Abstract: Rich source of information is present in social media. Social media provides meaningful information of gleaning emotions of people and understanding public attitude and mood more deeply. Internet provides a huge platform for exchanging ideas, online learning, sharing ideas on social networking sites such as Google plus, Instagram, Twitter, Facebook. There is huge volume of data in the web with the advancement of web technologies and its growth. Social networking sites give exposure, allow people to express and share their view about the topic discussion with various communities and can post message across the world. Lot of work is been progressed in the field of sentimental analysis of twitter data. The emphasis of this survey is to analyse the in the tweets where opinions are highly unstructured, heterogeneous opinions are either positive, negative or neutral. In this paper we present a survey and comparative analysis of existing techniques for opinion mining. there are two main parts in framework initially, ensemble classifier schema, combines knowledge based, generic domain independent with machine learning methods, used to perceive contents of emotions in user generated data. To model the emotional level of the topic based on emotional recognition, a graph-based method is used to create the topic emotional graph visualizing public moods and emotion on topic. Encouraging results are observed.
Keywords: Sentiment analysis, Twitter, Predictive Analysis, Affinity analysis, Pre-processing, R Language, Social Network
|
DOI:
10.17148/IJARCCE.2018.71014
