Social Media Data Opinion Miner and Sentiment Classifier using Deep Learning Methodology
Social Media Data opinion Mining and Sentiment analysis is one of the important application of NLP (Natural Language Processing) to analyze online social media users’ conversations and discussions , and find out the in depth context of a topic, a brand, a celebrity or a theme. There are various methods and algorithms used in the area of sentiment analysis and opinion mining. The Dictionary Based Approaches only identifies the number of positive words and the negative words in the sentence and takes a decision based on their difference. Thus a phrase “not bad” would became negative phrase. Using doc2vec, a deep learning algorithm which is used in this paper to drive the context from phrases. This algorithm calculates a value of probability of positiveness for each sentence. The range of values 0 -1 are considered as completely negative sentences, and the values from 0.35 to 0.65 are considered as completely positive sentences and the values which are in the middle are considered as neutral. There are various methods used to analyze and determine the user opinions from the massive amount of online data produced. This paper emphases on usage of deep learning based approach to develop a system to discover public opinions and sentiments. Experimental results prove that the accuracy on both training and testing datasets are high when compared with other machining learning approaches.