Stream data analysis using advanced machine learning approaches

Abstract

Stream data analysis involves the process of identifying the potential patterns newlinefrom the data streams from various sources. Many of the data stream sources newlineare social media applications. Social networks evolved as a mandate for newlinesharing information instantly. Social networking applications are prominent newlineamong the internet user communities. Many social media websites are used newlinefor sharing the information instantly. newlineTwitter is one of the vibrant social networking websites for sharing newlinesmall textual information within a short span of time. It is essential to identify newlinethe type of information shared on these websites. The twitter stream data newlineanalysis mainly involves the sentiment analysis process using various trained newlinemachine learning classifiers applied on a large collection of tweets. The newlineclassifiers are trained using a maximum number of polarity oriented words for newlineeffectively classifying the tweets. The trained classifiers at sentence level newlineoutperformed the keyword based classification method. The classified tweets newlineare further analyzed for identifying the top tweets. The experimental results newlineshow that the sentiment analysis process predicted polarities of tweet and newlineeffectively identified the top tweets. In addition to the polarity prediction, newlinescore calculation of sentiment content on tweets is essential. newlineSentiment score calculation is carried out with sentiment corpus newlineapproach for calculating the score effectively. Especially, the grammatical newlinetype of the word used in a tweet and the relationship between the words are newlineproperly identified. The tweet tagger and corpus sentiment score assignment newlineis distinctively used when compared to other previous works. newline