot change the analysis of the sentences or meaning of that sentence but it create the confusion while determining the result. So we can apply the pre-processing method to remove all these things. After that the feature extraction method is apply on the processed dataset to extract their aspects. Later, this aspects are used to calculate the positive and negative polarity in sentence. Then the model is trained on training dataset using supervised learning method. The training consist of the pairs of input and the corresponding answer vector and the current model is run with the training dataset and produces a result. By using machine learning algorithm or NLP algorithms, the classification will give the better accuracy & these analyses will be helpful for product developer and end-user.
Keywords: Sentiment Analysis, Machine Learning, Feature Extraction
| DOI: 10.17148/IJARCCE.2018.71139