Certain investigation on predicting the users behavior in social networks using enhanced graph based semi supervised learning algorithm

Abstract

newline newline Social media offers a flexible environment where user can share their thoughts and feelings via text emoji s Images etc The volume velocity and Veracity of the data generated are huge high speed and with the uncertainty of accuracy Big Data is a kind of data with colossal Volume Variety and a high Velocity of data cohort The data generated through high Velocity does not support a modest mechanism to capture and process the information In order to capture the streaming data and utilize the information for prediction of user behavioural analysis is challenging Clustering and classification for streaming data are immensely complicated and results in an incorrect prediction Hence this complication results in a massive loss in the Healthcare sector concerning the prediction of suitable medicine for the appropriate disease Also, in other sectors it results in massive loss in optimized decision making for a natural disaster by government organization bandwidth allocation in social networking sites and manufacturing the products in industries etc Hence Sentimental Analysis or Opinion Mining is used to regulate the detection of subjective data such as opinions attitudes emotions and feelings Sentimental Analysis is an area of enormous potential where machines can learn or train by algorithms which makes better decision comparatively while the techniques of Sentimental Analysis can vary from very simple nuanced to complicate nuanced therefore existing research focus only on binary sentiment classification Since the data generated is enormous and with high speed Apache Hadoop is used for storing through Hadoop Distributed File System HDFS and processing through MapReduce Hadoop offers the storage of the data in replication to avoid data failures The MapReduce framework is a combination of two sections namely the Map and Reduce Primarily the input data could be segmented into a various section using replica key value pairs Apache Flume isa progressively reliable distributed and configurable tool